Showing 1–50 of 390 results for author: Gonzalez, J

Generative Proxy: Synthesizing Proxy-Based Interfaces for Real-World Interaction Across AR Glasses

Authors: Xianhao Carton Liu, William Chastek, Eric J Gonzalez, Mar Gonzalez-Franco, Chen Zhu-Tian

Abstract: Interacting with real-world objects in AR is difficult, especially when targets are distant, cluttered, or occluded. These challenges are amplified on emerging lightweight AR glasses, which often lack binocular or large field of view on display, but also continuous inputs, such as hand or eye tracking. Proxy-based interfaces offer an alternative by allowing users to interact with virtual abstracti… ▽ More Interacting with real-world objects in AR is difficult, especially when targets are distant, cluttered, or occluded. These challenges are amplified on emerging lightweight AR glasses, which often lack binocular or large field of view on display, but also continuous inputs, such as hand or eye tracking. Proxy-based interfaces offer an alternative by allowing users to interact with virtual abstractions of physical objects that can be repositioned, reorganized, and adapted to the task and device. However, designing such interfaces is currently manual and highly device-specific. We present Generative Proxy, a method for automatically generating proxy-based interfaces from three specifications: scene, intent, and device capabilities. We formulate generation as a constrained synthesis problem that first produces valid interfaces for the target device and task, then ranks candidates using semantic and articulatory distance inspired by direct manipulation theory. We demonstrate Generative Proxy across diverse scenes, device profiles, and user intents. Expert evaluation shows initial evidence that generated proxy UIs are useful and usable, highlighting proxy-based abstraction as a promising interaction paradigm for future AR glasses. △ Less

Submitted 9 August, 2026; originally announced August 2026.

Comments: 11 pages, 11 figures, published in UIST '26, Detroit, MI, USA, Nov 2 - 5, 2026; UIST '26: ACM Symposium on User Interface Software and Technology

Fantastic Adaptive Taxonomies and How to Use Them

Authors: Mert Cemri, Andrei Cojocaru, Melissa Pan, Shu Liu, Shubham Agarwal, Alexander Krentsel, Jay Tang, Kannan Ramchandran, Joseph E. Gonzalez, Matei Zaharia, Alex Dimakis, Ion Stoica

Abstract: An agent system's execution traces record how it fails, and procedures that improve such a system without changing model weights (trajectory selection, prompt and workflow optimization, runtime monitoring) read these traces for feedback. Yet raw traces are a poor medium for accumulating that feedback: long, instance-specific, and lacking a stable vocabulary for recurring failures. We argue that an… ▽ More An agent system's execution traces record how it fails, and procedures that improve such a system without changing model weights (trajectory selection, prompt and workflow optimization, runtime monitoring) read these traces for feedback. Yet raw traces are a poor medium for accumulating that feedback: long, instance-specific, and lacking a stable vocabulary for recurring failures. We argue that an agent system should instead maintain an explicit representation of how it fails, induced from its own behavior and reusable wherever failure feedback is needed. AdaMAST builds this representation by converting a target system's traces into a compact, evidence-grounded failure taxonomy: named failure codes organized along three fixed axes (system-level, role-specific, and domain-specific), with every name, definition, and evidence pattern induced from the traces; no code is hand-authored, no trace human-annotated. The taxonomy is not merely a post-hoc diagnostic but a shared feedback interface, improving agents in three ways. In agent-system search, taxonomy-coded diagnoses of failed candidates outperform free-form reflection on all five benchmarks we test. At runtime, taxonomy feedback raises SWE-agent's resolution on SWE-bench Verified Mini from 60% with free-text reflection to 70%, and improves Claude Code from 64.0% to 70.7% as a runtime skill. In trajectory selection, AdaMAST-Judge, a verifier built on the induced codes, improves best-of-5 accuracy on Terminal-Bench 2.0 by 8-15 points over Pass@1. The vocabulary itself is compact (an order-of-magnitude compression that preserves trace distinctions), human-faithful (matching expert failure annotations more closely than a hand-crafted reference vocabulary), and adaptive (taxonomies induced for different domains share few codes). Adaptive failure taxonomies close the loop between the traces agents produce and the procedures that improve them. △ Less

Submitted 29 July, 2026; v1 submitted 17 July, 2026; originally announced July 2026.

Underwriting the Agent Economy: The Blueprint for an AI Insurance Stack

Authors: Cristian Trout, Sanmi Koyejo, Sasha Romanosky, Giorgio Ripamonti, Lynn Thompson, Desiree Spain, Alex Taylor, Kevin Casey, Stephen Casper, Matthew Botvinick, Sean McGregor, Miles Brundage, A. Feder Cooper, Patricia Paskov, Adrien Ecoffet, Ben Bucknall, Kevin Wei, Markus Anderljung, Lukasz Szpruch, Bri Treece, Tom Zick, Gabriel Weil, Ugur Ozer, Kevin Kalinich, Jesus Gonzalez , et al. (12 additional authors not shown)

Abstract: From maritime trade to commercial nuclear power, insurance has been the enabler of major economic and technological developments by pricing risk, limiting downside, and spreading best practices. The emerging AI agent economy, projected to handle trillions of dollars in transactions by 2030, looks to be the next such development. Yet insurers' exposure to AI agent risk currently sits largely unpric… ▽ More From maritime trade to commercial nuclear power, insurance has been the enabler of major economic and technological developments by pricing risk, limiting downside, and spreading best practices. The emerging AI agent economy, projected to handle trillions of dollars in transactions by 2030, looks to be the next such development. Yet insurers' exposure to AI agent risk currently sits largely unpriced across existing insurance lines; between this silent coverage and growing exclusions, coverage is not fit for purpose. Furthermore, insurability is trending the wrong way: AI agent capabilities appear to be outpacing reliability, leading to rising incident severity; concentration among a few foundation model providers threatens correlated losses; and traditional actuarial modeling will struggle to keep pace with a technology evolving as rapidly as frontier AI. This report argues that affirmative AI coverage with limits in the billions is achievable by 2030, but only with industry-wide coordination. Drawing on successful historical precedents such as Underwriters Laboratories, the Closed Claims Project, and others, we lay out an eight-component AI insurance stack spanning incident data collection, catastrophe modeling, standards, contract design, risk selection, pricing, monitoring, and claims management. Building out this infrastructure is what will enable insurers to cover and manage AI agent risk sustainably and at scale. Finally, we discuss coverage for catastrophic risk from frontier AI ("AI CAT"), including CBRN, critical infrastructure collapse, and loss of control scenarios. Addressing these tail risks will require purpose-built instruments, potentially including a frontier model developer mutual, catastrophe bonds, bespoke liability regimes, and government backstops. △ Less

Submitted 14 July, 2026; v1 submitted 13 July, 2026; originally announced July 2026.

A Strong Balanced-Softmax Classifier-Retraining Baseline for Long-Tailed Recognition

Authors: Juan Terven, Diana Margarita Córdova Esparza, Julio Alejandro Romero Gonzalez, Edgar Arturo Chávez Urbiola, Francisco Javier Willars Rodriguez, Juan Bautista Hurtado Ramos, Alfonso Ramirez Pedraza

Abstract: Long-tailed recognition methods often modify losses, margins, or representations to reduce the dominance of frequent classes. We ask whether, after Balanced Softmax training, the remaining tail error can be reduced by retraining only the classifier. We evaluate BS-cRT, a two-stage procedure that trains a backbone and cosine classifier with Balanced Softmax, freezes the backbone, and updates only t… ▽ More Long-tailed recognition methods often modify losses, margins, or representations to reduce the dominance of frequent classes. We ask whether, after Balanced Softmax training, the remaining tail error can be reduced by retraining only the classifier. We evaluate BS-cRT, a two-stage procedure that trains a backbone and cosine classifier with Balanced Softmax, freezes the backbone, and updates only the classifier on balanced episodic batches. The second stage keeps the empirical-prior Balanced Softmax objective and uses raw cosine logits at inference. Across CIFAR-100-LT, CIFAR-10-LT, ImageNet-LT, and Places-LT, this classifier-only step consistently improves Few-shot accuracy over the matched Balanced Softmax checkpoint. At imbalance factor 100, Few-shot gains are +5.15 points on CIFAR-100-LT and +5.83 on CIFAR-10-LT; on ImageNet-LT and Places-LT, gains are +6.92 and +9.78 points, respectively, with a Top-1/Few-shot trade-off on ImageNet-LT. We also analyze Counterfactual Boundary Risk Minimization (CBRM), a boundary-probe extension using prototype-based features near decision boundaries. CBRM identifies two failure modes: scaled-logit cosine margins destabilize training, and corrected hardest-negative probes remain head-class anchored. The results support BS-cRT as a practical classifier-side baseline and indicate that boundary supervision must account for class frequency. △ Less

Submitted 10 July, 2026; originally announced July 2026.

Comments: 18 pages, 6 figures, 7 tables

ACM Class: I.4.0

Sculptable Mesh Structures for Room-Scale Form-Finding

Authors: Jesse T. Gonzalez, Yanzhen Zhang, Dian Zhu, Alice Yu, Sapna Tayal, Nazm Furniturewala, Ziying Qi, Somin Ella Moon, Leyi Han, Alexandra Ion, Scott E. Hudson

Abstract: It can be hard to design a physical structure entirely within the confines of a computer monitor. To better capture the interplay between real-world objects and a designer's work-in-progress, practitioners will often go through a sequence of low-fidelity prototypes (paper, clay, foam) before arriving at a form that satisfies both functional and aesthetic concerns. While necessary, this model-makin… ▽ More It can be hard to design a physical structure entirely within the confines of a computer monitor. To better capture the interplay between real-world objects and a designer's work-in-progress, practitioners will often go through a sequence of low-fidelity prototypes (paper, clay, foam) before arriving at a form that satisfies both functional and aesthetic concerns. While necessary, this model-making process can be quite time-consuming, particularly at larger scales, and the resulting geometry can be difficult to translate into a CAD environment, where it will be further refined. This paper introduces a user-adjustable, room-scale, "shape-aware" mesh structure for low-fidelity prototyping. A user physically manipulates the mesh by lengthening and shortening the edges, altering the overall curvature and sculpting coarse forms. The edges are equipped with resistive length sensors, and transmit their configuration to a central computer. The structure can later be reproduced in software, connecting this prototyping stage to the larger computational design pipeline. △ Less

Submitted 9 July, 2026; originally announced July 2026.

