Showing 1–50 of 314 results for author: Lee, I

What do Language Models Learn and When? The Implicit Curriculum Hypothesis

Authors: Emmy Liu, Kaiser Sun, Millicent Li, Isabelle Lee, Lindia Tjuatja, Jen-tse Huang, Graham Neubig

Abstract: Large language models (LLMs) can perform remarkably complex tasks, yet the fine-grained details of how these capabilities emerge during pretraining remain poorly understood. Scaling laws on validation loss tell us how much a model improves with additional compute, but not what skills it acquires in which order. To remedy this, we propose the Implicit Curriculum Hypothesis: pretraining follows a co… ▽ More Large language models (LLMs) can perform remarkably complex tasks, yet the fine-grained details of how these capabilities emerge during pretraining remain poorly understood. Scaling laws on validation loss tell us how much a model improves with additional compute, but not what skills it acquires in which order. To remedy this, we propose the Implicit Curriculum Hypothesis: pretraining follows a compositional and predictable curriculum across models and data mixtures. We test this by designing a suite of simple, composable tasks spanning retrieval, morphological transformations, coreference, logical reasoning, and mathematics. Using these tasks, we track emergence points across four model families spanning sizes from 410M-13B parameters. We find that emergence orderings of when models reach fixed accuracy thresholds are strikingly consistent ($ρ= .81$ across 45 model pairs), and that composite tasks most often emerge after their component tasks. Furthermore, we find that this structure is encoded in model representations: tasks with similar function vector representations also tend to follow similar trajectories in training. By using the space of representations derived from our task set, we can effectively predict the training trajectories of simple held-out compositional tasks throughout the course of pretraining ($R^2 = .68$-$.84$ across models) without previously evaluating them. Together, these results suggest that pretraining is more structured than loss curves reveal: skills emerge in a compositional order that is consistent across models and readable from their internals. △ Less

Submitted 9 April, 2026; originally announced April 2026.

The Format Tax

Authors: Ivan Yee Lee, Loris D'Antoni, Taylor Berg-Kirkpatrick

Abstract: Asking a large language model to respond in JSON should be a formatting choice, not a capability tax. Yet we find that structured output requirements -- JSON, XML, LaTeX, Markdown -- substantially degrade reasoning and writing performance across open-weight models. The research response has focused on constrained decoding, but sampling bias accounts for only a fraction of the degradation. The domi… ▽ More Asking a large language model to respond in JSON should be a formatting choice, not a capability tax. Yet we find that structured output requirements -- JSON, XML, LaTeX, Markdown -- substantially degrade reasoning and writing performance across open-weight models. The research response has focused on constrained decoding, but sampling bias accounts for only a fraction of the degradation. The dominant cost enters at the prompt: format-requesting instructions alone cause most of the accuracy loss, before any decoder constraint is applied. This diagnosis points to a simple principle: decouple reasoning from formatting. Whether by generating freeform first and reformatting in a second pass, or by enabling extended thinking within a single generation, separating the two concerns substantially recovers lost accuracy. Across six open-weight models, four API models, four formats, and tasks spanning math, science, logic, and writing, decoupling recovers most lost accuracy. Notably, most recent closed-weight models show little to no format tax, suggesting the problem is not inherent to structured generation but a gap that current open-weight models have yet to close. Code is available at https://github.com/ivnle/the-format-tax. △ Less

Submitted 4 April, 2026; originally announced April 2026.

Video-Oasis: Rethinking Evaluation of Video Understanding

Authors: Geuntaek Lim, Minho Shim, Sungjune Park, Jaeyun Lee, Inwoong Lee, Taeoh Kim, Dongyoon Wee, Yukyung Choi

Abstract: The inherent complexity of video understanding makes it difficult to attribute whether performance gains stem from visual perception, linguistic reasoning, or knowledge priors. While many benchmarks have emerged to assess high-level reasoning, the essential criteria that constitute video understanding remain largely overlooked. Instead of introducing yet another benchmark, we take a step back to r… ▽ More The inherent complexity of video understanding makes it difficult to attribute whether performance gains stem from visual perception, linguistic reasoning, or knowledge priors. While many benchmarks have emerged to assess high-level reasoning, the essential criteria that constitute video understanding remain largely overlooked. Instead of introducing yet another benchmark, we take a step back to re-examine the current landscape of video understanding. In this work, we provide Video-Oasis, a sustainable diagnostic suite designed to systematically evaluate existing evaluations and distill spatio-temporal challenges for video understanding. Our analysis reveals two critical findings: (1) 54% of existing benchmark samples are solvable without visual input or temporal context, and (2) on the remaining samples, state-of-the-art models exhibit performance barely exceeding random guessing. To bridge this gap, we investigate which algorithmic design choices contribute to robust video understanding, providing practical guidelines for future research. We hope our work serves as a standard guideline for benchmark construction and the rigorous evaluation of architecture development. Code is available at https://github.com/sejong-rcv/Video-Oasis. △ Less

Submitted 31 March, 2026; originally announced March 2026.

Class-Distribution Guided Active Learning for 3D Occupancy Prediction in Autonomous Driving

Authors: Wonjune Kim, In-Jae Lee, Sihwan Hwang, Sanmin Kim, Dongsuk Kum

Abstract: 3D occupancy prediction provides dense spatial understanding critical for safe autonomous driving. However, this task suffers from a severe class imbalance due to its volumetric representation, where safety-critical objects (bicycles, traffic cones, pedestrians) occupy minimal voxels compared to dominant backgrounds. Additionally, voxel-level annotation is costly, yet dedicating effort to dominant… ▽ More 3D occupancy prediction provides dense spatial understanding critical for safe autonomous driving. However, this task suffers from a severe class imbalance due to its volumetric representation, where safety-critical objects (bicycles, traffic cones, pedestrians) occupy minimal voxels compared to dominant backgrounds. Additionally, voxel-level annotation is costly, yet dedicating effort to dominant classes is inefficient. To address these challenges, we propose a class-distribution guided active learning framework for selecting training samples to annotate in autonomous driving datasets. Our approach combines three complementary criteria to select the training samples. Inter-sample diversity prioritizes samples whose predicted class distributions differ from those of the labeled set, intra-set diversity prevents redundant sampling within each acquisition cycle, and frequency-weighted uncertainty emphasizes rare classes by reweighting voxel-level entropy with inverse per-sample class proportions. We ensure evaluation validity by using a geographically disjoint train/validation split of Occ3D-nuScenes, which reduces train-validation overlap and mitigates potential map memorization. With only 42.4% labeled data, our framework reaches 26.62 mIoU, comparable to full supervision and outperforming active learning baselines at the same budget. We further validate generality on SemanticKITTI using a different architecture, demonstrating consistent effectiveness across datasets. △ Less

Submitted 28 March, 2026; originally announced March 2026.

Comments: IEEE RA-L 2026

Journal ref: IEEE Robotics and Automation Letters (2026)

Beyond identifiability: Learning causal representations with few environments and finite samples

Authors: Inbeom Lee, Tongtong Jin, Bryon Aragam

Abstract: We provide explicit, finite-sample guarantees for learning causal representations from data with a sublinear number of environments. Causal representation learning seeks to provide a rigourous foundation for the general representation learning problem by bridging causal models with latent factor models in order to learn interpretable representations with causal semantics. Despite a blossoming theo… ▽ More We provide explicit, finite-sample guarantees for learning causal representations from data with a sublinear number of environments. Causal representation learning seeks to provide a rigourous foundation for the general representation learning problem by bridging causal models with latent factor models in order to learn interpretable representations with causal semantics. Despite a blossoming theory of identifiability in causal representation learning, estimation and finite-sample bounds are less well understood. We show that causal representations can be learned with only a logarithmic number of unknown, multi-node interventions, and that the intervention targets need not be carefully designed in advance. Through a careful perturbation analysis, we provide a new analysis of this problem that guarantees consistent recovery of (a) the latent causal graph, (b) the mixing matrix and representations, and (c) \emph{unknown} intervention targets. △ Less

Submitted 26 March, 2026; originally announced March 2026.

Revealing Multi-View Hallucination in Large Vision-Language Models

Authors: Wooje Park, Insu Lee, Soohyun Kim, Jaeyun Jang, Minyoung Noh, Kyuhong Shim, Byonghyo Shim

Abstract: Large vision-language models (LVLMs) are increasingly being applied to multi-view image inputs captured from diverse viewpoints. However, despite this growing use, current LVLMs often confuse or mismatch visual information originating from different instances or viewpoints, a phenomenon we term multi-view hallucination. To systematically analyze this problem, we construct MVH-Bench, a benchmark co… ▽ More Large vision-language models (LVLMs) are increasingly being applied to multi-view image inputs captured from diverse viewpoints. However, despite this growing use, current LVLMs often confuse or mismatch visual information originating from different instances or viewpoints, a phenomenon we term multi-view hallucination. To systematically analyze this problem, we construct MVH-Bench, a benchmark comprising 4.8k question-answer pairs targeting two types of hallucination: cross-instance and cross-view. Empirical results show that recent LVLMs struggle to correctly associate visual evidence with its corresponding instance or viewpoint. To overcome this limitation, we propose Reference Shift Contrastive Decoding (RSCD), a training-free decoding technique that suppresses visual interference by generating negative logits through attention masking. Experiments on MVH-Bench with Qwen2.5-VL and LLaVA-OneVision demonstrate that RSCD consistently improves performance by up to 21.1 and 34.6 points over existing hallucination mitigation methods, highlighting the effectiveness of our approach. △ Less

Submitted 25 March, 2026; originally announced March 2026.

CAF-Score: Calibrating CLAP with LALMs for Reference-free Audio Captioning Evaluation

Authors: Insung Lee, Taeyoung Jeong, Haejun Yoo, Du-Seong Chang, Myoung-Wan Koo

Abstract: While Large Audio-Language Models (LALMs) have advanced audio captioning, robust evaluation remains difficult. Reference-based metrics are expensive and often fail to assess acoustic fidelity, while Contrastive Language-Audio Pretraining (CLAP)-based approaches frequently overlook syntactic errors and fine-grained details. We propose CAF-Score, a reference-free metric that calibrates CLAP's coarse… ▽ More While Large Audio-Language Models (LALMs) have advanced audio captioning, robust evaluation remains difficult. Reference-based metrics are expensive and often fail to assess acoustic fidelity, while Contrastive Language-Audio Pretraining (CLAP)-based approaches frequently overlook syntactic errors and fine-grained details. We propose CAF-Score, a reference-free metric that calibrates CLAP's coarse-grained semantic alignment with the fine-grained comprehension and syntactic awareness of LALMs. By combining contrastive audio-text embeddings with LALM reasoning, CAF-Score effectively detects syntactic inconsistencies and subtle hallucinations. Experiments on the BRACE benchmark demonstrate that our approach achieves the highest correlation with human judgments, even outperforming reference-based baselines in challenging scenarios. These results highlight the efficacy of CAF-Score for reference-free audio captioning evaluation. Code and results are available at https://github.com/inseong00/CAF-Score. △ Less

Submitted 19 March, 2026; originally announced March 2026.