Journal ref: UIST 2025: Proceedings of the 38th Annual ACM Symposium on User Interface Software and Technology

PIXELRAG: Web Screenshots Beat Text for Retrieval-Augmented Generation

Authors: Yichuan Wang, Zhifei Li, Zirui Wang, Paul Teiletche, Lesheng Jin, Matei Zaharia, Joseph E. Gonzalez, Sewon Min

Abstract: Augmenting large language models (LLMs) with retrieved web text has become a dominant paradigm, yet the web is not natively textual: existing systems depend on complex parsing pipelines that linearize HTML and discard layout, visual structure, and formatting. We introduce PixelRAG, a new retrieval-augmented method that represents websites in their native visual form and performs retrieval and read… ▽ More Augmenting large language models (LLMs) with retrieved web text has become a dominant paradigm, yet the web is not natively textual: existing systems depend on complex parsing pipelines that linearize HTML and discard layout, visual structure, and formatting. We introduce PixelRAG, a new retrieval-augmented method that represents websites in their native visual form and performs retrieval and reading entirely in pixel space, enabling an end-to-end architecture that eliminates text abstraction. PixelRAG is, to our knowledge, the first pipeline to operate over a full Wikipedia corpus in this form, scaling to a datastore of 30 million screenshot images with an efficient visual retrieval index. Built on an existing visual embedding model (i.e., Qwen3-VL-Embedding), PixelRAG further fine-tunes this model on screenshot data with carefully curated contrastive training data. Retrieved screenshots are then fed directly as pixel inputs to a VLM, without intermediate text conversion. PixelRAG consistently outperforms both no-retrieval and text-based RAG baselines, most surprisingly on widely studied text-centric tasks such as NQ and SimpleQA. It also achieves strong gains on multimodal open-domain QA (e.g., MMSearch), benchmarks over noisy news corpora (e.g., LiveVQA), and agentic benchmarks (e.g., MoNaCo), improving accuracy by up to 18.1% over text-based baselines. Finally, pixel representations enable a new efficiency lever for RAG through image compression, achieving up to 3x token cost reduction at lower resolutions while maintaining accuracy. Our results challenge the necessity of text representations in web retrieval, suggesting that web RAG can operate directly in the web's native visual form while improving both performance and efficiency. △ Less

Submitted 1 June, 2026; originally announced June 2026.

Comments: Our code is available at https://github.com/StarTrail-org/PixelRAG

Comparing Controller-Free Pointing Techniques Across Depth for 2D Selection in Augmented Reality

Authors: Samiha Sultana, J. Felipe Gonzalez, Robert J. Teather

Abstract: This paper presents a systematic evaluation of five controller-free pointing techniques for 2D target selection in AR, using ISO 9241-411. We compared them across multiple depths (2 m, 6 m, 10 m) in terms of movement time, accuracy, throughput, and workload (NASA TLX). Head- and eye-based pointing significantly outperformed the hand-based methods (Finger, Wrist, and Arm); Head input was the most a… ▽ More This paper presents a systematic evaluation of five controller-free pointing techniques for 2D target selection in AR, using ISO 9241-411. We compared them across multiple depths (2 m, 6 m, 10 m) in terms of movement time, accuracy, throughput, and workload (NASA TLX). Head- and eye-based pointing significantly outperformed the hand-based methods (Finger, Wrist, and Arm); Head input was the most accurate and remained the most consistent across depth. Depth significantly impacted performance, with complex interactions with target size and distance. Our results offer a comprehensive empirical basis for selecting appropriate controller-free techniques in depth-varying AR tasks. △ Less

Submitted 6 June, 2026; originally announced June 2026.

Journal ref: Proceedings of the Graphics Interface Conference 2026

Continual Learning Bench: Evaluating Frontier AI Systems in Real-World Stateful Environments

Authors: Parth Asawa, Christopher M. Glaze, Gabriel Orlanski, Ramya Ramakrishnan, Benji Xu, Asim Biswal, Vincent Sunn Chen, Frederic Sala, Matei Zaharia, Joseph E. Gonzalez

Abstract: Continual learning, the ability of AI systems to improve through sequential experience, has attracted substantial interest, but no high-quality benchmark exists to evaluate it. We introduce Continual Learning Bench (CL-Bench), the first difficult, expert-validated benchmark designed to measure whether LLM-based systems genuinely improve with experience. CL-Bench spans six diverse domains (software… ▽ More Continual learning, the ability of AI systems to improve through sequential experience, has attracted substantial interest, but no high-quality benchmark exists to evaluate it. We introduce Continual Learning Bench (CL-Bench), the first difficult, expert-validated benchmark designed to measure whether LLM-based systems genuinely improve with experience. CL-Bench spans six diverse domains (software engineering, signal processing, disease outbreak forecasting, database querying, strategic game-playing, and demand forecasting), each validated by domain experts and designed so that tasks share a learnable latent structure (codebase layout, disease outbreak dynamics, opponent strategies) that a stateful system can discover online but a stateless one cannot. We evaluate frontier models across several agent architectures, from naive in-context learning (ICL) to dedicated memory systems, introducing a gain metric to isolate learning from prior capabilities. We find that these systems leave headroom for improved continual learning: agents frequently overfit to immediate observations or fail to reuse knowledge across instances, and dedicated memory systems do not fix this -- in fact, naive ICL outperforms systems dedicated to memory management. CL-Bench is the first benchmark to evaluate continual learning across diverse real-world domains with expert-validated tasks and isolate online learning from underlying model capability, showing a need for better continual learning systems. △ Less

Submitted 3 June, 2026; originally announced June 2026.

PyCC.id: A package for hypothesis-driven equation discovery with structural identifiability

Authors: Federico J. Gonzalez

Abstract: Data-driven equation discovery is fundamentally an inverse problem that seeks to infer the governing differential equations of a system directly from time-series measurements. A known issue is the ill-conditioned nature of the inverse problem, which frequently produces multiple mathematical models that fit the data similarly well. One path to address this issue is by incorporating known hypotheses… ▽ More Data-driven equation discovery is fundamentally an inverse problem that seeks to infer the governing differential equations of a system directly from time-series measurements. A known issue is the ill-conditioned nature of the inverse problem, which frequently produces multiple mathematical models that fit the data similarly well. One path to address this issue is by incorporating known hypotheses and constraints into the training phase beforehand. While this approach effectively reduces the search space, it still results in multiple candidate models, forcing practitioners to rely on post-hoc manual filtering based on their own domain expertise. A recent approach incorporates structural `skeletons' inspired by characteristic curves (CCs), defining a hypothesis-driven methodology. In this methodology, practitioners define a skeleton, which is associated with a family of ordinary differential equations (ODEs), and then add their hypotheses and priors based on their domain knowledge to refine the obtained model iteratively. An important advantage of this approach is that some skeletons have demonstrable structural identifiability properties, which are useful for checking whether the skeleton is correct or should be discarded. Furthermore, this formalism enables the use of multiple equation discovery paradigms due to its modularity (such as neural networks, symbolic regression, and sparse regression). In this work, we present the Python library PyCC, which condenses these efforts into a flexible tool that allows researchers and engineers to seamlessly define their skeletons and hypotheses to discover ODEs from time-dependent data. △ Less

Submitted 6 May, 2026; originally announced June 2026.

Comments: The software package is available at: https://github.com/FedejGon/pyCC.id

Stateful Visual Encoders for Vision-Language Models

Authors: Zirui Wang, Junwei Yu, Adam Yala, David M. Chan, Joseph E. Gonzalez, Trevor Darrell

Abstract: Vision-language models (VLMs) are increasingly used in multi-image, multi-turn agentic settings where decisions depend on visual changes. However, in existing open-weight VLMs, visual comparisons happen only inside the language model, while the visual encoder itself remains stateless: each image is encoded independently, without access to the prior visual context. As a result, small but task-criti… ▽ More Vision-language models (VLMs) are increasingly used in multi-image, multi-turn agentic settings where decisions depend on visual changes. However, in existing open-weight VLMs, visual comparisons happen only inside the language model, while the visual encoder itself remains stateless: each image is encoded independently, without access to the prior visual context. As a result, small but task-critical changes may be attenuated before the language model has a chance to compare them, especially when those changes do not affect the high-level semantics of the scene. We introduce a Stateful Visual Encoder, which conditions each visual representation on prior visual features. Under supervised finetuning, VLMs equipped with stateful encoders achieve consistent improvements on controlled tasks involving cross-image spatial aggregation, multi-object visual differencing, and visual trajectory behavior cloning. These improvements are consistent across input resolutions, language model sizes, and VLM backbones. Finally, we validate our model on real-world tasks, including longitudinal radiology, fine-grained image comparison, and remote sensing, where stateful encoders consistently improve generalist VLM baselines and can match or surpass specialized models in selected domains. Project page: https://statefulvisualencoders.github.io/ △ Less

Submitted 3 June, 2026; originally announced June 2026.

Comments: Project page: https://statefulvisualencoders.github.io/

An Exploration of Collision-based Enemy Morphology Generation

Authors: Johor Jara Gonzalez, Matthew Guzdial

Abstract: Despite a great deal of prior research into Procedural Content Generation (PCG), relatively little prior work has explored generating enemies for video games. In particular, there is almost no work on generating enemy morphologies, the basic body plan or collision information for in-game enemies, despite the existence of related morphology generation work in robotics. In this paper, we explore thr… ▽ More Despite a great deal of prior research into Procedural Content Generation (PCG), relatively little prior work has explored generating enemies for video games. In particular, there is almost no work on generating enemy morphologies, the basic body plan or collision information for in-game enemies, despite the existence of related morphology generation work in robotics. In this paper, we explore three different novel approaches to generate enemy morphologies based on player collision information. We found that each approach provides different strengths and weaknesses, but all had equivalent or better performance than an evolutionary baseline adapted from prior robotics morphology work. △ Less

Submitted 1 June, 2026; originally announced June 2026.

Further evidence towards the Fourier Entropy-Influence conjecture

Authors: María José González, Paul MacManus, María Cristina Pereyra

Abstract: The Fourier Entropy-Influence (FEI) conjecture states that the Fourier entropy of Boolean functions is uniformly bounded by their total influence. It has been verified for canonical examples such as disjoint tribes and for some classes of Boolean functions such as symmetric functions and read-$k$ decision trees (with a constant that depends linearly on $k$). In this note we present new classes of… ▽ More The Fourier Entropy-Influence (FEI) conjecture states that the Fourier entropy of Boolean functions is uniformly bounded by their total influence. It has been verified for canonical examples such as disjoint tribes and for some classes of Boolean functions such as symmetric functions and read-$k$ decision trees (with a constant that depends linearly on $k$). In this note we present new classes of Boolean functions that verify the FEI conjecture. The key element is an inequality controlling the difference between the entropy of a function $f$ and the average of the entropies of $f^{\pm}$, the sub-functions obtained by setting $x_m=\pm1$ for some $m$, by the $m$-influence of $f$. If this key inequality were to hold for Boolean functions, then the full FEI conjecture would follow by induction. We introduce the notion of a stopping binary tree and observe that functions that satisfy the key inequality at the branching nodes of the tree and the FEI conjecture at the stopping nodes will satisfy the FEI conjecture. We identify some classes of functions that fit this framework: the $δ$-tribes functions, the monotone Boolean functions with the tribe separation property, and the Boolean functions with the semi-separation property, and, and, along the way, demonstrate some results that we hope the experts in this fascinating field might find useful. △ Less

Submitted 9 June, 2026; v1 submitted 29 May, 2026; originally announced June 2026.

Comments: 30 pages

MSC Class: 94D10 06E30 43A75

An Approach for Thyroid Nodule Analysis Using Thermographic Images

Authors: J. R. González, É. O. Rodrigues, C. P. Damião, C. A. P. Fontes, A. C. Silva, A. C. Paiva, H. Li, C. Du, A. Conci

Abstract: Thyroid cancer is said to be the second most common type of cancer in female individuals and the third in males by 2030, according to projections. In general, detecting cancer in its early stages improves the chance of survival of the individual. Thermography is a diagnostic tool that has been increasingly used to detect cancer and abnormalities, including that of thyroid. Various methods to segme… ▽ More Thyroid cancer is said to be the second most common type of cancer in female individuals and the third in males by 2030, according to projections. In general, detecting cancer in its early stages improves the chance of survival of the individual. Thermography is a diagnostic tool that has been increasingly used to detect cancer and abnormalities, including that of thyroid. Various methods to segment and detect hot regions in thermograms and, consequently, to detect suspicious tissues present in these images have been proposed. It is well known that medical diagnosis yields a great deal of information. Thus, physicians have to comprehensively analyse and evaluate this information in a short period of time, which is infeasible in most cases. In this work, we perform a general review of thermography , focusing on the thyroid analysis. We propose protocols for image acquisiton and an autonomous registration for thyroid images. We also perform analyses of the image data, which include feature extraction, image processing, and a possible approach for classification of healthy or unhealthy patients. In summary, this work presents a pilot project for detection of tumors in our university hospital, which is part of an effort to support preventive medical actions in our endocrinology department. Under some future adjustments, this project will be submitted for approval by the ethics and research committee of Hospital Universitário Antonio Pedro at Universidade Federal Fluminense (HUAP-UFF) and to the Brazilian Ministry of Health Ethical committee under the name: Evaluation of the importance of thermography to aid diagnosis of thyroid nodules of patients in HUAP-UFF (in Portuguese: Avaliação da importância da termografia no auxílio à investigação diagnóstica de nódulos tireoidianos em pacientes acompanhados no HUAP-UFF). △ Less

Submitted 27 May, 2026; originally announced May 2026.

Journal ref: Application of Infrared to Biomedical Sciences 2017

Recon: Reconstruction-Guided Reasoning Synthesis for User Modeling

Authors: Alan Zhu, Mihran Miroyan, Carolyn Wang, Andrew Zhou, Lisa Dunlap, Narges Norouzi, Joseph E. Gonzalez

Abstract: User modeling aims to use language models (LMs) to mimic an individual's behavior from a corpus of past context-action pairs (e.g., conversation turns), enabling the simulation of users in settings like behavioral science, human-AI collaboration, and market research. Recent approaches augment these corpora with synthesized reasoning traces, typically generated by conditioning on both context and a… ▽ More User modeling aims to use language models (LMs) to mimic an individual's behavior from a corpus of past context-action pairs (e.g., conversation turns), enabling the simulation of users in settings like behavioral science, human-AI collaboration, and market research. Recent approaches augment these corpora with synthesized reasoning traces, typically generated by conditioning on both context and action. However, such conditioning constitutes post-hoc rationalization rather than reasoning: the trace is guaranteed to justify the action, but may not encode the underlying latent causal decision paths. We propose Recon, which uses action reconstruction to score reasoning traces by their predictive power: given a context and candidate reasoning, a reconstruction model predicts the action, and reconstruction fidelity determines reasoning quality. Across four domains, Recon achieves a 54.7% win rate over Backward Synthesis, a standard post-hoc rationalization baseline. Further, we find that training a reasoning synthesis model with rewards derived from Recon improves downstream user modeling performance, achieving a win rate of up to 70.0% over baselines. We further show that Recon-synthesized reasoning transfers across models, and improves user modeling beyond the reconstruction model. Our work demonstrates that post-hoc rationalization is insufficient for reasoning synthesis, and that useful and interpretable reasoning should naturally elicit the action from the context. △ Less

Submitted 26 May, 2026; originally announced May 2026.

Inference Time Context Sparsity: Illusion or Opportunity?

Authors: Sahil Joshi, Prithvi Dixit, Agniva Chowdhury, Anshumali Shrivastava, Joseph E. Gonzalez, Ion Stoica, Kumar Krishna Agrawal, Aditya Desai

Abstract: Sparsity has long been a central theme in LLM efficiency, but its role in context processing remains unresolved. As LLM workloads shift toward longer contexts and agentic interactions, the compute and memory bottlenecks of attention become increasingly critical, raising the question of whether these constraints are fundamental. Our position is that these constraints are artificial and unnecessary,… ▽ More Sparsity has long been a central theme in LLM efficiency, but its role in context processing remains unresolved. As LLM workloads shift toward longer contexts and agentic interactions, the compute and memory bottlenecks of attention become increasingly critical, raising the question of whether these constraints are fundamental. Our position is that these constraints are artificial and unnecessary, and that the future of LLM inference lies in extreme but principled sparsity along the context dimension. This position is supported by several strands of empirical and theoretical evidence. First, we find the insistence on dense attention unreasonable, since in a long context a query effectively projects O(N) attention information into a hidden space of dimension d << N, making the process inherently lossy. Second, we perform an extensive study of sparsity in LLMs spanning 20 models across five model families, varying context lengths, and different sparsity levels. We empirically demonstrate a strong trend: current LLMs, despite not being trained for context sparsity, are remarkably robust to inference-time decode sparsity across tasks of varying complexity, including retrieval, multi-hop QA, mathematical reasoning, and agentic coding. Importantly, we also show that current hardware is already sufficient to realize substantial gains from this sparsity. For example, our sparse decode kernels accelerate large-context processing by up to 10x over FlashInfer at 50x sparsity levels on hardware such as the H100. Overall, these results position extreme context sparsity not as a heuristic, but as a principled foundation for LLM inference, training, and architecture design: one that is both feasible and beneficial, and a compelling direction for future systems. △ Less

Submitted 22 May, 2026; originally announced May 2026.

Comments: 19 pages, 8 figures

optimize_anything: A Universal API for Optimizing any Text Parameter

Authors: Lakshya A Agrawal, Donghyun Lee, Shangyin Tan, Wenjie Ma, Karim Elmaaroufi, Rohit Sandadi, Sanjit A. Seshia, Koushik Sen, Dan Klein, Ion Stoica, Joseph E. Gonzalez, Omar Khattab, Alexandros G. Dimakis, Matei Zaharia

Abstract: Can a single LLM-based optimization system match specialized tools across fundamentally different domains? We show that when optimization problems are formulated as improving a text artifact evaluated by a scoring function, a single AI-based optimization system-supporting single-task search, multi-task search with cross-problem transfer, and generalization to unseen inputs-achieves state-of-the-ar… ▽ More Can a single LLM-based optimization system match specialized tools across fundamentally different domains? We show that when optimization problems are formulated as improving a text artifact evaluated by a scoring function, a single AI-based optimization system-supporting single-task search, multi-task search with cross-problem transfer, and generalization to unseen inputs-achieves state-of-the-art results across six diverse tasks. Our system discovers agent architectures that nearly triple Gemini Flash's ARC-AGI accuracy (32.5% to 89.5%), finds scheduling algorithms that cut cloud costs by 40%, generates CUDA kernels where 87% match or beat PyTorch, and outperforms AlphaEvolve's reported circle packing solution (n=26). Ablations across three domains reveal that actionable side information yields faster convergence and substantially higher final scores than score-only feedback, and that multi-task search outperforms independent optimization given equivalent per-problem budget through cross-task transfer, with benefits scaling with the number of related tasks. Together, we show for the first time that text optimization with LLM-based search is a general-purpose problem-solving paradigm, unifying tasks traditionally requiring domain-specific algorithms under a single framework. We open-source optimize\_anything with support for multiple backends as part of the GEPA project at https://github.com/gepa-ai/gepa . △ Less

Submitted 19 May, 2026; originally announced May 2026.

Comments: 16 pages, 11 figures; Blog: https://gepa-ai.github.io/gepa/blog/2026/02/18/introducing-optimize-anything/

MSC Class: 68T05; 68T07; 68T20; 68T50; 68W50; 90C26; 90C59; 52C15 ACM Class: I.2.6; I.2.7; I.2.8; I.2.11; D.1.2; D.2.2; G.1.6; F.2.2

Journal ref: Proceedings of the ACM Conference on AI and Agentic Systems (CAIS 26), May 26-29, 2026, San Jose, CA, USA

Concurrency without Model Changes: Future-based Asynchronous Function Calling for LLMs

Authors: Guangyu Feng, Huanzhi Mao, Prabal Dutta, Joseph E. Gonzalez

Abstract: Function calling, also known as tool use, is a core capability of modern LLM agents but is typically constrained by synchronous execution semantics. Under these semantics, LLM decoding is blocked until each function call completes, resulting in increasing end-to-end latency. In this work, we introduce AsyncFC, a pure execution-layer framework that decouples LLM decoding from function execution, en… ▽ More Function calling, also known as tool use, is a core capability of modern LLM agents but is typically constrained by synchronous execution semantics. Under these semantics, LLM decoding is blocked until each function call completes, resulting in increasing end-to-end latency. In this work, we introduce AsyncFC, a pure execution-layer framework that decouples LLM decoding from function execution, enabling overlap between model decoding and function execution as well as inter-function parallelism when dependencies permit. AsyncFC layers over existing models and unmodified function implementations, requiring no fine-tuning or changes to the standard synchronous function-calling protocol. Across standard function-calling benchmarks and adapted software engineering benchmarks, AsyncFC significantly reduces end-to-end task completion time while preserving task accuracy. Furthermore, these results reveal that LLMs possess a native capability to reason over symbolic futures that represent unresolved execution results, enabling an asynchronous paradigm for model-tool interaction. △ Less

Submitted 14 May, 2026; originally announced May 2026.

FrontierSmith: Synthesizing Open-Ended Coding Problems at Scale

Authors: Runyuan He, Qiuyang Mang, Shang Zhou, Kaiyuan Liu, Hanchen Li, Huanzhi Mao, Qizheng Zhang, Zerui Li, Bo Peng, Lufeng Cheng, Tianfu Fu, Yichuan Wang, Wenhao Chai, Jingbo Shang, Alex Dimakis, Joseph E. Gonzalez, Alvin Cheung

Abstract: Many real-world coding challenges are open-ended and admit no known optimal solution. Yet, recent progress in LLM coding has focused on well-defined tasks such as feature implementation, bug fixing, and competitive programming. Open-ended coding remains a weak spot for LLMs, largely because open-ended training problems are scarce and expensive to construct. Our goal is to synthesize open-ended cod… ▽ More Many real-world coding challenges are open-ended and admit no known optimal solution. Yet, recent progress in LLM coding has focused on well-defined tasks such as feature implementation, bug fixing, and competitive programming. Open-ended coding remains a weak spot for LLMs, largely because open-ended training problems are scarce and expensive to construct. Our goal is to synthesize open-ended coding problems at scale to train stronger LLM coders. We introduce FrontierSmith, an automated system for iteratively evolving open-ended problems from existing closed-ended coding tasks. Starting from competitive programming problems, FrontierSmith generates candidate open-ended variants by changing the problems'goals, restricting outputs, and generalizing inputs. It then uses a quantitative idea divergence metric to select problems that elicit genuinely diverse approaches from different solvers. Agents then generate test cases and verifiers for the surviving candidates. On two open-ended coding benchmarks, training on our synthesized data yields substantial gains over the base models: Qwen3.5-9B improves by +8.82 score on FrontierCS and +306.36 (Elo-rating-based performance) on ALE-bench; Qwen3.5-27B improves by +12.12 and +309.12, respectively. The synthesized problems also make agents take more turns and use more tokens, similar to human-curated ones, suggesting that closed-ended seeds can be a practical starting point for long-horizon coding data. △ Less

Submitted 14 May, 2026; originally announced May 2026.