Comments: A condensed version of this work has been submitted to Interspeech 2026. Section 10 is an extended analysis added in this version

Toward Clinically Ready Foundation Models in Medical Image Analysis: Adaptation Mechanisms and Deployment Trade-offs

Authors: Karma Phuntsho, Abdullah, Kyungmi Lee, Ickjai Lee, Euijoon Ahn

Abstract: Foundation models (FMs) have demonstrated strong transferability across medical imaging tasks, yet their clinical utility depends critically on how pretrained representations are adapted to domain-specific data, supervision regimes, and deployment constraints. Prior surveys primarily emphasize architectural advances and application coverage, while the mechanisms of adaptation and their implication… ▽ More Foundation models (FMs) have demonstrated strong transferability across medical imaging tasks, yet their clinical utility depends critically on how pretrained representations are adapted to domain-specific data, supervision regimes, and deployment constraints. Prior surveys primarily emphasize architectural advances and application coverage, while the mechanisms of adaptation and their implications for robustness, calibration, and regulatory feasibility remain insufficiently structured. This review introduces a strategy-centric framework for FM adaptation in medical image analysis (MIA). We conceptualize adaptation as a post-pretraining intervention and organize existing approaches into five mechanisms: parameter-, representation-, objective-, data-centric, and architectural/sequence-level adaptation. For each mechanism, we analyze trade-offs in adaptation depth, label efficiency, domain robustness, computational cost, auditability, and regulatory burden. We synthesize evidence across classification, segmentation, and detection tasks, highlighting how adaptation strategies influence clinically relevant failure modes rather than only aggregate benchmark performance. Finally, we examine how adaptation choices interact with validation protocols, calibration stability, multi-institutional deployment, and regulatory oversight. By reframing adaptation as a process of controlled representational change under clinical constraints, this review provides practical guidance for designing FM-based systems that are robust, auditable, and compatible with clinical deployment. △ Less

Submitted 15 March, 2026; originally announced March 2026.

Report for NSF Workshop on Algorithm-Hardware Co-design for Medical Applications

Authors: Peipei Zhou, Zheng Dong, Insup Lee, Aidong Zhang, Robert Dick, Majid Sarrafzadeh, Xiaodong Wu, Weisong Shi, Zhuoping Yang, Jingtong Hu, Yiyu Shi

Abstract: This report summarizes the discussions and recommendations from the NSF Workshop on Algorithm-Hardware Co-design for Medical Applications, held on September 26-27, 2024, in Pittsburgh, PA. The workshop assembled an interdisciplinary cohort of researchers, clinicians, and industry leaders to examine foundational challenges and develop a strategic roadmap for algorithm-hardware co-design in medical… ▽ More This report summarizes the discussions and recommendations from the NSF Workshop on Algorithm-Hardware Co-design for Medical Applications, held on September 26-27, 2024, in Pittsburgh, PA. The workshop assembled an interdisciplinary cohort of researchers, clinicians, and industry leaders to examine foundational challenges and develop a strategic roadmap for algorithm-hardware co-design in medical computing. The workshop focuses on four thematic areas: (1) teleoperations, telehealth, and surgical operations; (2) wearable and implantable medicine, including implantable living pharmacies; (3) home ICU, hospital systems, and elderly care; and (4) medical sensing, imaging, and reconstruction. This report calls for a fundamental shift in how next-generation medical technologies are conceived, designed, validated, and translated into practice. The report recommends that NSF sustain investment in shared standardized data infrastructures and compute infrastructures, develop clinic workflow-aware systems and human-AI collaboration frameworks, promote scalable validation ecosystems grounded in objective, continuous measures, and physics-informed, and enable safe, accountable, and resilient platforms, including virtual-physical healthcare ecosystems, to de-risk translational pathways. The workshop information can be found on the website: https://sites.google.com/view/nsfworkshop. △ Less

Submitted 11 March, 2026; originally announced March 2026.

Bio-Inspired Self-Supervised Learning for Wrist-worn IMU Signals

Authors: Prithviraj Tarale, Kiet Chu, Abhishek Varghese, Kai-Chun Liu, Maxwell A Xu, Mohit Iyyer, Sunghoon I. Lee

Abstract: Wearable accelerometers have enabled large-scale health and wellness monitoring, yet learning robust human-activity representations has been constrained by the scarcity of labeled data. While self-supervised learning offers a potential remedy, existing approaches treat sensor streams as unstructured time series, overlooking the underlying biological structure of human movement, a factor we argue i… ▽ More Wearable accelerometers have enabled large-scale health and wellness monitoring, yet learning robust human-activity representations has been constrained by the scarcity of labeled data. While self-supervised learning offers a potential remedy, existing approaches treat sensor streams as unstructured time series, overlooking the underlying biological structure of human movement, a factor we argue is critical for effective Human Activity Recognition (HAR). We introduce a novel tokenization strategy grounded in the submovement theory of motor control, which posits that continuous wrist motion is composed of superposed elementary basis functions called submovements. We define our token as the movement segment, a unit of motion composed of a finite sequence of submovements that is readily extractable from wrist accelerometer signals. By treating these segments as tokens, we pretrain a Transformer encoder via masked movement-segment reconstruction to model the temporal dependencies of movement segments, shifting the learning focus beyond local waveform morphology. Pretrained on the NHANES corpus (approximately 28k hours; approximately 11k participants; approximately 10M windows), our representations outperform strong wearable SSL baselines across six subject-disjoint HAR benchmarks. Furthermore, they demonstrate stronger data efficiency in data-scarce settings. Code and pretrained weights will be made publicly available. △ Less

Submitted 11 March, 2026; originally announced March 2026.

GuideTWSI: A Diverse Tactile Walking Surface Indicator Dataset from Synthetic and Real-World Images for Blind and Low-Vision Navigation

Authors: Hochul Hwang, Soowan Yang, Anh N. H. Nguyen, Parth Goel, Krisha Adhikari, Sunghoon I. Lee, Joydeep Biswas, Nicholas A. Giudice, Donghyun Kim

Abstract: Tactile Walking Surface Indicators (TWSIs) are safety-critical landmarks that blind and low-vision (BLV) pedestrians use to locate crossings and hazard zones. From our observation sessions with BLV guide dog handlers, trainers, and an O&M specialist, we confirmed the critical importance of reliable and accurate TWSI segmentation for navigation assistance of BLV individuals. Achieving such reliabil… ▽ More Tactile Walking Surface Indicators (TWSIs) are safety-critical landmarks that blind and low-vision (BLV) pedestrians use to locate crossings and hazard zones. From our observation sessions with BLV guide dog handlers, trainers, and an O&M specialist, we confirmed the critical importance of reliable and accurate TWSI segmentation for navigation assistance of BLV individuals. Achieving such reliability requires large-scale annotated data. However, TWSIs are severely underrepresented in existing urban perception datasets, and even existing dedicated paving datasets are limited: they lack robot-relevant viewpoints (e.g., egocentric or top-down) and are geographically biased toward East Asian directional bars - raised parallel strips used for continuous guidance along sidewalks. This narrow focus overlooks truncated domes - rows of round bumps used primarily in North America and Europe as detectable warnings at curbs, crossings, and platform edges. As a result, models trained only on bar-centric data struggle to generalize to dome-based warnings, leading to missed detections and false stops in safety-critical environments. △ Less

Submitted 7 March, 2026; originally announced March 2026.

SafeGen-LLM: Enhancing Safety Generalization in Task Planning for Robotic Systems

Authors: Jialiang Fan, Weizhe Xu, Mengyu Liu, Oleg Sokolsky, Insup Lee, Fanxin Kong

Abstract: Safety-critical task planning in robotic systems remains challenging: classical planners suffer from poor scalability, Reinforcement Learning (RL)-based methods generalize poorly, and base Large Language Models (LLMs) cannot guarantee safety. To address this gap, we propose safety-generalizable large language models, named SafeGen-LLM. SafeGen-LLM can not only enhance the safety satisfaction of ta… ▽ More Safety-critical task planning in robotic systems remains challenging: classical planners suffer from poor scalability, Reinforcement Learning (RL)-based methods generalize poorly, and base Large Language Models (LLMs) cannot guarantee safety. To address this gap, we propose safety-generalizable large language models, named SafeGen-LLM. SafeGen-LLM can not only enhance the safety satisfaction of task plans but also generalize well to novel safety properties in various domains. We first construct a multi-domain Planning Domain Definition Language 3 (PDDL3) benchmark with explicit safety constraints. Then, we introduce a two-stage post-training framework: Supervised Fine-Tuning (SFT) on a constraint-compliant planning dataset to learn planning syntax and semantics, and Group Relative Policy Optimization (GRPO) guided by fine-grained reward machines derived from formal verification to enforce safety alignment and by curriculum learning to better handle complex tasks. Extensive experiments show that SafeGen-LLM achieves strong safety generalization and outperforms frontier proprietary baselines across multi-domain planning tasks and multiple input formats (e.g., PDDLs and natural language). △ Less

Submitted 27 February, 2026; originally announced February 2026.

Comments: 12 pages, 6 figures

Mining Type Constructs Using Patterns in AI-Generated Code

Authors: Imgyeong Lee, Tayyib Ul Hassan, Abram Hindle

Abstract: Artificial Intelligence (AI) increasingly automates various parts of the software development tasks. Although AI has enhanced the productivity of development tasks, it remains unstudied whether AI essentially outperforms humans in type-related programming tasks, such as employing type constructs properly for type safety, during its tasks. Moreover, there is no systematic study that evaluates wheth… ▽ More Artificial Intelligence (AI) increasingly automates various parts of the software development tasks. Although AI has enhanced the productivity of development tasks, it remains unstudied whether AI essentially outperforms humans in type-related programming tasks, such as employing type constructs properly for type safety, during its tasks. Moreover, there is no systematic study that evaluates whether AI agents overuse or misuse the type constructs under the complicated type systems to the same extent as humans. In this study, we present the first empirical analysis to answer these questions in the domain of TypeScript projects. Our findings show that, in contrast to humans, AI agents are 9x more prone to use the 'any' keyword. In addition, we observed that AI agents use advanced type constructs, including those that ignore type checks, more often compared to humans. Surprisingly, even with all these issues, Agentic pull requests (PRs) have 1.8x higher acceptance rates compared to humans for TypeScript. We encourage software developers to carefully confirm the type safety of their codebases whenever they coordinate with AI agents in the development process. △ Less

Submitted 19 February, 2026; originally announced February 2026.

Benchmarking Self-Supervised Models for Cardiac Ultrasound View Classification

Authors: Youssef Megahed, Salma I. Megahed, Robin Ducharme, Inok Lee, Adrian D. C. Chan, Mark C. Walker, Steven Hawken

Abstract: Reliable interpretation of cardiac ultrasound images is essential for accurate clinical diagnosis and assessment. Self-supervised learning has shown promise in medical imaging by leveraging large unlabelled datasets to learn meaningful representations. In this study, we evaluate and compare two self-supervised learning frameworks, USF-MAE, developed by our team, and MoCo v3, on the recently introd… ▽ More Reliable interpretation of cardiac ultrasound images is essential for accurate clinical diagnosis and assessment. Self-supervised learning has shown promise in medical imaging by leveraging large unlabelled datasets to learn meaningful representations. In this study, we evaluate and compare two self-supervised learning frameworks, USF-MAE, developed by our team, and MoCo v3, on the recently introduced CACTUS dataset (37,736 images) for automated simulated cardiac view (A4C, PL, PSAV, PSMV, Random, and SC) classification. Both models used 5-fold cross-validation, enabling robust assessment of generalization performance across multiple random splits. The CACTUS dataset provides expert-annotated cardiac ultrasound images with diverse views. We adopt an identical training protocol for both models to ensure a fair comparison. Both models are configured with a learning rate of 0.0001 and a weight decay of 0.01. For each fold, we record performance metrics including ROC-AUC, accuracy, F1-score, and recall. Our results indicate that USF-MAE consistently outperforms MoCo v3 across metrics. The average testing AUC for USF-MAE is 99.99% (+/-0.01% 95% CI), compared to 99.97% (+/-0.01%) for MoCo v3. USF-MAE achieves a mean testing accuracy of 99.33% (+/-0.18%), higher than the 98.99% (+/-0.28%) reported for MoCo v3. Similar trends are observed for the F1-score and recall, with improvements statistically significant across folds (paired t-test, p=0.0048 < 0.01). This proof-of-concept analysis suggests that USF-MAE learns more discriminative features for cardiac view classification than MoCo v3 when applied to this dataset. The enhanced performance across multiple metrics highlights the potential of USF-MAE for improving automated cardiac ultrasound classification. △ Less

Submitted 16 February, 2026; originally announced February 2026.

Comments: 10 pages, 3 figures, 3 tables

From Reasoning to Pixels: Benchmarking the Alignment Gap in Unified Multimodal Models

Authors: Cheng Yang, Chufan Shi, Bo Shui, Yaokang Wu, Muzi Tao, Huijuan Wang, Ivan Yee Lee, Yong Liu, Xuezhe Ma, Taylor Berg-Kirkpatrick

Abstract: Unified multimodal models (UMMs) aim to integrate multimodal understanding and generation within a unified architecture, yet it remains unclear to what extent their representations are truly aligned across modalities. To investigate this question, we use reasoning-guided image generation as a diagnostic task, where models produce textual reasoning first and then generate images. We introduce UReas… ▽ More Unified multimodal models (UMMs) aim to integrate multimodal understanding and generation within a unified architecture, yet it remains unclear to what extent their representations are truly aligned across modalities. To investigate this question, we use reasoning-guided image generation as a diagnostic task, where models produce textual reasoning first and then generate images. We introduce UReason, a benchmark for evaluating cross-modal alignment in this paradigm, consisting of 2,000 manually curated instances spanning five reasoning-intensive tasks: Code, Arithmetic, Spatial, Attribute and Text. To enable controlled analysis, we develop an evaluation framework that compares direct generation, reasoning-guided generation and de-contextualized generation, which conditions only on the refined prompt extracted from reasoning. Across eight widely used UMMs, while we find that reasoning-guided generation yields improvements over direct generation, somewhat surprisingly, de-contextualized generation consistently outperforms reasoning-guided generation by a large margin. Our results suggest that the intended visual semantics in textual reasoning are not reliably reflected in the generated images. This finding indicates that, despite unified design and training, current UMMs still do not robustly align representations across modalities. Overall, UReason serves as a practical litmus test for cross-modal alignment and provides a challenging benchmark for developing next-generation, more tightly aligned UMMs. △ Less

Submitted 7 April, 2026; v1 submitted 9 February, 2026; originally announced February 2026.

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

Forget to Generalize: Iterative Adaptation for Generalization in Federated Learning

Authors: Abdulrahman Alotaibi, Irene Tenison, Miriam Kim, Isaac Lee, Lalana Kagal

Abstract: The Web is naturally heterogeneous with user devices, geographic regions, browsing patterns, and contexts all leading to highly diverse, unique datasets. Federated Learning (FL) is an important paradigm for the Web because it enables privacy-preserving, collaborative machine learning across diverse user devices, web services and clients without needing to centralize sensitive data. However, its pe… ▽ More The Web is naturally heterogeneous with user devices, geographic regions, browsing patterns, and contexts all leading to highly diverse, unique datasets. Federated Learning (FL) is an important paradigm for the Web because it enables privacy-preserving, collaborative machine learning across diverse user devices, web services and clients without needing to centralize sensitive data. However, its performance degrades severely under non-IID client distributions that is prevalent in real-world web systems. In this work, we propose a new training paradigm - Iterative Federated Adaptation (IFA) - that enhances generalization in heterogeneous federated settings through generation-wise forget and evolve strategy. Specifically, we divide training into multiple generations and, at the end of each, select a fraction of model parameters (a) randomly or (b) from the later layers of the model and reinitialize them. This iterative forget and evolve schedule allows the model to escape local minima and preserve globally relevant representations. Extensive experiments on CIFAR-10, MIT-Indoors, and Stanford Dogs datasets show that the proposed approach improves global accuracy, especially when the data cross clients are Non-IID. This method can be implemented on top any federated algorithm to improve its generalization performance. We observe an average of 21.5%improvement across datasets. This work advances the vision of scalable, privacy-preserving intelligence for real-world heterogeneous and distributed web systems. △ Less

Submitted 4 February, 2026; originally announced February 2026.

Why Can't I Open My Drawer? Mitigating Object-Driven Shortcuts in Zero-Shot Compositional Action Recognition

Authors: Geo Ahn, Inwoong Lee, Taeoh Kim, Minho Shim, Dongyoon Wee, Jinwoo Choi

Abstract: Zero-Shot Compositional Action Recognition (ZS-CAR) requires recognizing novel verb-object combinations composed of previously observed primitives. In this work, we tackle a key failure mode: models predict verbs via object-driven shortcuts (i.e., relying on the labeled object class) rather than temporal evidence. We argue that sparse compositional supervision and verb-object learning asymmetry ca… ▽ More Zero-Shot Compositional Action Recognition (ZS-CAR) requires recognizing novel verb-object combinations composed of previously observed primitives. In this work, we tackle a key failure mode: models predict verbs via object-driven shortcuts (i.e., relying on the labeled object class) rather than temporal evidence. We argue that sparse compositional supervision and verb-object learning asymmetry can promote object-driven shortcut learning. Our analysis with proposed diagnostic metrics shows that existing methods overfit to training co-occurrence patterns and underuse temporal verb cues, resulting in weak generalization to unseen compositions. To address object-driven shortcuts, we propose Robust COmpositional REpresentations (RCORE) with two components. Co-occurrence Prior Regularization (CPR) adds explicit supervision for unseen compositions and regularizes the model against frequent co-occurrence priors by treating them as hard negatives. Temporal Order Regularization for Composition (TORC) enforces temporal-order sensitivity to learn temporally grounded verb representations. Across Sth-com and EK100-com, RCORE reduces shortcut diagnostics and consequently improves compositional generalization. △ Less

Submitted 7 April, 2026; v1 submitted 22 January, 2026; originally announced January 2026.

Comments: The code is available at https://github.com/KHU-VLL/RCORE

Deep Learning Based Facial Retargeting Using Local Patches

Authors: Yeonsoo Choi, Inyup Lee, Sihun Cha, Seonghyeon Kim, Sunjin Jung, Junyong Noh

Abstract: In the era of digital animation, the quest to produce lifelike facial animations for virtual characters has led to the development of various retargeting methods. While the retargeting facial motion between models of similar shapes has been very successful, challenges arise when the retargeting is performed on stylized or exaggerated 3D characters that deviate significantly from human facial struc… ▽ More In the era of digital animation, the quest to produce lifelike facial animations for virtual characters has led to the development of various retargeting methods. While the retargeting facial motion between models of similar shapes has been very successful, challenges arise when the retargeting is performed on stylized or exaggerated 3D characters that deviate significantly from human facial structures. In this scenario, it is important to consider the target character's facial structure and possible range of motion to preserve the semantics assumed by the original facial motions after the retargeting. To achieve this, we propose a local patch-based retargeting method that transfers facial animations captured in a source performance video to a target stylized 3D character. Our method consists of three modules. The Automatic Patch Extraction Module extracts local patches from the source video frame. These patches are processed through the Reenactment Module to generate correspondingly re-enacted target local patches. The Weight Estimation Module calculates the animation parameters for the target character at every frame for the creation of a complete facial animation sequence. Extensive experiments demonstrate that our method can successfully transfer the semantic meaning of source facial expressions to stylized characters with considerable variations in facial feature proportion. △ Less

Submitted 13 January, 2026; originally announced January 2026.

Comments: Eurographics 25

Journal ref: Computer Graphics Forum 2024

SpatialMosaic: A Multiview VLM Dataset for Partial Visibility

Authors: Kanghee Lee, Injae Lee, Minseok Kwak, Jungi Hong, Kwonyoung Ryu, Jaesik Park

Abstract: The rapid progress of Multimodal Large Language Models (MLLMs) has unlocked the potential for enhanced 3D scene understanding and spatial reasoning. A recent line of work explores learning spatial reasoning directly from multi-view images, enabling MLLMs to understand 3D scenes without explicit 3D reconstructions. Nevertheless, key challenges that frequently arise in real-world environments, such… ▽ More The rapid progress of Multimodal Large Language Models (MLLMs) has unlocked the potential for enhanced 3D scene understanding and spatial reasoning. A recent line of work explores learning spatial reasoning directly from multi-view images, enabling MLLMs to understand 3D scenes without explicit 3D reconstructions. Nevertheless, key challenges that frequently arise in real-world environments, such as partial visibility, occlusion, and low-overlap conditions that require spatial reasoning from fragmented visual cues, remain under-explored. To address these limitations, we propose a scalable multi-view data generation and annotation pipeline that constructs realistic spatial reasoning QAs, resulting in SpatialMosaic, a comprehensive instruction-tuning dataset featuring 2M QA pairs. We further introduce SpatialMosaic-Bench, a challenging benchmark for evaluating multi-view spatial reasoning under complex and diverse scenarios, consisting of 1M QA pairs across 6 tasks. Our proposed dataset spans both indoor and outdoor scenes, enabling comprehensive evaluation in diverse real-world scenarios. In addition, we introduce a new baseline for multi-view settings, SpatialMosaicVLM, a hybrid framework that integrates 3D reconstruction models as geometry encoders within VLMs for robust spatial reasoning. Extensive experiments demonstrate that our proposed dataset effectively enhances spatial reasoning under challenging multi-view conditions, validating the effectiveness of our data generation pipeline in constructing realistic and challenging QAs. Code and dataset will be available soon. △ Less

Submitted 9 April, 2026; v1 submitted 29 December, 2025; originally announced December 2025.