Generating synthetic computed tomography for radiotherapy: SynthRAD2025 challenge report

Authors: Viktor Rogowski, Maarten L. Terpstra, Niklas Wahl, Florian Kamp, Erik van der Bijl, Arthur Jr. Galapon, Christopher Kurz, Bowen Xin, Zhengxiang Sun, Hollie Min, Gregg Belous, Jason Dowling, Yan Xia, Siyuan Mei, Fuxin Fan, Arthur Longuefosse, Javier Sequeiro Gonzalez, Miguel Diaz Benito, Alvaro Garcia Martin, Fabien Baldacci, Valentin Boussot, Cédric Hémon, Jean-Claude Nunes, Jean-Louis Dillenseger, Zhiyuan Zhang , et al. (8 additional authors not shown)

Abstract: Radiation therapy (RT) requires precise dose delivery over multiple fractions, with CT fundamental for treatment planning due to its electron density information. Repeated CT acquisitions impose radiation exposure and logistical burdens, MRI lacks electron density, and cone-beam CT (CBCT) requires correction for dose calculation. Synthetic CT (sCT) generation addresses these by converting MRI or C… ▽ More Radiation therapy (RT) requires precise dose delivery over multiple fractions, with CT fundamental for treatment planning due to its electron density information. Repeated CT acquisitions impose radiation exposure and logistical burdens, MRI lacks electron density, and cone-beam CT (CBCT) requires correction for dose calculation. Synthetic CT (sCT) generation addresses these by converting MRI or CBCT into CT-equivalent images with accurate Hounsfield Unit (HU) values, enabling MRI-only RT and CBCT-based adaptive workflows. Building on SynthRAD2023, SynthRAD2025 benchmarked sCT methods on 2,362 patients from five European centers across head and neck, thorax, and abdomen. Two tasks: MRI-to-CT (890 cases) and CBCT-to-CT (1,472 cases), evaluated via image similarity (MAE, PSNR, MS-SSIM), segmentation (Dice, HD95), and dosimetric metrics from photon and proton plans. With 803 participants and 12/13 valid submissions, Task 1 top performance reached MAE $64.8\pm21.3$ HU, PSNR $\sim$30 dB, MS-SSIM $\sim$0.936, Dice 0.79, photon $γ_{2\%/2\text{mm}}>98\%$, proton $γ\approx85\%$. Task 2 improved: MAE $48.3\pm13.4$ HU, PSNR 32.6 dB, MS-SSIM 0.968, Dice 0.86, photon $γ>99\%$, proton $γ\approx89\%$. Strong image--segmentation correlations ($ρ=0.78$--$0.79$) but moderate dose correlations confirmed image quality is insufficient as a dosimetric surrogate. Head-and-neck cases were most consistent; thoracic and abdominal cases showed greater variability. Residual errors at tissue interfaces propagate along beam paths, affecting proton dose more than photon. SynthRAD2025 demonstrates that deep learning yields clinically relevant sCTs, especially for CBCT-to-CT, while identifying persistent MRI-to-CT challenges and underscoring dose-based evaluation as essential for clinical validation. △ Less

Submitted 13 May, 2026; originally announced May 2026.

Comments: 59 pages total: 26 pages main article + supplementary material; 8 figures in the main manuscript and 3 supplementary figures. Currently under review at the journal Medical Image Analysis (MIA)

Learning, Fast and Slow: Towards LLMs That Adapt Continually

Authors: Rishabh Tiwari, Kusha Sareen, Lakshya A Agrawal, Joseph E. Gonzalez, Matei Zaharia, Kurt Keutzer, Inderjit S Dhillon, Rishabh Agarwal, Devvrit Khatri

Abstract: Large language models (LLMs) are trained for downstream tasks by updating their parameters (e.g., via RL). However, updating parameters forces them to absorb task-specific information, which can result in catastrophic forgetting and loss of plasticity. In contrast, in-context learning with fixed LLM parameters can cheaply and rapidly adapt to task-specific requirements (e.g., prompt optimization),… ▽ More Large language models (LLMs) are trained for downstream tasks by updating their parameters (e.g., via RL). However, updating parameters forces them to absorb task-specific information, which can result in catastrophic forgetting and loss of plasticity. In contrast, in-context learning with fixed LLM parameters can cheaply and rapidly adapt to task-specific requirements (e.g., prompt optimization), but cannot by itself typically match the performance gains available through updating LLM parameters. There is no good reason for restricting learning to being in-context or in-weights. Moreover, humans also likely learn at different time scales (e.g., System 1 vs 2). To this end, we introduce a fast-slow learning framework for LLMs, with model parameters as "slow" weights and optimized context as "fast" weights. These fast "weights" can learn from textual feedback to absorb the task-specific information, while allowing slow weights to stay closer to the base model and persist general reasoning behaviors. Fast-Slow Training (FST) is up to 3x more sample-efficient than only slow learning (RL) across reasoning tasks, while consistently reaching a higher performance asymptote. Moreover, FST-trained models remain closer to the base LLM (up to 70% less KL divergence), resulting in less catastrophic forgetting than RL-training. This reduced drift also preserves plasticity: after training on one task, FST trained models adapt more effectively to a subsequent task than parameter-only trained models. In continual learning scenarios, where task domains change on the fly, FST continues to acquire each new task while parameter-only RL stalls. △ Less

Submitted 14 May, 2026; v1 submitted 12 May, 2026; originally announced May 2026.

Comments: 29 pages, 14 figures, including appendix; Blog post: https://gepa-ai.github.io/gepa/blog/2026/05/11/learning-fast-and-slow/

ACM Class: I.2.6; I.2.7; I.2.8; I.2.4

Cooperate to Compete: Strategic Coordination in Multi-Agent Conquest

Authors: Abigail O'Neill, Alan Zhu, Mihran Miroyan, Narges Norouzi, Joseph E. Gonzalez

Abstract: Language Model (LM)-based agents remain largely untested in mixed-motive settings where agents must leverage short-term cooperation for long-term competitive goals (e.g., multi-party politics). We introduce Cooperate to Compete (C2C), a multi-agent environment where players can engage in private negotiations while competing to be the first to achieve their secret objective. Players have asymmetric… ▽ More Language Model (LM)-based agents remain largely untested in mixed-motive settings where agents must leverage short-term cooperation for long-term competitive goals (e.g., multi-party politics). We introduce Cooperate to Compete (C2C), a multi-agent environment where players can engage in private negotiations while competing to be the first to achieve their secret objective. Players have asymmetric objectives and negotiations are non-binding, allowing alliances to form and break as players' short-term interests align and diverge. We run AI only games and conduct a user study pitting human players against AI opponents. We identify significant differences between human and AI negotiation behaviors, finding that humans favor lower-complexity deals and are significantly less reliable partners compared to LM-based agents. We also find that humans are more aggressive negotiators, accepting deals without a counteroffer only 56.3% of the time compared to 67.6% for LM-based agents. Through targeted prompting inspired by these findings, we modify agents' negotiation behavior and improve win rates from 22.2% to 32.7%. We run over 1,100 games with over 16,000 private conversations totaling 15.2 million tokens and over 150,000 player actions. Our results establish C2C as a testbed for studying and building LM-based agents that can navigate the sophisticated coordination required for real-world deployments. The game, code, and dataset may be found at https://negotiationgame.io/c2c. △ Less

Submitted 27 April, 2026; originally announced April 2026.

Neural posterior estimation of the neutrino direction in IceCube using transformer-encoded normalizing flows on the sphere

Authors: R. Abbasi, M. Ackermann, J. Adams, J. A. Aguilar, M. Ahlers, J. M. Alameddine, S. Ali, N. M. Amin, K. Andeen, C. Argüelles, Y. Ashida, S. Athanasiadou, S. N. Axani, R. Babu, X. Bai, A. Balagopal V., S. W. Barwick, V. Basu, R. Bay, J. J. Beatty, J. Becker Tjus, P. Behrens, J. Beise, C. Bellenghi, S. Benkel , et al. (389 additional authors not shown)

Abstract: IceCube is a cubic-kilometer-scale neutrino detector located at the geographic South Pole. A precise directional reconstruction of IceCube neutrinos is vital for associations with astronomical objects. In this context, we discuss neural posterior estimation of the neutrino direction via a transformer encoder that maps to a normalizing flow on the 2-sphere. It achieves a new state-of-the-art angula… ▽ More IceCube is a cubic-kilometer-scale neutrino detector located at the geographic South Pole. A precise directional reconstruction of IceCube neutrinos is vital for associations with astronomical objects. In this context, we discuss neural posterior estimation of the neutrino direction via a transformer encoder that maps to a normalizing flow on the 2-sphere. It achieves a new state-of-the-art angular resolution for the two main event morphologies in IceCube - tracks and showers - while being significantly faster than traditional B-spline-based likelihood reconstructions. All-sky scans can be performed within seconds rather than hours, and take constant computation time, regardless of whether the posterior extent is arc-minutes or spans the whole sky. We utilize a combination of $C^2$-smooth rational-quadratic splines, scale transformations and rotations to define a novel spherical normalizing-flow distribution whose parameters are predicted as a whole as the output of the transformer encoder. We test several structural choices diverting from the vanilla transformer architecture. In particular, we find dual residual streams, nonlinear QKV projection and a separate class token with its own cross-attention processing to boost test-time performance. The angular resolution for both showers and tracks improves substantially over the whole trained energy range from 100 GeV to 100 PeV. At 100 TeV deposited energy, for example, the median angular resolution improves by a factor of $1.3$ for throughgoing tracks, by a factor of $1.7$ for showers and by a factor of $2.5$ for starting tracks compared to state-of-the art likelihood reconstructions based on B-splines. While previous machine-learning (ML) efforts have managed to obtain competitive shower resolutions, this is the first time an ML-based method outperforms likelihood-based muon reconstructions above 100 GeV. △ Less

Submitted 21 April, 2026; originally announced April 2026.

Semantic Reality: Interactive Context-Aware Visualization of Inter-Object Relationships in Augmented Reality

Authors: Xiaoan Liu, Eric J Gonzalez, Nels Numan, Andrea Colaço, Lucy Abramyan, Chen Zhu-Tian, Ryo Suzuki, Mar Gonzalez-Franco

Abstract: Bridging the physical and digital world through interaction remains a core challenge in augmented reality (AR). Existing systems target single objects, limiting support for planning, comparison, and assembly tasks that depend on relationships among multiple items. We present Semantic Reality, an AR system focused on surfacing inter-object connectivity and making it interactive. Leveraging multimod… ▽ More Bridging the physical and digital world through interaction remains a core challenge in augmented reality (AR). Existing systems target single objects, limiting support for planning, comparison, and assembly tasks that depend on relationships among multiple items. We present Semantic Reality, an AR system focused on surfacing inter-object connectivity and making it interactive. Leveraging multimodal reasoning, spatial anchoring, and physical action recognition, Semantic Reality maintains a persistent model of objects around the user and their relationships. Connections are visualized in-situ to highlight compatibility, reveal next steps, and reduce ambiguity during tasks. We contribute a connectivity-centered interaction paradigm and a system architecture that couples anchor tracking, action sensing, and model inference to construct a live connectivity graph. In an exploratory study comparing Semantic Reality to a single-object baseline, participants reported clearer inter-object understanding and higher engagement and satisfaction, without increased workload. A scenario study illustrates where connectivity aids planning, sequencing, and disambiguation. △ Less

Submitted 6 April, 2026; originally announced April 2026.