Anomaly Detection by Effectively Leveraging Synthetic Images

Authors: Sungho Kang, Hyunkyu Park, Yeonho Lee, Hanbyul Lee, Mijoo Jeong, YeongHyeon Park, Injae Lee, Juneho Yi

Abstract: Anomaly detection plays a vital role in industrial manufacturing. Due to the scarcity of real defect images, unsupervised approaches that rely solely on normal images have been extensively studied. Recently, diffusion-based generative models brought attention to training data synthesis as an alternative solution. In this work, we focus on a strategy to effectively leverage synthetic images to maxi… ▽ More Anomaly detection plays a vital role in industrial manufacturing. Due to the scarcity of real defect images, unsupervised approaches that rely solely on normal images have been extensively studied. Recently, diffusion-based generative models brought attention to training data synthesis as an alternative solution. In this work, we focus on a strategy to effectively leverage synthetic images to maximize the anomaly detection performance. Previous synthesis strategies are broadly categorized into two groups, presenting a clear trade-off. Rule-based synthesis, such as injecting noise or pasting patches, is cost-effective but often fails to produce realistic defect images. On the other hand, generative model-based synthesis can create high-quality defect images but requires substantial cost. To address this problem, we propose a novel framework that leverages a pre-trained text-guided image-to-image translation model and image retrieval model to efficiently generate synthetic defect images. Specifically, the image retrieval model assesses the similarity of the generated images to real normal images and filters out irrelevant outputs, thereby enhancing the quality and relevance of the generated defect images. To effectively leverage synthetic images, we also introduce a two stage training strategy. In this strategy, the model is first pre-trained on a large volume of images from rule-based synthesis and then fine-tuned on a smaller set of high-quality images. This method significantly reduces the cost for data collection while improving the anomaly detection performance. Experiments on the MVTec AD dataset demonstrate the effectiveness of our approach. △ Less

Submitted 29 December, 2025; originally announced December 2025.

Improved cystic hygroma detection from prenatal imaging using ultrasound-specific self-supervised representation learning

Authors: Youssef Megahed, Robin Ducharme, Inok Lee, Inbal Willner, Adrian D. C. Chan, Mark Walker, Steven Hawken

Abstract: Cystic hygroma is a high-risk prenatal ultrasound finding that portends high rates of chromosomal abnormalities, structural malformations, and adverse pregnancy outcomes. Automated detection can increase reproducibility and support scalable early screening programs, but supervised deep learning methods are limited by small labelled datasets. This study assesses whether ultrasound-specific self-sup… ▽ More Cystic hygroma is a high-risk prenatal ultrasound finding that portends high rates of chromosomal abnormalities, structural malformations, and adverse pregnancy outcomes. Automated detection can increase reproducibility and support scalable early screening programs, but supervised deep learning methods are limited by small labelled datasets. This study assesses whether ultrasound-specific self-supervised pretraining can facilitate accurate, robust deep learning detection of cystic hygroma in first-trimester ultrasound images. We fine-tuned the Ultrasound Self-Supervised Foundation Model with Masked Autoencoding (USF-MAE), pretrained on over 370,000 unlabelled ultrasound images, for binary classification of normal controls and cystic hygroma cases used in this study. Performance was evaluated on the same curated ultrasound dataset, preprocessing pipeline, and 4-fold cross-validation protocol as for the DenseNet-169 baseline, using accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (ROC-AUC). Model interpretability was analyzed qualitatively using Score-CAM visualizations. USF-MAE outperformed the DenseNet-169 baseline on all evaluation metrics. The proposed model yielded a mean accuracy of 0.96, sensitivity of 0.94, specificity of 0.98, and ROC-AUC of 0.98 compared to 0.93, 0.92, 0.94, and 0.94 for the DenseNet-169 baseline, respectively. Qualitative Score-CAM visualizations of model predictions demonstrated clinical relevance by highlighting expected regions in the fetal neck for both positive and negative cases. Paired statistical analysis using a Wilcoxon signed-rank test confirmed that performance improvements achieved by USF-MAE were statistically significant (p = 0.0057). △ Less

Submitted 18 January, 2026; v1 submitted 27 December, 2025; originally announced December 2025.

Comments: 13 pages, 6 figures, 2 tables

DDAVS: Disentangled Audio Semantics and Delayed Bidirectional Alignment for Audio-Visual Segmentation

Authors: Jingqi Tian, Yiheng Du, Haoji Zhang, Yuji Wang, Isaac Ning Lee, Xulong Bai, Tianrui Zhu, Jingxuan Niu, Yansong Tang

Abstract: Audio-Visual Segmentation (AVS) aims to localize sound-producing objects at the pixel level by jointly leveraging auditory and visual information. However, existing methods often suffer from multi-source entanglement and audio-visual misalignment, which lead to biases toward louder or larger objects while overlooking weaker, smaller, or co-occurring sources. To address these challenges, we propose… ▽ More Audio-Visual Segmentation (AVS) aims to localize sound-producing objects at the pixel level by jointly leveraging auditory and visual information. However, existing methods often suffer from multi-source entanglement and audio-visual misalignment, which lead to biases toward louder or larger objects while overlooking weaker, smaller, or co-occurring sources. To address these challenges, we propose DDAVS, a Disentangled Audio Semantics and Delayed Bidirectional Alignment framework. To mitigate multi-source entanglement, DDAVS employs learnable queries to extract audio semantics and anchor them within a structured semantic space derived from an audio prototype memory bank. This is further optimized through contrastive learning to enhance discriminability and robustness. To alleviate audio-visual misalignment, DDAVS introduces dual cross-attention with delayed modality interaction, improving the robustness of multimodal alignment. Extensive experiments on the AVS-Objects and VPO benchmarks demonstrate that DDAVS consistently outperforms existing approaches, exhibiting strong performance across single-source, multi-source, and multi-instance scenarios. These results validate the effectiveness and generalization ability of our framework under challenging real-world audio-visual segmentation conditions. Project page: https://trilarflagz.github.io/DDAVS-page/ △ Less

Submitted 23 December, 2025; originally announced December 2025.

Comments: https://trilarflagz.github.io/DDAVS-page/

Pattern-Guided Diffusion Models

Authors: Vivian Lin, Kuk Jin Jang, Wenwen Si, Insup Lee

Abstract: Diffusion models have shown promise in forecasting future data from multivariate time series. However, few existing methods account for recurring structures, or patterns, that appear within the data. We present Pattern-Guided Diffusion Models (PGDM), which leverage inherent patterns within temporal data for forecasting future time steps. PGDM first extracts patterns using archetypal analysis and e… ▽ More Diffusion models have shown promise in forecasting future data from multivariate time series. However, few existing methods account for recurring structures, or patterns, that appear within the data. We present Pattern-Guided Diffusion Models (PGDM), which leverage inherent patterns within temporal data for forecasting future time steps. PGDM first extracts patterns using archetypal analysis and estimates the most likely next pattern in the sequence. By guiding predictions with this pattern estimate, PGDM makes more realistic predictions that fit within the set of known patterns. We additionally introduce a novel uncertainty quantification technique based on archetypal analysis, and we dynamically scale the guidance level based on the pattern estimate uncertainty. We apply our method to two well-motivated forecasting applications, predicting visual field measurements and motion capture frames. On both, we show that pattern guidance improves PGDM's performance (MAE / CRPS) by up to 40.67% / 56.26% and 14.12% / 14.10%, respectively. PGDM also outperforms baselines by up to 65.58% / 84.83% and 93.64% / 92.55%. △ Less

Submitted 15 December, 2025; originally announced December 2025.

Comments: Under review

Self-Supervised Ultrasound Representation Learning for Renal Anomaly Prediction in Prenatal Imaging

Authors: Youssef Megahed, Inok Lee, Robin Ducharme, Kevin Dick, Adrian D. C. Chan, Steven Hawken, Mark C. Walker

Abstract: Prenatal ultrasound is the cornerstone for detecting congenital anomalies of the kidneys and urinary tract, but diagnosis is limited by operator dependence and suboptimal imaging conditions. We sought to assess the performance of a self-supervised ultrasound foundation model for automated fetal renal anomaly classification using a curated dataset of 969 two-dimensional ultrasound images. A pretrai… ▽ More Prenatal ultrasound is the cornerstone for detecting congenital anomalies of the kidneys and urinary tract, but diagnosis is limited by operator dependence and suboptimal imaging conditions. We sought to assess the performance of a self-supervised ultrasound foundation model for automated fetal renal anomaly classification using a curated dataset of 969 two-dimensional ultrasound images. A pretrained Ultrasound Self-Supervised Foundation Model with Masked Autoencoding (USF-MAE) was fine-tuned for binary and multi-class classification of normal kidneys, urinary tract dilation, and multicystic dysplastic kidney. Models were compared with a DenseNet-169 convolutional baseline using cross-validation and an independent test set. USF-MAE consistently improved upon the baseline across all evaluation metrics in both binary and multi-class settings. USF-MAE achieved an improvement of about 1.87% (AUC) and 7.8% (F1-score) on the validation set, 2.32% (AUC) and 4.33% (F1-score) on the independent holdout test set. The largest gains were observed in the multi-class setting, where the improvement in AUC was 16.28% and 46.15% in F1-score. To facilitate model interpretability, Score-CAM visualizations were adapted for a transformer architecture and show that model predictions were informed by known, clinically relevant renal structures, including the renal pelvis in urinary tract dilation and cystic regions in multicystic dysplastic kidney. These results show that ultrasound-specific self-supervised learning can generate a useful representation as a foundation for downstream diagnostic tasks. The proposed framework offers a robust, interpretable approach to support the prenatal detection of renal anomalies and demonstrates the promise of foundation models in obstetric imaging. △ Less

Submitted 15 December, 2025; originally announced December 2025.

Comments: 14 pages, 8 figures, 4 tables

GuideNav: User-Informed Development of a Vision-Only Robotic Navigation Assistant For Blind Travelers

Authors: Hochul Hwang, Soowan Yang, Jahir Sadik Monon, Nicholas A Giudice, Sunghoon Ivan Lee, Joydeep Biswas, Donghyun Kim

Abstract: While commendable progress has been made in user-centric research on mobile assistive systems for blind and low-vision (BLV) individuals, references that directly inform robot navigation design remain rare. To bridge this gap, we conducted a comprehensive human study involving interviews with 26 guide dog handlers, four white cane users, nine guide dog trainers, and one O\&M trainer, along with 15… ▽ More While commendable progress has been made in user-centric research on mobile assistive systems for blind and low-vision (BLV) individuals, references that directly inform robot navigation design remain rare. To bridge this gap, we conducted a comprehensive human study involving interviews with 26 guide dog handlers, four white cane users, nine guide dog trainers, and one O\&M trainer, along with 15+ hours of observing guide dog-assisted walking. After de-identification, we open-sourced the dataset to promote human-centered development and informed decision-making for assistive systems for BLV people. Building on insights from this formative study, we developed GuideNav, a vision-only, teach-and-repeat navigation system. Inspired by how guide dogs are trained and assist their handlers, GuideNav autonomously repeats a path demonstrated by a sighted person using a robot. Specifically, the system constructs a topological representation of the taught route, integrates visual place recognition with temporal filtering, and employs a relative pose estimator to compute navigation actions - all without relying on costly, heavy, power-hungry sensors such as LiDAR. In field tests, GuideNav consistently achieved kilometer-scale route following across five outdoor environments, maintaining reliability despite noticeable scene variations between teach and repeat runs. A user study with 3 guide dog handlers and 1 guide dog trainer further confirmed the system's feasibility, marking (to our knowledge) the first demonstration of a quadruped mobile system retrieving a path in a manner comparable to guide dogs. △ Less

Submitted 5 December, 2025; originally announced December 2025.