Comments: 15 pages, 15 figures, 4 tables

ACM Class: H.5.2; H.2.11; I.2.10

Combee: Scaling Prompt Learning for Self-Improving Language Model Agents

Authors: Hanchen Li, Runyuan He, Qizheng Zhang, Changxiu Ji, Qiuyang Mang, Xiaokun Chen, Lakshya A Agrawal, Wei-Liang Liao, Eric Yang, Alvin Cheung, James Zou, Kunle Olukotun, Ion Stoica, Joseph E. Gonzalez

Abstract: Recent advances in prompt learning allow large language model agents to acquire task-relevant knowledge from inference-time context without parameter changes. For example, existing methods (like ACE or GEPA) can learn system prompts to improve accuracy based on previous agent runs. However, these methods primarily focus on single-agent or low-parallelism settings. This fundamentally limits their a… ▽ More Recent advances in prompt learning allow large language model agents to acquire task-relevant knowledge from inference-time context without parameter changes. For example, existing methods (like ACE or GEPA) can learn system prompts to improve accuracy based on previous agent runs. However, these methods primarily focus on single-agent or low-parallelism settings. This fundamentally limits their ability to efficiently learn from a large set of collected agentic traces. It would be efficient and beneficial to run prompt learning in parallel to accommodate the growing trend of learning from many agentic traces or parallel agent executions. Yet without a principled strategy for scaling, current methods suffer from quality degradation with high parallelism. To improve both the efficiency and quality of prompt learning, we propose Combee, a novel framework to scale parallel prompt learning for self-improving agents. Combee speeds up learning and enables running many agents in parallel while learning from their aggregate traces without quality degradation. To achieve this, Combee leverages parallel scans and employs an augmented shuffle mechanism; Combee also introduces a dynamic batch size controller to balance quality and delay. Evaluations on AppWorld, Terminal-Bench, Formula, and FiNER demonstrate that Combee achieves up to 17x speedup over previous methods with comparable or better accuracy and equivalent cost. △ Less

Submitted 5 April, 2026; originally announced April 2026.

VisionClaw: Always-On AI Agents through Smart Glasses

Authors: Xiaoan Liu, DaeHo Lee, Eric J Gonzalez, Mar Gonzalez-Franco, Ryo Suzuki

Abstract: We present VisionClaw, an always-on wearable AI agent that integrates live egocentric perception with agentic task execution. Running on Meta Ray-Ban smart glasses, VisionClaw continuously perceives real-world context and enables in-situ, speech-driven action initiation and delegation via OpenClaw AI agents. Therefore, users can directly execute tasks through the smart glasses, such as adding real… ▽ More We present VisionClaw, an always-on wearable AI agent that integrates live egocentric perception with agentic task execution. Running on Meta Ray-Ban smart glasses, VisionClaw continuously perceives real-world context and enables in-situ, speech-driven action initiation and delegation via OpenClaw AI agents. Therefore, users can directly execute tasks through the smart glasses, such as adding real-world objects to an Amazon cart, generating notes from physical documents, receiving meeting briefings on the go, creating events from posters, or controlling IoT devices. We evaluate VisionClaw through a controlled laboratory study (N=12) and a longitudinal deployment study (N=5). Results show that integrating perception and execution enables faster task completion and reduces interaction overhead compared to non-always-on and non-agent baselines. Beyond performance gains, deployment findings reveal a shift in interaction: tasks are initiated opportunistically during ongoing activities, and execution is increasingly delegated rather than manually controlled. These results suggest a new paradigm for wearable AI agents, where perception and action are continuously coupled to support situated, hands-free interaction. △ Less

Submitted 8 April, 2026; v1 submitted 3 April, 2026; originally announced April 2026.

Comments: 17 pages, 11 figures, plus appendix

ACM Class: H.5.2; H.2.11; I.2.10

Sticky and Magnetic: Evaluating Error Correction and User Adaptation in Gaze and Pinch Interaction

Authors: Jazmin Collins, Prasanthi Gurumurthy, Eric J. Gonzalez, Mar Gonzalez-Franco

Abstract: The gaze-and-pinch framework offers a high-fidelity interaction modality for spatial computing in virtual reality (VR), yet it remains vulnerable to coordination errors--timing misalignments between gaze fixation and pinch gestures. These errors are categorized into two types: late triggers (gaze leaves a target before pinch) and early triggers (pinch before gaze arrival on target). While late tri… ▽ More The gaze-and-pinch framework offers a high-fidelity interaction modality for spatial computing in virtual reality (VR), yet it remains vulnerable to coordination errors--timing misalignments between gaze fixation and pinch gestures. These errors are categorized into two types: late triggers (gaze leaves a target before pinch) and early triggers (pinch before gaze arrival on target). While late triggers are well-studied, early triggers lack robust solutions. We investigate two heuristics--STICKY selection (temporal buffer) and MAGNETIC selection (spatial field)--to mitigate these errors. A within-subjects study (N = 9) on the Samsung Galaxy XR evaluated these heuristics against a baseline. Findings indicate that while throughput and selection time remained stable, the heuristics fundamentally shifted user behavior and significantly reduced errors during selection. Notably, MAGNETIC selection induced an "offloading" effect where users traded precision for speed. Additionally, the heuristics reclassified ambiguous failures as explainable coordination errors. We provide recommendations for selection heuristics that enhance interaction speed and cognitive agency in virtual reality. △ Less

Submitted 27 March, 2026; originally announced March 2026.

Comments: 5 page, 5 figures, Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26), April 13-17, 2026, Barcelona, Spain. ACM

ACM Class: H.5.2; I.3.7

Journal ref: In Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA '26). 2026

SurfaceXR: Fusing Smartwatch IMUs and Egocentric Hand Pose for Seamless Surface Interactions

Authors: Vasco Xu, Brian Chen, Eric J. Gonzalez, Andrea Colaço, Henry Hoffmann, Mar Gonzalez-Franco, Karan Ahuja

Abstract: Mid-air gestures in Extended Reality (XR) often cause fatigue and imprecision. Surface-based interactions offer improved accuracy and comfort, but current egocentric vision methods struggle due to hand tracking challenges and unreliable surface plane estimation. We introduce SurfaceXR, a sensor fusion approach combining headset-based hand tracking with smartwatch IMU data to enable robust inputs o… ▽ More Mid-air gestures in Extended Reality (XR) often cause fatigue and imprecision. Surface-based interactions offer improved accuracy and comfort, but current egocentric vision methods struggle due to hand tracking challenges and unreliable surface plane estimation. We introduce SurfaceXR, a sensor fusion approach combining headset-based hand tracking with smartwatch IMU data to enable robust inputs on everyday surfaces. Our insight is that these modalities are complementary: hand tracking provides 3D positional data while IMUs capture high-frequency motion. A 21-participant study validates SurfaceXR's effectiveness for touch tracking and 8-class gesture recognition, demonstrating significant improvements over single-modality approaches. △ Less

Submitted 19 March, 2026; originally announced March 2026.

Comments: Accepted to IEEE VR 2026 as a TVCG journal paper

M$^2$RNN: Non-Linear RNNs with Matrix-Valued States for Scalable Language Modeling

Authors: Mayank Mishra, Shawn Tan, Ion Stoica, Joseph Gonzalez, Tri Dao

Abstract: Transformers are highly parallel but are limited to computations in the TC$^0$ complexity class, excluding tasks such as entity tracking and code execution that provably require greater expressive power. Motivated by this limitation, we revisit non-linear Recurrent Neural Networks (RNNs) for language modeling and introduce Matrix-to-Matrix RNN (M$^2$RNN): an architecture with matrix-valued hidden… ▽ More Transformers are highly parallel but are limited to computations in the TC$^0$ complexity class, excluding tasks such as entity tracking and code execution that provably require greater expressive power. Motivated by this limitation, we revisit non-linear Recurrent Neural Networks (RNNs) for language modeling and introduce Matrix-to-Matrix RNN (M$^2$RNN): an architecture with matrix-valued hidden states and expressive non-linear state transitions. We demonstrate that the language modeling performance of non-linear RNNs is limited by their state size, and show how the state size expansion mechanism enables efficient use of tensor cores. Empirically, M$^2$RNN achieves perfect state tracking generalization at sequence lengths not seen during training. These benefits also translate to large-scale language modeling. In hybrid settings that interleave recurrent layers with attention, Hybrid M$^2$RNN outperforms equivalent Gated DeltaNet hybrids by $0.4$-$0.5$ perplexity points on a 7B MoE model, while using $3\times$ smaller state sizes for the recurrent layers. Notably, replacing even a single recurrent layer with M$^2$RNN in an existing hybrid architecture yields accuracy gains comparable to Hybrid M$^2$RNN with minimal impact on training throughput. Further, the Hybrid Gated DeltaNet models with a single M$^2$RNN layer also achieve superior long-context generalization, outperforming state-of-the-art hybrid linear attention architectures by up to $8$ points on LongBench. Together, these results establish non-linear RNN layers as a compelling building block for efficient and scalable language models. △ Less

Submitted 13 May, 2026; v1 submitted 15 March, 2026; originally announced March 2026.

Towards Fluent Interaction with Cyber-Physical Architecture

Authors: Jesse T. Gonzalez, Neeta Khanuja, Michael Li, Maggie Guo, Layomi Olaitan, Emily Lau, Jennifer Pugh, Alexandra Ion, Scott E. Hudson

Abstract: What happens when your walls begin to move? This paper explores the design of human-robot interaction for architectural-scale, shape-changing environments. We present findings from two studies: (1) a series of speculative design workshops (N=20) that uncovered aspirational visions for these spaces, and (2) a task-based Wizard-of-Oz elicitation study (N=12) that grounded these visions in the challe… ▽ More What happens when your walls begin to move? This paper explores the design of human-robot interaction for architectural-scale, shape-changing environments. We present findings from two studies: (1) a series of speculative design workshops (N=20) that uncovered aspirational visions for these spaces, and (2) a task-based Wizard-of-Oz elicitation study (N=12) that grounded these visions in the challenges of practical interaction. Our workshop findings reveal a complex landscape of user desires, exposing critical tensions between proactive automation and the preservation of user autonomy, and between personalization and public ownership. Our elicitation study reveals a set of core interaction challenges related to multimodal collaboration; and, most critically: suggests the need for a modality-agnostic model of evolving user intent. We conclude with a set of grounded proposals for creating robotic environments that are collaborative and trusted partners in everyday life. △ Less

Submitted 13 March, 2026; originally announced March 2026.

Journal ref: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems

Navig-AI-tion: Navigation by Contextual AI and Spatial Audio

Authors: Mathias N. Lystbæk, Haley Adams, Ranjith Kagathi Ananda, Eric J Gonzalez, Luca Ballan, Qiuxuan Wu, Andrea Colaço, Peter Tan, Mar Gonzalez-Franco

Abstract: Audio-only walking navigation can leave users disoriented, relying on vague cardinal directions and lacking real-time environmental context, leading to frequent errors. To address this, we present a novel system that integrates a Vision Language Model (VLM) with a spatial audio cue. Our system extracts environmental landmarks to anchor navigation instructions and, crucially, provides a directional… ▽ More Audio-only walking navigation can leave users disoriented, relying on vague cardinal directions and lacking real-time environmental context, leading to frequent errors. To address this, we present a novel system that integrates a Vision Language Model (VLM) with a spatial audio cue. Our system extracts environmental landmarks to anchor navigation instructions and, crucially, provides a directional spatial audio signal when the user faces the wrong direction, indicating the precise turn direction. In a user study (n=12), the spatial audio cue with VLM reduced route deviations compared to both VLM-only and Google Maps (audio-only) baseline systems. Users reported that the spatial audio cue effectively supported orientation and that landmark-anchored instructions provided a better navigation experience over audio-only Google Maps. This work serves as an initial look at the utility of future audio-only navigation systems for incorporating directional cues, especially real-time corrective spatial audio. △ Less

Submitted 8 April, 2026; v1 submitted 13 March, 2026; originally announced March 2026.