Robust Precoding Designs of RSMA for Multiuser MIMO Systems

Authors: Wentao Zhou, Yijie Mao, Di Zhang, Mérouane Debbah, Inkyu Lee

Abstract: Rate-splitting multiple access (RSMA) has been studied for multiuser multiple-input multiple-output (MUMIMO) systems especially in the presence of imperfect channel state information (CSI) at the transmitter. However, its precoding designs that maximize the sum rate normally have high computational complexity. To implement an efficient RSMA scheme for the MU-MIMO system, in this work, we propose a… ▽ More Rate-splitting multiple access (RSMA) has been studied for multiuser multiple-input multiple-output (MUMIMO) systems especially in the presence of imperfect channel state information (CSI) at the transmitter. However, its precoding designs that maximize the sum rate normally have high computational complexity. To implement an efficient RSMA scheme for the MU-MIMO system, in this work, we propose a novel robust precoding design, which can handle imperfect CSI. Specifically, we first adopt the generalized mutual information to construct a lower bound of the objective function in the sum rate maximization problem. Then, we apply a smooth lower bound of the non-smooth sum rate objective function to construct a new optimization problem. By revealing the relationship between the generalized signal-to-interference-plus-noise ratio and the minimum mean square error matrices, we transform the constructed problem into a tractable one. After decomposing the transformed problem into three subproblems, we investigate a new alternating precoding design based on sequential solutions. Simulation results demonstrate that the proposed precoding scheme achieves comparable performance to conventional methods, while significantly reducing the computational complexity. △ Less

Submitted 4 December, 2025; originally announced December 2025.

Comments: This work has been accepted for publication in IEEE Transactions on Wireless Communications

Optical Context Compression Is Just (Bad) Autoencoding

Authors: Ivan Yee Lee, Cheng Yang, Taylor Berg-Kirkpatrick

Abstract: DeepSeek-OCR shows that rendered text can be reconstructed from a small number of vision tokens, sparking excitement about using vision as a compression medium for long textual contexts. But this pipeline requires rendering token embeddings to pixels and compressing from there -- discarding learned representations in favor of an image the vision encoder must then recover from. We ask whether this… ▽ More DeepSeek-OCR shows that rendered text can be reconstructed from a small number of vision tokens, sparking excitement about using vision as a compression medium for long textual contexts. But this pipeline requires rendering token embeddings to pixels and compressing from there -- discarding learned representations in favor of an image the vision encoder must then recover from. We ask whether this detour helps. Comparing DeepSeek-OCR's vision encoder against near-zero-parameter mean pooling and a learned hierarchical encoder, we find it does not. For reconstruction, simple direct methods match or surpass vision at every compression ratio. For language modeling, vision performs comparably to truncation -- a baseline that simply discards context -- and loses to the hierarchical encoder at every compression ratio. As expected, all compression methods outperform truncation for factual recall, but vision never surpasses the best direct baseline. The excitement around optical context compression outpaces the evidence. Code and checkpoints are available at https://github.com/ivnle/bad-autoencoding. △ Less

Submitted 4 April, 2026; v1 submitted 3 December, 2025; originally announced December 2025.

OpenBox: Annotate Any Bounding Boxes in 3D

Authors: In-Jae Lee, Mungyeom Kim, Kwonyoung Ryu, Pierre Musacchio, Jaesik Park

Abstract: Unsupervised and open-vocabulary 3D object detection has recently gained attention, particularly in autonomous driving, where reducing annotation costs and recognizing unseen objects are critical for both safety and scalability. However, most existing approaches uniformly annotate 3D bounding boxes, ignore objects' physical states, and require multiple self-training iterations for annotation refin… ▽ More Unsupervised and open-vocabulary 3D object detection has recently gained attention, particularly in autonomous driving, where reducing annotation costs and recognizing unseen objects are critical for both safety and scalability. However, most existing approaches uniformly annotate 3D bounding boxes, ignore objects' physical states, and require multiple self-training iterations for annotation refinement, resulting in suboptimal quality and substantial computational overhead. To address these challenges, we propose OpenBox, a two-stage automatic annotation pipeline that leverages a 2D vision foundation model. In the first stage, OpenBox associates instance-level cues from 2D images processed by a vision foundation model with the corresponding 3D point clouds via cross-modal instance alignment. In the second stage, it categorizes instances by rigidity and motion state, then generates adaptive bounding boxes with class-specific size statistics. As a result, OpenBox produces high-quality 3D bounding box annotations without requiring self-training. Experiments on the Waymo Open Dataset, the Lyft Level 5 Perception dataset, and the nuScenes dataset demonstrate improved accuracy and efficiency over baselines. △ Less

Submitted 1 December, 2025; originally announced December 2025.

Comments: Accepted by NeurIPS 2025

Dynamic-eDiTor: Training-Free Text-Driven 4D Scene Editing with Multimodal Diffusion Transformer

Authors: Dong In Lee, Hyungjun Doh, Seunggeun Chi, Runlin Duan, Sangpil Kim, Karthik Ramani

Abstract: Recent progress in 4D representations, such as Dynamic NeRF and 4D Gaussian Splatting (4DGS), has enabled dynamic 4D scene reconstruction. However, text-driven 4D scene editing remains under-explored due to the challenge of ensuring both multi-view and temporal consistency across space and time during editing. Existing studies rely on 2D diffusion models that edit frames independently, often causi… ▽ More Recent progress in 4D representations, such as Dynamic NeRF and 4D Gaussian Splatting (4DGS), has enabled dynamic 4D scene reconstruction. However, text-driven 4D scene editing remains under-explored due to the challenge of ensuring both multi-view and temporal consistency across space and time during editing. Existing studies rely on 2D diffusion models that edit frames independently, often causing motion distortion, geometric drift, and incomplete editing. We introduce Dynamic-eDiTor, a training-free text-driven 4D editing framework leveraging Multimodal Diffusion Transformer (MM-DiT) and 4DGS. This mechanism consists of Spatio-Temporal Sub-Grid Attention (STGA) for locally consistent cross-view and temporal fusion, and Context Token Propagation (CTP) for global propagation via token inheritance and optical-flow-guided token replacement. Together, these components allow Dynamic-eDiTor to perform seamless, globally consistent multi-view video without additional training and directly optimize pre-trained source 4DGS. Extensive experiments on multi-view video dataset DyNeRF demonstrate that our method achieves superior editing fidelity and both multi-view and temporal consistency prior approaches. Project page for results and code: https://di-lee.github.io/dynamic-eDiTor/ △ Less

Submitted 29 November, 2025; originally announced December 2025.

Comments: 4D Scene Editing

Ternary-Input Binary-Weight CNN Accelerator Design for Miniature Object Classification System with Query-Driven Spatial DVS

Authors: Yuyang Li, Swasthik Muloor, Jack Laudati, Nickolas Dematteis, Yidam Park, Hana Kim, Nathan Chang, Inhee Lee

Abstract: Miniature imaging systems are essential for space-constrained applications but are limited by memory and power constraints. While machine learning can reduce data size by extracting key features, its high energy demands often exceed the capacity of small batteries. This paper presents a CNN hardware accelerator optimized for object classification in miniature imaging systems. It processes data fro… ▽ More Miniature imaging systems are essential for space-constrained applications but are limited by memory and power constraints. While machine learning can reduce data size by extracting key features, its high energy demands often exceed the capacity of small batteries. This paper presents a CNN hardware accelerator optimized for object classification in miniature imaging systems. It processes data from a spatial Dynamic Vision Sensor (DVS), reconfigurable to a temporal DVS via pixel sharing, minimizing sensor area. By using ternary DVS outputs and a ternary-input, binary-weight neural network, the design reduces computation and memory needs. Fabricated in 28 nm CMOS, the accelerator cuts data size by 81% and MAC operations by 27%. It achieves 440 ms inference time at just 1.6 mW power consumption, improving the Figure-of-Merit (FoM) by 7.3x over prior CNN accelerators for miniature systems. △ Less

Submitted 28 November, 2025; originally announced December 2025.

Comments: 6 pages.12 figures & 2 table

Enhanced Conditional Generation of Double Perovskite by Knowledge-Guided Language Model Feedback

Authors: Inhyo Lee, Junhyeong Lee, Jongwon Park, KyungTae Lim, Seunghwa Ryu

Abstract: Double perovskites (DPs) are promising candidates for sustainable energy technologies due to their compositional tunability and compatibility with low-energy fabrication, yet their vast design space poses a major challenge for conditional materials discovery. This work introduces a multi-agent, text gradient-driven framework that performs DP composition generation under natural-language conditions… ▽ More Double perovskites (DPs) are promising candidates for sustainable energy technologies due to their compositional tunability and compatibility with low-energy fabrication, yet their vast design space poses a major challenge for conditional materials discovery. This work introduces a multi-agent, text gradient-driven framework that performs DP composition generation under natural-language conditions by integrating three complementary feedback sources: LLM-based self-evaluation, DP-specific domain knowledge-informed feedback, and ML surrogate-based feedback. Analogous to how knowledge-informed machine learning improves the reliability of conventional data-driven models, our framework incorporates domain-informed text gradients to guide the generative process toward physically meaningful regions of the DP composition space. Systematic comparison of three incremental configurations, (i) pure LLM generation, (ii) LLM generation with LLM reasoning-based feedback, and (iii) LLM generation with domain knowledge-guided feedback, shows that iterative guidance from knowledge-informed gradients improves stability-condition satisfaction without additional training data, achieving over 98% compositional validity and up to 54% stable or metastable candidates, surpassing both the LLM-only baseline (43%) and prior GAN-based results (27%). Analyses of ML-based gradients further reveal that they enhance performance in in-distribution (ID) regions but become unreliable in out-of-distribution (OOD) regimes. Overall, this work provides the first systematic analysis of multi-agent, knowledge-guided text gradients for DP discovery and establishes a generalizable blueprint for MAS-driven generative materials design aimed at advancing sustainable technologies. △ Less

Submitted 2 December, 2025; v1 submitted 27 November, 2025; originally announced November 2025.