Comments: 5 pages, 2 figures, to be published in Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA '26), 6 pages appendix

Monitoring and Prediction of Mood in Elderly People during Daily Life Activities

Authors: Daniel Bautista-Salinas, Joaquín Roca González, Inmaculada Méndez, Oscar Martinez Mozos

Abstract: We present an intelligent wearable system to monitor and predict mood states of elderly people during their daily life activities. Our system is composed of a wristband to record different physiological activities together with a mobile app for ecological momentary assessment (EMA). Machine learning is used to train a classifier to automatically predict different mood states based on the smart ban… ▽ More We present an intelligent wearable system to monitor and predict mood states of elderly people during their daily life activities. Our system is composed of a wristband to record different physiological activities together with a mobile app for ecological momentary assessment (EMA). Machine learning is used to train a classifier to automatically predict different mood states based on the smart band only. Our approach shows promising results on mood accuracy and provides results comparable with the state of the art in the specific detection of happiness and activeness. △ Less

Submitted 11 March, 2026; originally announced March 2026.

Comments: This is the authors' manuscript. The final published article is available at https://doi.org/10.1109/EMBC.2019.8857847

Journal ref: Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Berlin, Germany, 2019, pp. 6930-6934

World Mouse: Exploring Interactions with a Cross-Reality Cursor

Authors: Esen K. Tütüncü, Mar Gonzalez-Franco, Khushman Patel, Eric J. Gonzalez

Abstract: As Extended Reality (XR) systems increasingly map and understand the physical world, interacting with these blended representations remains challenging. The current push for "natural" inputs has its trade-offs: touch is limited by human reach and fatigue, while gaze often lacks the precision for fine interaction. To bridge this gap, we introduce World Mouse, a cross-reality cursor that reinterpret… ▽ More As Extended Reality (XR) systems increasingly map and understand the physical world, interacting with these blended representations remains challenging. The current push for "natural" inputs has its trade-offs: touch is limited by human reach and fatigue, while gaze often lacks the precision for fine interaction. To bridge this gap, we introduce World Mouse, a cross-reality cursor that reinterprets the familiar 2D desktop mouse for complex 3D scenes. The system is driven by two core mechanisms: within-object interaction, which uses surface normals for precise cursor placement, and between-object navigation, which leverages interpolation to traverse empty space. Unlike previous virtual-only approaches, World Mouse leverages semantic segmentation and mesh reconstruction to treat physical objects as interactive surfaces. Through a series of prototypes, including object manipulation and screen-to-world transitions, we illustrate how cross-reality cursors may enable seamless interactions across real and virtual environments. △ Less

Submitted 11 March, 2026; originally announced March 2026.

Comments: 7 pages, 4 figures. CHI '26, April 13-17, 2026, Barcelona, Spain

SVG-EAR: Parameter-Free Linear Compensation for Sparse Video Generation via Error-aware Routing

Authors: Xuanyi Zhou, Qiuyang Mang, Shuo Yang, Haocheng Xi, Jintao Zhang, Huanzhi Mao, Joseph E. Gonzalez, Kurt Keutzer, Ion Stoica, Alvin Cheung

Abstract: Diffusion Transformers (DiTs) have become a leading backbone for video generation, yet their quadratic attention cost remains a major bottleneck. Sparse attention reduces this cost by computing only a subset of attention blocks. However, prior methods often either drop the remaining blocks, which incurs information loss, or rely on learned predictors to approximate them, introducing training overh… ▽ More Diffusion Transformers (DiTs) have become a leading backbone for video generation, yet their quadratic attention cost remains a major bottleneck. Sparse attention reduces this cost by computing only a subset of attention blocks. However, prior methods often either drop the remaining blocks, which incurs information loss, or rely on learned predictors to approximate them, introducing training overhead and potential output distribution shifting. In this paper, we show that the missing contributions can be recovered without training: after semantic clustering, keys and values within each block exhibit strong similarity and can be well summarized by a small set of cluster centroids. Based on this observation, we introduce SVG-EAR, a parameter-free linear compensation branch that uses the centroid to approximate skipped blocks and recover their contributions. While centroid compensation is accurate for most blocks, it can fail on a small subset. Standard sparsification typically selects blocks by attention scores, which indicate where the model places its attention mass, but not where the approximation error would be largest. SVG-EAR therefore performs error-aware routing: a lightweight probe estimates the compensation error for each block, and we compute exactly the blocks with the highest error-to-cost ratio while compensating for skipped blocks. We provide theoretical guarantees that relate attention reconstruction error to clustering quality, and empirically show that SVG-EAR improves the quality-efficiency trade-off and increases throughput at the same generation fidelity on video diffusion tasks. Overall, SVG-EAR establishes a clear Pareto frontier over prior approaches, achieving up to 1.77$\times$ and 1.93$\times$ speedups while maintaining PSNRs of up to 29.759 and 31.043 on Wan2.2 and HunyuanVideo, respectively. △ Less

Submitted 5 July, 2026; v1 submitted 9 March, 2026; originally announced March 2026.

MV-Fashion: Towards Enabling Virtual Try-On and Size Estimation with Multi-View Paired Data

Authors: Hunor Laczkó, Libang Jia, Loc-Phat Truong, Diego Hernández, Sergio Escalera, Jordi Gonzalez, Meysam Madadi

Abstract: Existing 4D human datasets fall short for fashion-specific research, lacking either realistic garment dynamics or task-specific annotations. Synthetic datasets suffer from a realism gap, whereas real-world captures lack the detailed annotations and paired data required for virtual try-on (VTON) and size estimation tasks. To bridge this gap, we introduce MV-Fashion, a large-scale, multi-view video… ▽ More Existing 4D human datasets fall short for fashion-specific research, lacking either realistic garment dynamics or task-specific annotations. Synthetic datasets suffer from a realism gap, whereas real-world captures lack the detailed annotations and paired data required for virtual try-on (VTON) and size estimation tasks. To bridge this gap, we introduce MV-Fashion, a large-scale, multi-view video dataset engineered for domain-specific fashion analysis. MV-Fashion features 3,273 sequences (72.5 million frames) from 80 diverse subjects wearing 3-10 outfits each. It is designed to capture complex, real-world garment dynamics, including multiple layers and varied styling (e.g. rolled sleeves, tucked shirt). A core contribution is a rich data representation that includes pixel-level semantic annotations, ground-truth material properties like elasticity, and 3D point clouds. Crucially for VTON applications, MV-Fashion provides paired data: multi-view synchronized captures of worn garments alongside their corresponding flat, catalogue images. We leverage this dataset to establish baselines for fashion-centric tasks, including virtual try-on, clothing size estimation, and novel view synthesis. The dataset is available at https://hunorlaczko.github.io/MV-Fashion . △ Less

Submitted 9 March, 2026; originally announced March 2026.

Break the Window: Exploring Spatial Decomposition of Webpages in XR

Authors: Chenyang Zhang, Tianjian Wei, Haoyang Yang, Mar Gonzalez-Franco, Yalong Yang, Eric J Gonzalez

Abstract: Most XR web browsers still present webpages as a single floating window, carrying over desktop design assumptions into immersive space. We explore an alternative by breaking the browser window and distributing a webpage into spatial UI chunks within a mixed-reality workspace. We present Break-the-Window (BTW), an exploratory prototype that spatially decomposes live, fully functional webpages into… ▽ More Most XR web browsers still present webpages as a single floating window, carrying over desktop design assumptions into immersive space. We explore an alternative by breaking the browser window and distributing a webpage into spatial UI chunks within a mixed-reality workspace. We present Break-the-Window (BTW), an exploratory prototype that spatially decomposes live, fully functional webpages into movable panels supporting mid-air and surface-attached placement, as well as direct touch and ray-based interaction. Through a formative study with XR practitioners and an exploratory qualitative study with 15 participants, we observed how spatial decomposition supports distributed attention and spatial meaning-making, while also surfacing challenges around coordination effort, interaction precision, and the lack of shared spatial UI conventions. This work invites discussion on how web interfaces might be reimagined for spatial computing beyond the single-window paradigm. △ Less

Submitted 2 March, 2026; originally announced March 2026.

Comments: 6 pages, 5 figures. Accepted as a CHI 2026 Extended Abstract (Poster). To appear in Proceedings of the ACM CHI Conference on Human Factors in Computing Systems (CHI '26)

SageBwd: A Trainable Low-bit Attention

Authors: Jintao Zhang, Marco Chen, Haoxu Wang, Kai Jiang, Ion Stoica, Joseph E. Gonzalez, Jianfei Chen, Jun Zhu

Abstract: Low-bit attention, such as SageAttention, has emerged as an effective approach for accelerating model inference, but its applicability to training remains poorly understood. In prior work, we introduced SageBwd, a trainable INT8 attention that quantizes six of seven attention matrix multiplications while preserving fine-tuning performance. However, SageBwd exhibited a persistent performance gap to… ▽ More Low-bit attention, such as SageAttention, has emerged as an effective approach for accelerating model inference, but its applicability to training remains poorly understood. In prior work, we introduced SageBwd, a trainable INT8 attention that quantizes six of seven attention matrix multiplications while preserving fine-tuning performance. However, SageBwd exhibited a persistent performance gap to full-precision attention (FPA) during pre-training. In this work, we investigate why this gap occurs and demonstrate that SageBwd matches full-precision attention during pretraining. Through experiments and theoretical analysis, we reach a few important insights and conclusions: (i) QK-norm is necessary for stable training at large tokens per step, (ii) quantization errors primarily arise from the backward-pass score gradient dS, (iii) reducing tokens per step enables SageBwd to match FPA performance in pre-training, and (iv) K-smoothing remains essential for training stability, while Q-smoothing provides limited benefit during pre-training. △ Less

Submitted 2 March, 2026; originally announced March 2026.

DS SERVE: A Framework for Efficient and Scalable Neural Retrieval

Authors: Jinjian Liu, Yichuan Wang, Xinxi Lyu, Rulin Shao, Joseph E. Gonzalez, Matei Zaharia, Sewon Min

Abstract: We present DS-Serve, a framework that transforms large-scale text datasets, comprising half a trillion tokens, into a high-performance neural retrieval system. DS-Serve offers both a web interface and API endpoints, achieving low latency with modest memory overhead on a single node. The framework also supports inference-time trade-offs between latency, accuracy, and result diversity. We anticipate… ▽ More We present DS-Serve, a framework that transforms large-scale text datasets, comprising half a trillion tokens, into a high-performance neural retrieval system. DS-Serve offers both a web interface and API endpoints, achieving low latency with modest memory overhead on a single node. The framework also supports inference-time trade-offs between latency, accuracy, and result diversity. We anticipate that DS-Serve will be broadly useful for a range of applications, including large-scale retrieval-augmented generation (RAG), training data attribution, training search agents, and beyond. △ Less

Submitted 16 December, 2025; originally announced February 2026.