3D-Consistent Multi-View Editing by Correspondence Guidance

Authors: Josef Bengtson, David Nilsson, Dong In Lee, Yaroslava Lochman, Fredrik Kahl

Abstract: Recent advancements in diffusion and flow models have greatly improved text-based image editing, yet methods that edit images independently often produce geometrically and photometrically inconsistent results across different views of the same scene. Such inconsistencies are particularly problematic for editing of 3D representations such as NeRFs or Gaussian splat models. We propose a training-fre… ▽ More Recent advancements in diffusion and flow models have greatly improved text-based image editing, yet methods that edit images independently often produce geometrically and photometrically inconsistent results across different views of the same scene. Such inconsistencies are particularly problematic for editing of 3D representations such as NeRFs or Gaussian splat models. We propose a training-free guidance framework that enforces multi-view consistency during the image editing process. The key idea is that corresponding points should look similar after editing. To achieve this, we introduce a consistency loss that guides the denoising process toward coherent edits. The framework is flexible and can be combined with widely varying image editing methods, supporting both dense and sparse multi-view editing setups. Experimental results show that our approach significantly improves 3D consistency compared to existing multi-view editing methods. We also show that this increased consistency enables high-quality Gaussian splat editing with sharp details and strong fidelity to user-specified text prompts. Please refer to our project page for video results: https://3d-consistent-editing.github.io/ △ Less

Submitted 20 March, 2026; v1 submitted 27 November, 2025; originally announced November 2025.

Comments: Added experiments with FLUX.1 editing method

Learning to Generate Human-Human-Object Interactions from Textual Descriptions

Authors: Jeonghyeon Na, Sangwon Baik, Inhee Lee, Junyoung Lee, Hanbyul Joo

Abstract: The way humans interact with each other, including interpersonal distances, spatial configuration, and motion, varies significantly across different situations. To enable machines to understand such complex, context-dependent behaviors, it is essential to model multiple people in relation to the surrounding scene context. In this paper, we present a novel research problem to model the correlations… ▽ More The way humans interact with each other, including interpersonal distances, spatial configuration, and motion, varies significantly across different situations. To enable machines to understand such complex, context-dependent behaviors, it is essential to model multiple people in relation to the surrounding scene context. In this paper, we present a novel research problem to model the correlations between two people engaged in a shared interaction involving an object. We refer to this formulation as Human-Human-Object Interactions (HHOIs). To overcome the lack of dedicated datasets for HHOIs, we present a newly captured HHOIs dataset and a method to synthesize HHOI data by leveraging image generative models. As an intermediary, we obtain individual human-object interaction (HOIs) and human-human interaction (HHIs) from the HHOIs, and with these data, we train an text-to-HOI and text-to-HHI model using score-based diffusion model. Finally, we present a unified generative framework that integrates the two individual model, capable of synthesizing complete HHOIs in a single advanced sampling process. Our method extends HHOI generation to multi-human settings, enabling interactions involving more than two individuals. Experimental results show that our method generates realistic HHOIs conditioned on textual descriptions, outperforming previous approaches that focus only on single-human HOIs. Furthermore, we introduce multi-human motion generation involving objects as an application of our framework. △ Less

Submitted 24 December, 2025; v1 submitted 25 November, 2025; originally announced November 2025.

Comments: Project Page: https://tlb-miss.github.io/hhoi/

PIM or CXL-PIM? Understanding Architectural Trade-offs Through Large-Scale Benchmarking

Authors: I-Ting Lee, Bao-Kai Wang, Liang-Chi Chen, Wen Sheng Lim, Da-Wei Chang, Yu-Ming Chang, Chieng-Chung Ho

Abstract: Processing-in-memory (PIM) reduces data movement by executing near memory, but our large-scale characterization on real PIM hardware shows that end-to-end performance is often limited by disjoint host and device address spaces that force explicit staging transfers. In contrast, CXL-PIM provides a unified address space and cache-coherent access at the cost of higher access latency. These opposing i… ▽ More Processing-in-memory (PIM) reduces data movement by executing near memory, but our large-scale characterization on real PIM hardware shows that end-to-end performance is often limited by disjoint host and device address spaces that force explicit staging transfers. In contrast, CXL-PIM provides a unified address space and cache-coherent access at the cost of higher access latency. These opposing interface models create workload-dependent tradeoffs that are not captured by small-scale studies. This work presents a side-by-side, large-scale comparison of PIM and CXL-PIM using measurements from real PIM hardware and trace-driven CXL modeling. We identify when unified-address access amortizes link latency enough to overcome transfer bottlenecks, and when tightly coupled PIM remains preferable. Our results reveal phase- and dataset-size regimes in which the relative ranking between the two architectures reverses, offering practical guidance for future near-memory system design. △ Less

Submitted 18 November, 2025; v1 submitted 18 November, 2025; originally announced November 2025.

Conformal Constrained Policy Optimization for Cost-Effective LLM Agents

Authors: Wenwen Si, Sooyong Jang, Insup Lee, Osbert Bastani

Abstract: While large language models (LLMs) have recently made tremendous progress towards solving challenging AI problems, they have done so at increasingly steep computational and API costs. We propose a novel strategy where we combine multiple LLM models with varying cost/accuracy tradeoffs in an agentic manner, where models and tools are run in sequence as determined by an orchestration model to minimi… ▽ More While large language models (LLMs) have recently made tremendous progress towards solving challenging AI problems, they have done so at increasingly steep computational and API costs. We propose a novel strategy where we combine multiple LLM models with varying cost/accuracy tradeoffs in an agentic manner, where models and tools are run in sequence as determined by an orchestration model to minimize cost subject to a user-specified level of reliability; this constraint is formalized using conformal prediction to provide guarantees. To solve this problem, we propose Conformal Constrained Policy Optimization (CCPO), a training paradigm that integrates constrained policy optimization with off-policy reinforcement learning and recent advances in online conformal prediction. CCPO jointly optimizes a cost-aware policy (score function) and an adaptive threshold. Across two multi-hop question answering benchmarks, CCPO achieves up to a 30% cost reduction compared to other cost-aware baselines and LLM-guided methods without compromising reliability. Our approach provides a principled and practical framework for deploying LLM agents that are significantly more cost-effective while maintaining reliability. △ Less

Submitted 22 March, 2026; v1 submitted 14 November, 2025; originally announced November 2025.

Quantifying and Improving Adaptivity in Conformal Prediction through Input Transformations

Authors: Sooyong Jang, Insup Lee

Abstract: Conformal prediction constructs a set of labels instead of a single point prediction, while providing a probabilistic coverage guarantee. Beyond the coverage guarantee, adaptiveness to example difficulty is an important property. It means that the method should produce larger prediction sets for more difficult examples, and smaller ones for easier examples. Existing evaluation methods for adaptive… ▽ More Conformal prediction constructs a set of labels instead of a single point prediction, while providing a probabilistic coverage guarantee. Beyond the coverage guarantee, adaptiveness to example difficulty is an important property. It means that the method should produce larger prediction sets for more difficult examples, and smaller ones for easier examples. Existing evaluation methods for adaptiveness typically analyze coverage rate violation or average set size across bins of examples grouped by difficulty. However, these approaches often suffer from imbalanced binning, which can lead to inaccurate estimates of coverage or set size. To address this issue, we propose a binning method that leverages input transformations to sort examples by difficulty, followed by uniform-mass binning. Building on this binning, we introduce two metrics to better evaluate adaptiveness. These metrics provide more reliable estimates of coverage rate violation and average set size due to balanced binning, leading to more accurate adaptivity assessment. Through experiments, we demonstrate that our proposed metric correlates more strongly with the desired adaptiveness property compared to existing ones. Furthermore, motivated by our findings, we propose a new adaptive prediction set algorithm that groups examples by estimated difficulty and applies group-conditional conformal prediction. This allows us to determine appropriate thresholds for each group. Experimental results on both (a) an Image Classification (ImageNet) (b) a medical task (visual acuity prediction) show that our method outperforms existing approaches according to the new metrics. △ Less

Submitted 16 November, 2025; v1 submitted 14 November, 2025; originally announced November 2025.

Deep Learning Analysis of Prenatal Ultrasound for Identification of Ventriculomegaly

Authors: Youssef Megahed, Inok Lee, Robin Ducharme, Aylin Erman, Olivier X. Miguel, Kevin Dick, Adrian D. C. Chan, Steven Hawken, Mark Walker, Felipe Moretti

Abstract: The proposed study aimed to develop a deep learning model capable of detecting ventriculomegaly on prenatal ultrasound images. Ventriculomegaly is a prenatal condition characterized by dilated cerebral ventricles of the fetal brain and is important to diagnose early, as it can be associated with an increased risk for fetal aneuploidies and/or underlying genetic syndromes. An Ultrasound Self-Superv… ▽ More The proposed study aimed to develop a deep learning model capable of detecting ventriculomegaly on prenatal ultrasound images. Ventriculomegaly is a prenatal condition characterized by dilated cerebral ventricles of the fetal brain and is important to diagnose early, as it can be associated with an increased risk for fetal aneuploidies and/or underlying genetic syndromes. An Ultrasound Self-Supervised Foundation Model with Masked Autoencoding (USF-MAE), recently developed by our group, was fine-tuned for a binary classification task to distinguish fetal brain ultrasound images as either normal or showing ventriculomegaly. The USF-MAE incorporates a Vision Transformer encoder pretrained on more than 370,000 ultrasound images from the OpenUS-46 corpus. For this study, the pretrained encoder was adapted and fine-tuned on a curated dataset of fetal brain ultrasound images to optimize its performance for ventriculomegaly detection. Model evaluation was conducted using 5-fold cross-validation and an independent test cohort, and performance was quantified using accuracy, precision, recall, specificity, F1-score, and area under the receiver operating characteristic curve (AUC). The proposed USF-MAE model reached an F1-score of 91.76% on the 5-fold cross-validation and 91.78% on the independent test set, with much higher scores than those obtained by the baseline models by 19.37% and 16.15% compared to VGG-19, 2.31% and 2.56% compared to ResNet-50, and 5.03% and 11.93% compared to ViT-B/16, respectively. The model also showed a high mean test precision of 94.47% and an accuracy of 97.24%. The Eigen-CAM (Eigen Class Activation Map) heatmaps showed that the model was focusing on the ventricle area for the diagnosis of ventriculomegaly, which has explainability and clinical plausibility. △ Less

Submitted 20 November, 2025; v1 submitted 10 November, 2025; originally announced November 2025.

Comments: 13 pages, 7 figures, 3 tables

EchoMark: Perceptual Acoustic Environment Transfer with Watermark-Embedded Room Impulse Response

Authors: Chenpei Huang, Lingfeng Yao, Kyu In Lee, Lan Emily Zhang, Xun Chen, Miao Pan

Abstract: Acoustic Environment Matching (AEM) is the task of transferring clean audio into a target acoustic environment, enabling engaging applications such as audio dubbing and auditory immersive virtual reality (VR). Recovering similar room impulse response (RIR) directly from reverberant speech offers more accessible and flexible AEM solution. However, this capability also introduces vulnerabilities of… ▽ More Acoustic Environment Matching (AEM) is the task of transferring clean audio into a target acoustic environment, enabling engaging applications such as audio dubbing and auditory immersive virtual reality (VR). Recovering similar room impulse response (RIR) directly from reverberant speech offers more accessible and flexible AEM solution. However, this capability also introduces vulnerabilities of arbitrary ``relocation" if misused by malicious user, such as facilitating advanced voice spoofing attacks or undermining the authenticity of recorded evidence. To address this issue, we propose EchoMark, the first deep learning-based AEM framework that generates perceptually similar RIRs with embedded watermark. Our design tackle the challenges posed by variable RIR characteristics, such as different durations and energy decays, by operating in the latent domain. By jointly optimizing the model with a perceptual loss for RIR reconstruction and a loss for watermark detection, EchoMark achieves both high-quality environment transfer and reliable watermark recovery. Experiments on diverse datasets validate that EchoMark achieves room acoustic parameter matching performance comparable to FiNS, the state-of-the-art RIR estimator. Furthermore, a high Mean Opinion Score (MOS) of 4.22 out of 5, watermark detection accuracy exceeding 99\%, and bit error rates (BER) below 0.3\% collectively demonstrate the effectiveness of EchoMark in preserving perceptual quality while ensuring reliable watermark embedding. △ Less

Submitted 31 March, 2026; v1 submitted 9 November, 2025; originally announced November 2025.

AI as intermediary in modern-day ritual: An immersive, interactive production of the roller disco musical Xanadu at UCLA

Authors: Mira Winick, Naisha Agarwal, Chiheb Boussema, Ingrid Lee, Camilo Vargas, Jeff Burke

Abstract: Interfaces for contemporary large language, generative media, and perception AI models are often engineered for single user interaction. We investigate ritual as a design scaffold for developing collaborative, multi-user human-AI engagement. We consider the specific case of an immersive staging of the musical Xanadu performed at UCLA in Spring 2025. During a two-week run, over five hundred audienc… ▽ More Interfaces for contemporary large language, generative media, and perception AI models are often engineered for single user interaction. We investigate ritual as a design scaffold for developing collaborative, multi-user human-AI engagement. We consider the specific case of an immersive staging of the musical Xanadu performed at UCLA in Spring 2025. During a two-week run, over five hundred audience members contributed sketches and jazzercise moves that vision language models translated to virtual scenery elements and from choreographic prompts. This paper discusses four facets of interaction-as-ritual within the show: audience input as offerings that AI transforms into components of the ritual; performers as ritual guides, demonstrating how to interact with technology and sorting audience members into cohorts; AI systems as instruments "played" by the humans, in which sensing, generative components, and stagecraft create systems that can be mastered over time; and reciprocity of interaction, in which the show's AI machinery guides human behavior as well as being guided by humans, completing a human-AI feedback loop that visibly reshapes the virtual world. Ritual served as a frame for integrating linear narrative, character identity, music and interaction. The production explored how AI systems can support group creativity and play, addressing a critical gap in prevailing single user AI design paradigms. △ Less

Submitted 8 November, 2025; originally announced November 2025.

Adaptation of Foundation Models for Medical Image Analysis: Strategies, Challenges, and Future Directions

Authors: Karma Phuntsho, Abdullah, Kyungmi Lee, Ickjai Lee, Euijoon Ahn

Abstract: Foundation models (FMs) have emerged as a transformative paradigm in medical image analysis, offering the potential to provide generalizable, task-agnostic solutions across a wide range of clinical tasks and imaging modalities. Their capacity to learn transferable representations from large-scale data has the potential to address the limitations of conventional task-specific models. However, adapt… ▽ More Foundation models (FMs) have emerged as a transformative paradigm in medical image analysis, offering the potential to provide generalizable, task-agnostic solutions across a wide range of clinical tasks and imaging modalities. Their capacity to learn transferable representations from large-scale data has the potential to address the limitations of conventional task-specific models. However, adaptation of FMs to real-world clinical practice remains constrained by key challenges, including domain shifts, limited availability of high-quality annotated data, substantial computational demands, and strict privacy requirements. This review presents a comprehensive assessment of strategies for adapting FMs to the specific demands of medical imaging. We examine approaches such as supervised fine-tuning, domain-specific pretraining, parameter-efficient fine-tuning, self-supervised learning, hybrid methods, and multimodal or cross-modal frameworks. For each, we evaluate reported performance gains, clinical applicability, and limitations, while identifying trade-offs and unresolved challenges that prior reviews have often overlooked. Beyond these established techniques, we also highlight emerging directions aimed at addressing current gaps. These include continual learning to enable dynamic deployment, federated and privacy-preserving approaches to safeguard sensitive data, hybrid self-supervised learning to enhance data efficiency, data-centric pipelines that combine synthetic generation with human-in-the-loop validation, and systematic benchmarking to assess robust generalization under real-world clinical variability. By outlining these strategies and associated research gaps, this review provides a roadmap for developing adaptive, trustworthy, and clinically integrated FMs capable of meeting the demands of real-world medical imaging. △ Less

Submitted 3 November, 2025; originally announced November 2025.

ORBIT -- Open Recommendation Benchmark for Reproducible Research with Hidden Tests

Authors: Jingyuan He, Jiongnan Liu, Vishan Vishesh Oberoi, Bolin Wu, Mahima Jagadeesh Patel, Kangrui Mao, Chuning Shi, I-Ta Lee, Arnold Overwijk, Chenyan Xiong

Abstract: Recommender systems are among the most impactful AI applications, interacting with billions of users every day, guiding them to relevant products, services, or information tailored to their preferences. However, the research and development of recommender systems are hindered by existing datasets that fail to capture realistic user behaviors and inconsistent evaluation settings that lead to ambigu… ▽ More Recommender systems are among the most impactful AI applications, interacting with billions of users every day, guiding them to relevant products, services, or information tailored to their preferences. However, the research and development of recommender systems are hindered by existing datasets that fail to capture realistic user behaviors and inconsistent evaluation settings that lead to ambiguous conclusions. This paper introduces the Open Recommendation Benchmark for Reproducible Research with HIdden Tests (ORBIT), a unified benchmark for consistent and realistic evaluation of recommendation models. ORBIT offers a standardized evaluation framework of public datasets with reproducible splits and transparent settings for its public leaderboard. Additionally, ORBIT introduces a new webpage recommendation task, ClueWeb-Reco, featuring web browsing sequences from 87 million public, high-quality webpages. ClueWeb-Reco is a synthetic dataset derived from real, user-consented, and privacy-guaranteed browsing data. It aligns with modern recommendation scenarios and is reserved as the hidden test part of our leaderboard to challenge recommendation models' generalization ability. ORBIT measures 12 representative recommendation models on its public benchmark and introduces a prompted LLM baseline on the ClueWeb-Reco hidden test. Our benchmark results reflect general improvements of recommender systems on the public datasets, with variable individual performances. The results on the hidden test reveal the limitations of existing approaches in large-scale webpage recommendation and highlight the potential for improvements with LLM integrations. ORBIT benchmark, leaderboard, and codebase are available at https://www.open-reco-bench.ai. △ Less

Submitted 29 October, 2025; originally announced October 2025.

Comments: Accepted to NeurIPS 2025 Datasets & Benchmarks track

Towards Better Health Conversations: The Benefits of Context-seeking

Authors: Rory Sayres, Yuexing Hao, Abbi Ward, Amy Wang, Beverly Freeman, Serena Zhan, Diego Ardila, Jimmy Li, I-Ching Lee, Anna Iurchenko, Siyi Kou, Kartikeya Badola, Jimmy Hu, Bhawesh Kumar, Keith Johnson, Supriya Vijay, Justin Krogue, Avinatan Hassidim, Yossi Matias, Dale R. Webster, Sunny Virmani, Yun Liu, Quang Duong, Mike Schaekermann

Abstract: Navigating health questions can be daunting in the modern information landscape. Large language models (LLMs) may provide tailored, accessible information, but also risk being inaccurate, biased or misleading. We present insights from 4 mixed-methods studies (total N=163), examining how people interact with LLMs for their own health questions. Qualitative studies revealed the importance of context… ▽ More Navigating health questions can be daunting in the modern information landscape. Large language models (LLMs) may provide tailored, accessible information, but also risk being inaccurate, biased or misleading. We present insights from 4 mixed-methods studies (total N=163), examining how people interact with LLMs for their own health questions. Qualitative studies revealed the importance of context-seeking in conversational AIs to elicit specific details a person may not volunteer or know to share. Context-seeking by LLMs was valued by participants, even if it meant deferring an answer for several turns. Incorporating these insights, we developed a "Wayfinding AI" to proactively solicit context. In a randomized, blinded study, participants rated the Wayfinding AI as more helpful, relevant, and tailored to their concerns compared to a baseline AI. These results demonstrate the strong impact of proactive context-seeking on conversational dynamics, and suggest design patterns for conversational AI to help navigate health topics. △ Less

Submitted 13 September, 2025; originally announced October 2025.

Readability $\ne$ Learnability: Rethinking the Role of Simplicity in Training Small Language Models

Authors: Ivan Lee, Taylor Berg-Kirkpatrick

Abstract: Recent studies suggest that very small language models (SLMs) can generate surprisingly coherent text when trained on simplified, child-directed corpora such as TinyStories. These findings have been interpreted as evidence that readability -- characterized by accessible vocabulary, familiar narrative structure, and simple syntax -- plays a key role in enabling such capabilities to emerge. In this… ▽ More Recent studies suggest that very small language models (SLMs) can generate surprisingly coherent text when trained on simplified, child-directed corpora such as TinyStories. These findings have been interpreted as evidence that readability -- characterized by accessible vocabulary, familiar narrative structure, and simple syntax -- plays a key role in enabling such capabilities to emerge. In this paper, we challenge that interpretation. We construct synthetic datasets with matched structure but varied readability, and find that readability alone does not predict coherence or learning efficiency in SLMs. Models trained on complex, adult-level text perform comparably to those trained on simplified language, and even exhibit faster development of coherence during training. Instead, we show that statistical simplicity, as measured by n-gram diversity, is a stronger predictor of learnability. Our findings caution against the growing trend of anthropomorphizing language model training -- drawing parallels to human cognitive development without empirical basis -- and argue for more precise reasoning about what properties actually support capability emergence in small models. △ Less

Submitted 15 October, 2025; originally announced October 2025.

Comments: Accepted to COLM 2025 (Spotlight)

RAPID: An Efficient Reinforcement Learning Algorithm for Small Language Models

Authors: Lianghuan Huang, Sagnik Anupam, Insup Lee, Shuo Li, Osbert Bastani

Abstract: Reinforcement learning (RL) has emerged as a promising strategy for finetuning small language models (SLMs) to solve targeted tasks such as math and coding. However, RL algorithms tend to be resource-intensive, taking a significant amount of time to train. We propose RAPID, a novel RL algorithm that can substantially reduce the running time of RL. Our key insight is that RL tends to be costly due… ▽ More Reinforcement learning (RL) has emerged as a promising strategy for finetuning small language models (SLMs) to solve targeted tasks such as math and coding. However, RL algorithms tend to be resource-intensive, taking a significant amount of time to train. We propose RAPID, a novel RL algorithm that can substantially reduce the running time of RL. Our key insight is that RL tends to be costly due to the need to perform both inference and backpropagation during training. To maximize use of computational resources, our algorithm performs inference in large batches, and then performs off-policy policy gradient updates in mini-batches. For off-policy updates, we incorporate group advantage estimation into the policy gradient algorithm, and derive an importance weighted estimator to correct for the bias arising from off-policy learning. Our experiments demonstrate that our algorithm can reduce running time by 11%-34% on three benchmarks compared to state-of-the-art RL algorithms while maintaining similar or better accuracy. △ Less

Submitted 3 October, 2025; originally announced October 2025.