K-Search: LLM Kernel Generation via Co-Evolving Intrinsic World Model

Authors: Shiyi Cao, Ziming Mao, Joseph E. Gonzalez, Ion Stoica

Abstract: Optimizing GPU kernels is critical for efficient modern machine learning systems yet remains challenging due to the complex interplay of design factors and rapid hardware evolution. Existing automated approaches typically treat Large Language Models (LLMs) merely as stochastic code generators within heuristic-guided evolutionary loops. These methods often struggle with complex kernels requiring co… ▽ More Optimizing GPU kernels is critical for efficient modern machine learning systems yet remains challenging due to the complex interplay of design factors and rapid hardware evolution. Existing automated approaches typically treat Large Language Models (LLMs) merely as stochastic code generators within heuristic-guided evolutionary loops. These methods often struggle with complex kernels requiring coordinated, multi-step structural transformations, as they lack explicit planning capabilities and frequently discard promising strategies due to inefficient or incorrect intermediate implementations. To address this, we propose Search via Co-Evolving World Model and build K-Search based on this method. By replacing static search heuristics with a co-evolving world model, our framework leverages LLMs' prior domain knowledge to guide the search, actively exploring the optimization space. This approach explicitly decouples high-level algorithmic planning from low-level program instantiation, enabling the system to navigate non-monotonic optimization paths while remaining resilient to temporary implementation defects. We evaluate K-Search on diverse, complex kernels from FlashInfer, including GQA, MLA, and MoE kernels. Our results show that K-Search significantly outperforms state-of-the-art evolutionary search methods, achieving an average 2.10x improvement and up to a 14.3x gain on complex MoE kernels. On the GPUMode TriMul task, K-Search achieves state-of-the-art performance on H100, reaching 1030us and surpassing both prior evolution and human-designed solutions. △ Less

Submitted 26 February, 2026; v1 submitted 22 February, 2026; originally announced February 2026.

Better Think Thrice: Learning to Reason Causally with Double Counterfactual Consistency

Authors: Victoria Lin, Xinnuo Xu, Rachel Lawrence, Risa Ueno, Amit Sharma, Javier Gonzalez, Niranjani Prasad

Abstract: Despite their strong performance on reasoning benchmarks, large language models (LLMs) have proven brittle when presented with counterfactual questions, suggesting weaknesses in their causal reasoning ability. While recent work has demonstrated that labeled counterfactual tasks can be useful benchmarks of LLMs' causal reasoning, producing such data at the scale required to cover the vast potential… ▽ More Despite their strong performance on reasoning benchmarks, large language models (LLMs) have proven brittle when presented with counterfactual questions, suggesting weaknesses in their causal reasoning ability. While recent work has demonstrated that labeled counterfactual tasks can be useful benchmarks of LLMs' causal reasoning, producing such data at the scale required to cover the vast potential space of counterfactuals is limited. In this work, we introduce double counterfactual consistency (DCC), a lightweight inference-time method for measuring and guiding the ability of LLMs to reason causally. Without requiring labeled counterfactual data, DCC verifies a model's ability to execute two important elements of causal reasoning: causal intervention and counterfactual prediction. Using DCC, we evaluate the causal reasoning abilities of various leading LLMs across a range of reasoning tasks and interventions. Moreover, we demonstrate the effectiveness of DCC as a training-free test-time rejection sampling criterion and show that it can directly improve performance on reasoning tasks across multiple model families. △ Less

Submitted 18 February, 2026; originally announced February 2026.

Arming Data Agents with Tribal Knowledge

Authors: Shubham Agarwal, Asim Biswal, Sepanta Zeighami, Alvin Cheung, Joseph Gonzalez, Aditya G. Parameswaran

Abstract: Natural language to SQL (NL2SQL) translation enables non-expert users to query relational databases through natural language. Recently, NL2SQL agents, powered by the reasoning capabilities of Large Language Models (LLMs), have significantly advanced NL2SQL translation. Nonetheless, NL2SQL agents still make mistakes when faced with large-scale real-world databases because they lack knowledge of how… ▽ More Natural language to SQL (NL2SQL) translation enables non-expert users to query relational databases through natural language. Recently, NL2SQL agents, powered by the reasoning capabilities of Large Language Models (LLMs), have significantly advanced NL2SQL translation. Nonetheless, NL2SQL agents still make mistakes when faced with large-scale real-world databases because they lack knowledge of how to correctly leverage the underlying data (e.g., knowledge about the intent of each column) and form misconceptions about the data when querying it, leading to errors. Prior work has studied generating facts about the database to provide more context to NL2SQL agents, but such approaches simply restate database contents without addressing the agent's misconceptions. In this paper, we propose Tk-Boost, a bolt-on framework for augmenting any NL2SQL agent with tribal knowledge: knowledge that corrects the agent's misconceptions in querying the database accumulated through experience using the database. To accumulate experience, Tk-Boost first asks the NL2SQL agent to answer a few queries on the database, identifies the agent's misconceptions by analyzing its mistakes on the database, and generates tribal knowledge to address them. To enable accurate retrieval, Tk-Boost indexes this knowledge with applicability conditions that specify the query features for which the knowledge is useful. When answering new queries, Tk-Boost uses this knowledge to provide feedback to the NL2SQL agent, resolving the agent's misconceptions during SQL generation, and thus improving the agent's accuracy. Extensive experiments across the BIRD and Spider 2.0 benchmarks with various NL2SQL agents shows Tk-Boost improves NL2SQL agents accuracy by up to 16.9% on Spider 2.0 and 13.7% on BIRD △ Less

Submitted 17 February, 2026; v1 submitted 13 February, 2026; originally announced February 2026.

SLA2: Sparse-Linear Attention with Learnable Routing and QAT

Authors: Jintao Zhang, Haoxu Wang, Kai Jiang, Kaiwen Zheng, Youhe Jiang, Ion Stoica, Jianfei Chen, Jun Zhu, Joseph E. Gonzalez

Abstract: Sparse-Linear Attention (SLA) combines sparse and linear attention to accelerate diffusion models and has shown strong performance in video generation. However, (i) SLA relies on a heuristic split that assigns computations to the sparse or linear branch based on attention-weight magnitude, which can be suboptimal. Additionally, (ii) after formally analyzing the attention error in SLA, we identify… ▽ More Sparse-Linear Attention (SLA) combines sparse and linear attention to accelerate diffusion models and has shown strong performance in video generation. However, (i) SLA relies on a heuristic split that assigns computations to the sparse or linear branch based on attention-weight magnitude, which can be suboptimal. Additionally, (ii) after formally analyzing the attention error in SLA, we identify a mismatch between SLA and a direct decomposition into sparse and linear attention. We propose SLA2, which introduces (I) a learnable router that dynamically selects whether each attention computation should use sparse or linear attention, (II) a more faithful and direct sparse-linear attention formulation that uses a learnable ratio to combine the sparse and linear attention branches, and (III) a sparse + low-bit attention design, where low-bit attention is introduced via quantization-aware fine-tuning to reduce quantization error. Experiments show that on video diffusion models, SLA2 can achieve 97% attention sparsity and deliver an 18.6x attention speedup while preserving generation quality. △ Less

Submitted 13 February, 2026; originally announced February 2026.

VisGym: Diverse, Customizable, Scalable Environments for Multimodal Agents

Authors: Zirui Wang, Junyi Zhang, Jiaxin Ge, Long Lian, Letian Fu, Lisa Dunlap, Ken Goldberg, XuDong Wang, Ion Stoica, David M. Chan, Sewon Min, Joseph E. Gonzalez

Abstract: Modern Vision-Language Models (VLMs) remain poorly characterized in multi-step visual interactions, particularly in how they integrate perception, memory, and action over long horizons. We introduce VisGym, a gymnasium of 17 environments for evaluating and training VLMs. The suite spans symbolic puzzles, real-image understanding, navigation, and manipulation, and provides flexible controls over di… ▽ More Modern Vision-Language Models (VLMs) remain poorly characterized in multi-step visual interactions, particularly in how they integrate perception, memory, and action over long horizons. We introduce VisGym, a gymnasium of 17 environments for evaluating and training VLMs. The suite spans symbolic puzzles, real-image understanding, navigation, and manipulation, and provides flexible controls over difficulty, input representation, planning horizon, and feedback. We also provide multi-step solvers that generate structured demonstrations, enabling supervised finetuning. Our evaluations show that all frontier models struggle in interactive settings, achieving low success rates in both the easy (46.6%) and hard (26.0%) configurations. Our experiments reveal notable limitations: models struggle to effectively leverage long context, performing worse with an unbounded history than with truncated windows. Furthermore, we find that several text-based symbolic tasks become substantially harder once rendered visually. However, explicit goal observations, textual feedback, and exploratory demonstrations in partially observable or unknown-dynamics settings for supervised finetuning yield consistent gains, highlighting concrete failure modes and pathways for improving multi-step visual decision-making. Code, data, and models can be found at: https://visgym.github.io/. △ Less

Submitted 23 January, 2026; originally announced January 2026.

Comments: Project page: https://visgym.github.io/

DeltaDorsal: Enhancing Hand Pose Estimation with Dorsal Features in Egocentric Views

Authors: William Huang, Siyou Pei, Leyi Zou, Eric J. Gonzalez, Ishan Chatterjee, Yang Zhang

Abstract: The proliferation of XR devices has made egocentric hand pose estimation a vital task, yet this perspective is inherently challenged by frequent finger occlusions. To address this, we propose a novel approach that leverages the rich information in dorsal hand skin deformation, unlocked by recent advances in dense visual featurizers. We introduce a dual-stream delta encoder that learns pose by cont… ▽ More The proliferation of XR devices has made egocentric hand pose estimation a vital task, yet this perspective is inherently challenged by frequent finger occlusions. To address this, we propose a novel approach that leverages the rich information in dorsal hand skin deformation, unlocked by recent advances in dense visual featurizers. We introduce a dual-stream delta encoder that learns pose by contrasting features from a dynamic hand with a baseline relaxed position. Our evaluation demonstrates that, using only cropped dorsal images, our method reduces the Mean Per Joint Angle Error (MPJAE) by 18% in self-occluded scenarios (fingers >= 50% occluded) compared to state-of-the-art techniques that depend on the whole hand's geometry and large model backbones. Consequently, our method not only enhances the reliability of downstream tasks like index finger pinch and tap estimation in occluded scenarios but also unlocks new interaction paradigms, such as detecting isometric force for a surface "click" without visible movement while minimizing model size. △ Less

Submitted 26 January, 2026; v1 submitted 21 January, 2026; originally announced January 2026.

Comments: 16 pages, 11 figures, Presented at ACM CHI 2026. For associated codebase, see https://github.com/hilab-open-source/deltadorsal

'1'-bit Count-based Sorting Unit to Reduce Link Power in DNN Accelerators

Authors: Ruichi Han, Yizhi Chen, Tong Lei, Jordi Altayo Gonzalez, Ahmed Hemani

Abstract: Interconnect power consumption remains a bottleneck in Deep Neural Network (DNN) accelerators. While ordering data based on '1'-bit counts can mitigate this via reduced switching activity, practical hardware sorting implementations remain underexplored. This work proposes the hardware implementation of a comparison-free sorting unit optimized for Convolutional Neural Networks (CNN). By leveraging… ▽ More Interconnect power consumption remains a bottleneck in Deep Neural Network (DNN) accelerators. While ordering data based on '1'-bit counts can mitigate this via reduced switching activity, practical hardware sorting implementations remain underexplored. This work proposes the hardware implementation of a comparison-free sorting unit optimized for Convolutional Neural Networks (CNN). By leveraging approximate computing to group population counts into coarse-grained buckets, our design achieves hardware area reductions while preserving the link power benefits of data reordering. Our approximate sorting unit achieves up to 35.4% area reduction while maintaining 19.50\% BT reduction compared to 20.42% of precise implementation. △ Less

Submitted 20 January, 2026; originally announced January 2026.