Who's Wearing? Ear Canal Biometric Key Extraction for User Authentication on Wireless Earbuds

Authors: Chenpei Huang, Lingfeng Yao, Hui Zhong, Kyu In Lee, Lan Zhang, Xiaoyong Yuan, Tomoaki Ohtsuki, Miao Pan

Abstract: Ear canal scanning/sensing (ECS) has emerged as a novel biometric authentication method for mobile devices paired with wireless earbuds. Existing studies have demonstrated the uniqueness of ear canals by training and testing machine learning classifiers on ECS data. However, implementing practical ECS-based authentication requires preventing raw biometric data leakage and designing computationally… ▽ More Ear canal scanning/sensing (ECS) has emerged as a novel biometric authentication method for mobile devices paired with wireless earbuds. Existing studies have demonstrated the uniqueness of ear canals by training and testing machine learning classifiers on ECS data. However, implementing practical ECS-based authentication requires preventing raw biometric data leakage and designing computationally efficient protocols suitable for resource-constrained earbuds. To address these challenges, we propose an ear canal key extraction protocol, \textbf{EarID}. Without relying on classifiers, EarID extracts unique binary keys directly on the earbuds during authentication. These keys further allow the use of privacy-preserving fuzzy commitment scheme that verifies the wearer's key on mobile devices. Our evaluation results demonstrate that EarID achieves a 98.7\% authentication accuracy, comparable to machine learning classifiers. The mobile enrollment time (160~ms) and earbuds processing time (226~ms) are negligible in terms of wearer's experience. Moreover, our approach is robust and attack-resistant, maintaining a false acceptance rate below 1\% across all adversarial scenarios. We believe the proposed EarID offers a practical and secure solution for next-generation wireless earbuds. △ Less

Submitted 2 October, 2025; originally announced October 2025.

CRAB: Camera-Radar Fusion for Reducing Depth Ambiguity in Backward Projection based View Transformation

Authors: In-Jae Lee, Sihwan Hwang, Youngseok Kim, Wonjune Kim, Sanmin Kim, Dongsuk Kum

Abstract: Recently, camera-radar fusion-based 3D object detection methods in bird's eye view (BEV) have gained attention due to the complementary characteristics and cost-effectiveness of these sensors. Previous approaches using forward projection struggle with sparse BEV feature generation, while those employing backward projection overlook depth ambiguity, leading to false positives. In this paper, to add… ▽ More Recently, camera-radar fusion-based 3D object detection methods in bird's eye view (BEV) have gained attention due to the complementary characteristics and cost-effectiveness of these sensors. Previous approaches using forward projection struggle with sparse BEV feature generation, while those employing backward projection overlook depth ambiguity, leading to false positives. In this paper, to address the aforementioned limitations, we propose a novel camera-radar fusion-based 3D object detection and segmentation model named CRAB (Camera-Radar fusion for reducing depth Ambiguity in Backward projection-based view transformation), using a backward projection that leverages radar to mitigate depth ambiguity. During the view transformation, CRAB aggregates perspective view image context features into BEV queries. It improves depth distinction among queries along the same ray by combining the dense but unreliable depth distribution from images with the sparse yet precise depth information from radar occupancy. We further introduce spatial cross-attention with a feature map containing radar context information to enhance the comprehension of the 3D scene. When evaluated on the nuScenes open dataset, our proposed approach achieves a state-of-the-art performance among backward projection-based camera-radar fusion methods with 62.4\% NDS and 54.0\% mAP in 3D object detection. △ Less

Submitted 6 September, 2025; originally announced September 2025.

Comments: Accepted by ICRA 2025

RICL: Adding In-Context Adaptability to Pre-Trained Vision-Language-Action Models

Authors: Kaustubh Sridhar, Souradeep Dutta, Dinesh Jayaraman, Insup Lee

Abstract: Multi-task ``vision-language-action'' (VLA) models have recently demonstrated increasing promise as generalist foundation models for robotics, achieving non-trivial performance out of the box on new tasks in new environments. However, for such models to be truly useful, an end user must have easy means to teach them to improve. For language and vision models, the emergent ability to perform in-con… ▽ More Multi-task ``vision-language-action'' (VLA) models have recently demonstrated increasing promise as generalist foundation models for robotics, achieving non-trivial performance out of the box on new tasks in new environments. However, for such models to be truly useful, an end user must have easy means to teach them to improve. For language and vision models, the emergent ability to perform in-context learning (ICL) has proven to be a versatile and highly useful interface to easily teach new tasks with no parameter finetuning. Unfortunately, VLAs pre-trained with imitation learning objectives do not naturally acquire ICL abilities. In this paper, we demonstrate that, with the right finetuning recipe and a small robot demonstration dataset, it is possible to inject in-context adaptability post hoc into such a VLA. After retraining for in-context learning (RICL), our system permits an end user to provide a small number (10-20) of demonstrations for a new task. RICL then fetches the most relevant portions of those demonstrations into the VLA context to exploit ICL, performing the new task and boosting task performance. We apply RICL to inject ICL into the $π_{0}$-FAST VLA, and show that it permits large in-context improvements for a variety of new manipulation tasks with only 20 demonstrations per task, without any parameter updates. When parameter updates on the target task demonstrations is possible, RICL finetuning further boosts performance. We release code and model weights for RICL-$π_{0}$-FAST alongside the paper to enable, for the first time, a simple in-context learning interface for new manipulation tasks. Website: https://ricl-vla.github.io. △ Less

Submitted 4 August, 2025; originally announced August 2025.

Comments: Conference on Robot Learning 2025 (CoRL 2025), 17 pages

Diffusion-Based User-Guided Data Augmentation for Coronary Stenosis Detection

Authors: Sumin Seo, In Kyu Lee, Hyun-Woo Kim, Jaesik Min, Chung-Hwan Jung

Abstract: Coronary stenosis is a major risk factor for ischemic heart events leading to increased mortality, and medical treatments for this condition require meticulous, labor-intensive analysis. Coronary angiography provides critical visual cues for assessing stenosis, supporting clinicians in making informed decisions for diagnosis and treatment. Recent advances in deep learning have shown great potentia… ▽ More Coronary stenosis is a major risk factor for ischemic heart events leading to increased mortality, and medical treatments for this condition require meticulous, labor-intensive analysis. Coronary angiography provides critical visual cues for assessing stenosis, supporting clinicians in making informed decisions for diagnosis and treatment. Recent advances in deep learning have shown great potential for automated localization and severity measurement of stenosis. In real-world scenarios, however, the success of these competent approaches is often hindered by challenges such as limited labeled data and class imbalance. In this study, we propose a novel data augmentation approach that uses an inpainting method based on a diffusion model to generate realistic lesions, allowing user-guided control of severity. Extensive evaluation on lesion detection and severity classification across various synthetic dataset sizes shows superior performance of our method on both a large-scale in-house dataset and a public coronary angiography dataset. Furthermore, our approach maintains high detection and classification performance even when trained with limited data, highlighting its clinical importance in improving the assessment of severity of stenosis and optimizing data utilization for more reliable decision support. △ Less

Submitted 1 August, 2025; originally announced August 2025.

Comments: Accepted at MICCAI 2025. Dataset available at https://github.com/medipixel/DiGDA

LiteFat: Lightweight Spatio-Temporal Graph Learning for Real-Time Driver Fatigue Detection

Authors: Jing Ren, Suyu Ma, Hong Jia, Xiwei Xu, Ivan Lee, Haytham Fayek, Xiaodong Li, Feng Xia

Abstract: Detecting driver fatigue is critical for road safety, as drowsy driving remains a leading cause of traffic accidents. Many existing solutions rely on computationally demanding deep learning models, which result in high latency and are unsuitable for embedded robotic devices with limited resources (such as intelligent vehicles/cars) where rapid detection is necessary to prevent accidents. This pape… ▽ More Detecting driver fatigue is critical for road safety, as drowsy driving remains a leading cause of traffic accidents. Many existing solutions rely on computationally demanding deep learning models, which result in high latency and are unsuitable for embedded robotic devices with limited resources (such as intelligent vehicles/cars) where rapid detection is necessary to prevent accidents. This paper introduces LiteFat, a lightweight spatio-temporal graph learning model designed to detect driver fatigue efficiently while maintaining high accuracy and low computational demands. LiteFat involves converting streaming video data into spatio-temporal graphs (STG) using facial landmark detection, which focuses on key motion patterns and reduces unnecessary data processing. LiteFat uses MobileNet to extract facial features and create a feature matrix for the STG. A lightweight spatio-temporal graph neural network is then employed to identify signs of fatigue with minimal processing and low latency. Experimental results on benchmark datasets show that LiteFat performs competitively while significantly decreasing computational complexity and latency as compared to current state-of-the-art methods. This work enables the development of real-time, resource-efficient human fatigue detection systems that can be implemented upon embedded robotic devices. △ Less

Submitted 13 August, 2025; v1 submitted 29 July, 2025; originally announced July 2025.

Comments: 8 pages, 4 figures

Projecting the New Body: How Body Image Evolves During Learning to Walk with a Wearable Robot

Authors: I-Chieh Lee, He Huang

Abstract: Advances in wearable robotics challenge the traditional definition of human motor systems, as wearable robots redefine body structure, movement capability, and perception of their own bodies. We measured gait performance and perceived body images via Selected Coefficient of Perceived Motion, SCoMo, after each training session. Based on human motor learning theory extended to wearer-robot systems,… ▽ More Advances in wearable robotics challenge the traditional definition of human motor systems, as wearable robots redefine body structure, movement capability, and perception of their own bodies. We measured gait performance and perceived body images via Selected Coefficient of Perceived Motion, SCoMo, after each training session. Based on human motor learning theory extended to wearer-robot systems, we hypothesized that learning the perceived body image when walking with a robotic leg co-evolves with the actual gait improvement and becomes more certain and more accurate to the actual motion. Our result confirmed that motor learning improved both physical and perceived gait pattern towards normal, indicating that via practice the wearers incorporated the robotic leg into their sensorimotor systems to enable wearer-robot movement coordination. However, a persistent discrepancy between perceived and actual motion remained, likely due to the absence of direct sensation and control of the prosthesis from wearers. Additionally, the perceptual overestimation at the later training sessions might limit further motor improvement. These findings suggest that enhancing the human sense of wearable robots and frequent calibrating perception of body image are essential for effective training with lower limb wearable robots and for developing more embodied assistive technologies. △ Less

Submitted 28 July, 2025; originally announced July 2025.