Comments: Accepted for oral presentation at the 2026 VLSI Symposium on Technology, Systems and Applications (VLSI TSA) on April 13-17, 2026, at the Ambassador Hotel, Hsinchu, Taiwan

RedunCut: Measurement-Driven Sampling and Accuracy Performance Modeling for Low-Cost Live Video Analytics

Authors: Gur-Eyal Sela, Kumar Krishna Agrawal, Bharathan Balaji, Joseph Gonzalez, Ion Stoica

Abstract: Live video analytics (LVA) runs continuously across massive camera fleets, but inference cost with modern vision models remains high. To address this, dynamic model size selection (DMSS) is an attractive approach: it is content-aware but treats models as black boxes, and could potentially reduce cost by up to 10x without model retraining or modification. Without ground truth labels at runtime, we… ▽ More Live video analytics (LVA) runs continuously across massive camera fleets, but inference cost with modern vision models remains high. To address this, dynamic model size selection (DMSS) is an attractive approach: it is content-aware but treats models as black boxes, and could potentially reduce cost by up to 10x without model retraining or modification. Without ground truth labels at runtime, we observe that DMSS methods use two stages per segment: (i) sampling a few models to calculate prediction statistics (e.g., confidences), then (ii) selection of the model size from those statistics. Prior systems fail to generalize to diverse workloads, particularly to mobile videos and lower accuracy targets. We identify that the failure modes stem from inefficient sampling whose cost exceeds its benefit, and inaccurate per-segment accuracy prediction. In this work, we present RedunCut, a new DMSS system that addresses both: It uses a measurement-driven planner that estimates the cost-benefit tradeoff of sampling, and a lightweight, data-driven performance model to improve accuracy prediction. Across road-vehicle, drone, and surveillance videos and multiple model families and tasks, RedunCut reduces compute cost by 14-62% at fixed accuracy and remains robust to limited historical data and to drift. △ Less

Submitted 30 December, 2025; originally announced December 2025.

Comments: 21 pages, 23 figures

TurboDiffusion: Accelerating Video Diffusion Models by 100-200 Times

Authors: Jintao Zhang, Kaiwen Zheng, Kai Jiang, Haoxu Wang, Ion Stoica, Joseph E. Gonzalez, Jianfei Chen, Jun Zhu

Abstract: We introduce TurboDiffusion, a video generation acceleration framework that can speed up end-to-end diffusion generation by 100-200x while maintaining video quality. TurboDiffusion mainly relies on several components for acceleration: (1) Attention acceleration: TurboDiffusion uses low-bit SageAttention and trainable Sparse-Linear Attention (SLA) to speed up attention computation. (2) Step distill… ▽ More We introduce TurboDiffusion, a video generation acceleration framework that can speed up end-to-end diffusion generation by 100-200x while maintaining video quality. TurboDiffusion mainly relies on several components for acceleration: (1) Attention acceleration: TurboDiffusion uses low-bit SageAttention and trainable Sparse-Linear Attention (SLA) to speed up attention computation. (2) Step distillation: TurboDiffusion adopts rCM for efficient step distillation. (3) W8A8 quantization: TurboDiffusion quantizes model parameters and activations to 8 bits to accelerate linear layers and compress the model. In addition, TurboDiffusion incorporates several other engineering optimizations. We conduct experiments on the Wan2.2-I2V-14B-720P, Wan2.1-T2V-1.3B-480P, Wan2.1-T2V-14B-720P, and Wan2.1-T2V-14B-480P models. Experimental results show that TurboDiffusion achieves 100-200x speedup for video generation even on a single RTX 5090 GPU, while maintaining comparable video quality. The GitHub repository, which includes model checkpoints and easy-to-use code, is available at https://github.com/thu-ml/TurboDiffusion. △ Less

Submitted 17 December, 2025; originally announced December 2025.

FrontierCS: Evolving Challenges for Evolving Intelligence

Authors: Qiuyang Mang, Wenhao Chai, Zhifei Li, Huanzhi Mao, Shang Zhou, Alexander Du, Hanchen Li, Shu Liu, Edwin Chen, Yichuan Wang, Xieting Chu, Zerui Cheng, Yuan Xu, Tian Xia, Zirui Wang, Tianneng Shi, Jianzhu Yao, Yilong Zhao, Qizheng Zhang, Charlie Ruan, Zeyu Shen, Kaiyuan Liu, Runyuan He, Dong Xing, Zerui Li , et al. (26 additional authors not shown)

Abstract: We introduce FrontierCS, a benchmark of 156 open-ended problems across diverse areas of computer science, designed and reviewed by experts, including CS PhDs and top-tier competitive programming participants and problem setters. Unlike existing benchmarks that focus on tasks with known optimal solutions, FrontierCS targets problems where the optimal solution is unknown, but the quality of a soluti… ▽ More We introduce FrontierCS, a benchmark of 156 open-ended problems across diverse areas of computer science, designed and reviewed by experts, including CS PhDs and top-tier competitive programming participants and problem setters. Unlike existing benchmarks that focus on tasks with known optimal solutions, FrontierCS targets problems where the optimal solution is unknown, but the quality of a solution can be objectively evaluated. Models solve these tasks by implementing executable programs rather than outputting a direct answer. FrontierCS includes algorithmic problems, which are often NP-hard variants of competitive programming problems with objective partial scoring, and research problems with the same property. For each problem we provide an expert reference solution and an automatic evaluator. Combining open-ended design, measurable progress, and expert curation, FrontierCS provides a benchmark at the frontier of computer-science difficulty. Empirically, we find that frontier reasoning models still lag far behind human experts on both the algorithmic and research tracks, that increasing reasoning budgets alone does not close this gap, and that models often over-optimize for generating merely workable code instead of discovering high-quality algorithms and system designs. △ Less

Submitted 17 December, 2025; originally announced December 2025.

Comments: Code with instruction: https://github.com/FrontierCS/Frontier-CS

Measuring Agents in Production

Authors: Melissa Z. Pan, Negar Arabzadeh, Riccardo Cogo, Yuxuan Zhu, Alexander Xiong, Lakshya A Agrawal, Huanzhi Mao, Emma Shen, Sid Pallerla, Liana Patel, Shu Liu, Tianneng Shi, Xiaoyuan Liu, Jared Quincy Davis, Emmanuele Lacavalla, Alessandro Basile, Shuyi Yang, Paul Castro, Daniel Kang, Koushik Sen, Dawn Song, Joseph E. Gonzalez, Ion Stoica, Matei Zaharia, Marquita Ellis

Abstract: LLM-based agents already operate in production across many industries, yet we lack an understanding of what technical methods make deployments successful. We present the first systematic study of Measuring Agents in Production, MAP, using first-hand data from agent developers. We conducted 20 case studies via in-depth interviews and surveyed 86 deployed systems practitioners across 26 domains. We… ▽ More LLM-based agents already operate in production across many industries, yet we lack an understanding of what technical methods make deployments successful. We present the first systematic study of Measuring Agents in Production, MAP, using first-hand data from agent developers. We conducted 20 case studies via in-depth interviews and surveyed 86 deployed systems practitioners across 26 domains. We investigate why organizations build agents, how they build them, how they evaluate them, and their top development challenges. Our study finds that production agents are built using simple, controllable approaches: 68% execute at most 10 steps before human intervention, 70% rely on prompting off-the-shelf models instead of weight tuning, and 74% depend primarily on human evaluation. Reliability (consistent correct behavior over time) remains the top development challenge, which practitioners currently address through systems-level design. MAP documents the current state of production agents, providing the research community with visibility into deployment realities and underexplored research avenues. △ Less

Submitted 4 June, 2026; v1 submitted 2 December, 2025; originally announced December 2025.

Comments: Accepted to the 43rd International Conference on Machine Learning (ICML 2026) as Oral Presentation

Dynamically Scaled Activation Steering

Authors: Alex Ferrando, Xavier Suau, Jordi Gonzàlez, Pau Rodriguez

Abstract: Activation steering has emerged as a powerful method for guiding the behavior of generative models towards desired outcomes such as toxicity mitigation. However, most existing methods apply interventions uniformly across all inputs, degrading model performance when steering is unnecessary. We introduce Dynamically Scaled Activation Steering (DSAS), a method-agnostic steering framework that decoupl… ▽ More Activation steering has emerged as a powerful method for guiding the behavior of generative models towards desired outcomes such as toxicity mitigation. However, most existing methods apply interventions uniformly across all inputs, degrading model performance when steering is unnecessary. We introduce Dynamically Scaled Activation Steering (DSAS), a method-agnostic steering framework that decouples when to steer from how to steer. DSAS adaptively modulates the strength of existing steering transformations across layers and inputs, intervening strongly only when undesired behavior is detected. At generation time, DSAS computes context-dependent scaling factors that selectively adjust the strength of any steering method. We also show how DSAS can be jointly optimized end-to-end together with the steering function. When combined with existing steering methods, DSAS consistently improves the Pareto front with respect to steering alone, achieving a better trade-off between toxicity mitigation and utility preservation. We further demonstrate DSAS's generality by applying it to a text-to-image diffusion model, showing how adaptive steering allows the modulation of specific concepts. Finally, DSAS introduces minimal computational overhead while improving interpretability, pinpointing which tokens require steering and by how much. △ Less

Submitted 30 July, 2026; v1 submitted 3 December, 2025; originally announced December 2025.

Closing the Performance Gap Between AI and Radiologists in Chest X-Ray Reporting

Authors: Harshita Sharma, Maxwell C. Reynolds, Valentina Salvatelli, Anne-Marie G. Sykes, Kelly K. Horst, Anton Schwaighofer, Maximilian Ilse, Olesya Melnichenko, Sam Bond-Taylor, Fernando Pérez-García, Vamshi K. Mugu, Alex Chan, Ceylan Colak, Shelby A. Swartz, Motassem B. Nashawaty, Austin J. Gonzalez, Heather A. Ouellette, Selnur B. Erdal, Beth A. Schueler, Maria T. Wetscherek, Noel Codella, Mohit Jain, Shruthi Bannur, Kenza Bouzid, Daniel C. Castro , et al. (4 additional authors not shown)

Abstract: AI-assisted report generation offers the opportunity to reduce radiologists' workload stemming from expanded screening guidelines, complex cases and workforce shortages, while maintaining diagnostic accuracy. In addition to describing pathological findings in chest X-ray reports, interpreting lines and tubes (L&T) is demanding and repetitive for radiologists, especially with high patient volumes.… ▽ More AI-assisted report generation offers the opportunity to reduce radiologists' workload stemming from expanded screening guidelines, complex cases and workforce shortages, while maintaining diagnostic accuracy. In addition to describing pathological findings in chest X-ray reports, interpreting lines and tubes (L&T) is demanding and repetitive for radiologists, especially with high patient volumes. We introduce MAIRA-X, a clinically evaluated multimodal AI model for longitudinal chest X-ray (CXR) report generation, that encompasses both clinical findings and L&T reporting. Developed using a large-scale, multi-site, longitudinal dataset of 3.1 million studies (comprising 6 million images from 806k patients) from Mayo Clinic, MAIRA-X was evaluated on three holdout datasets and the public MIMIC-CXR dataset, where it significantly improved AI-generated reports over the state of the art on lexical quality, clinical correctness, and L&T-related elements. A novel L&T-specific metrics framework was developed to assess accuracy in reporting attributes such as type, longitudinal change and placement. A first-of-its-kind retrospective user evaluation study was conducted with nine radiologists of varying experience, who blindly reviewed 600 studies from distinct subjects. The user study found comparable rates of critical errors (3.0% for original vs. 4.6% for AI-generated reports) and a similar rate of acceptable sentences (97.8% for original vs. 97.4% for AI-generated reports), marking a significant improvement over prior user studies with larger gaps and higher error rates. Our results suggest that MAIRA-X can effectively assist radiologists, particularly in high-volume clinical settings. △ Less

Submitted 21 November, 2025; originally announced November 2025.