Showing 1–50 of 414 results for author: Chung, S

UniVerse: Benchmarking and Enhancing LALMs on Culturally Inclusive Low-Resource Music Understanding

Authors: Ziya Zhou, Shangda Wu, Shenyang Xu, Yutong Zheng, Dafang Liang, Suin Chung, Danbinaerin Han, Junyan Jiang, Yongyi Zang, Ruibin Yuan, Rongxiu Zhong, Shilei Zhang, Junlan Feng, Jinglei Liu, Haotian Zhou, Zijin Li, Dasaem Jeong, Wei Xue, Yike Guo

Abstract: Recent advances in large audio-language models (LALMs) have significantly improved performance in tasks such as music captioning, genre classification, and sound event detection. However, limited attention has been paid to improving their adaptability across diverse musical traditions, particularly folk music rooted in distinct cultural contexts. Folk-music traditions are typically resource-scarce… ▽ More Recent advances in large audio-language models (LALMs) have significantly improved performance in tasks such as music captioning, genre classification, and sound event detection. However, limited attention has been paid to improving their adaptability across diverse musical traditions, particularly folk music rooted in distinct cultural contexts. Folk-music traditions are typically resource-scarce, unevenly represented across regions, and poorly documented. Even when such samples appear in large-scale pre-training, LALMs often fail to capture their structural and stylistic characteristics, partly due to the absence of dedicated evaluation protocols and training solutions. To address these limitations, we introduce UniVerse, a reproducible solution for low-resource music understanding. Specifically, we propose UniVerseBench, a benchmark of 5,042 Q&A pairs across more than 38 cultural and linguistic entities, constructed via an expert-guided yet highly automated pipeline. In parallel, we construct a fully automated, model-generated multi-turn dialogue training dataset UniVerseSet. By training LALMs on UniVerseSet, we systematically adapt and investigate representative multimodal imbalance learning strategies across both dense and Mixture-of-Experts (MoE) architectures. Experimental results indicate that fully automated data curation combined with imbalance-aware training yields non-trivial improvements, but models still struggle to capture fine-grained acoustic features, indicating a gap between surface-level alignment and deep musical comprehension. △ Less

Submitted 18 August, 2026; originally announced August 2026.

Comments: 21 pages, 7 figures, 8 tables

Pitch Contour Tokenization using VQ-VAE and Its Application on Korean Traditional Music Analysis

Authors: Seonguk Ju, Seola Cho, Sooin Chung, Danbinaerin Han, Dasaem Jeong

Abstract: Computational analysis of music often relies on discrete representations, yet many musical traditions are organized around continuous pitch movement that resists segmentation into note-like units. For such traditions, the discrete units that analysis would build on are not given in advance. We address this gap by learning a vocabulary of local pitch-contour patterns directly from unlabeled audio,… ▽ More Computational analysis of music often relies on discrete representations, yet many musical traditions are organized around continuous pitch movement that resists segmentation into note-like units. For such traditions, the discrete units that analysis would build on are not given in advance. We address this gap by learning a vocabulary of local pitch-contour patterns directly from unlabeled audio, using a VQ-VAE that quantizes fixed-length contour segments into a finite codebook. To make the learned tokens stable across segmentation positions and small variations in timing and pitch range, we train the model with a reconstruction objective evaluated under the best alignment among a set of candidate temporal and pitch-domain transformations. Applied to Korean traditional music, the learned tokens recover information about expert-defined sigimsae categories without supervision, and in pansori individual tokens align with the two principal modes, Gyemyeonjo and Ujo, supporting their use as units for corpus-level analysis of contour-centric traditions. △ Less

Submitted 11 August, 2026; originally announced August 2026.

Comments: 8 pages, 3 figures, 2 tables. Accepted at ISMIR 2026

Motif 3: Technical Report

Authors: Junghwan Lim, Joon Son Chung, Sungmin Lee, Wai Ting Cheung, Gihun Cho, Minsu Ha, Sangho Kang, Beomgyu Kim, Dongseok Kim, Jangwoong Kim, Taehyun Kim, Taewhan Kim, Jeesoo Lee, Jeongdoo Lee, Junhyeok Lee, Dongpin Oh, Hyeyeon Cho, Dahye Choi, Jaeheui Her, Hanbin Jung, Changjin Kang, Minjae Kim, Youngrok Kim, Hyukjin Kweon, Hongjoo Lee , et al. (2 additional authors not shown)

Abstract: We introduce Motif 3, a decoder-only Mixture-of-Experts language model with 314 billion total parameters and 13.2 billion activated per token. Each sparse MoE layer contains 384 routed experts, with eight selected per token. This fine-grained sparsity provides substantial expert capacity while limiting computation. Motif 3 is built around Grouped Differential Latent Attention (GDLA), which integra… ▽ More We introduce Motif 3, a decoder-only Mixture-of-Experts language model with 314 billion total parameters and 13.2 billion activated per token. Each sparse MoE layer contains 384 routed experts, with eight selected per token. This fine-grained sparsity provides substantial expert capacity while limiting computation. Motif 3 is built around Grouped Differential Latent Attention (GDLA), which integrates grouped differential attention with the compressed key-value representation of Multi-head Latent Attention. The architecture further incorporates modified manifold-constrained hyper-connections, Expert Specific PolyNorm activations, and multi-token prediction to improve optimization stability, expert specialization, and inference efficiency. We pretrain Motif 3 on approximately 12.5 trillion tokens spanning web documents, STEM, code, mathematics, multilingual content, and domain-specialized corpora. Expert-balancing and numerical-stabilization techniques support stable training at scale, while selective MXFP8 computation and communication, memory-efficient fused kernels, and window-aware context parallelism enable training with context lengths up to 256K tokens. Our post-training pipeline combines general supervised fine-tuning, six specialist teachers trained with reinforcement learning, a software-engineering teacher trained with supervised fine-tuning, and Multi-teacher On-Policy Distillation. The resulting unified model consolidates complementary capabilities in reasoning, coding, tool use, professional work, long-context understanding, calibrated abstention, and instruction following. Across a broad evaluation suite, Motif 3 demonstrates competitive performance against leading open weight models, including strong results on long-horizon agentic tasks, mathematical reasoning, scientific knowledge, and hallucination-sensitive evaluation. △ Less

Submitted 10 August, 2026; originally announced August 2026.

Frame-Level Pansori Mode Classification with Complementary Audio Representations

Authors: Sangheon Park, Seonguk Ju, Suin Chung, Danbinaerin Han, Dasaem Jeong

Abstract: Pansori is a traditional Korean vocal genre whose mode system (jo) is defined not by scale alone but by the entanglement of pitch collection, microtonal ornament (sigimsae), and vocal timbre. In this study, we introduce a 46-hour frame-level pansori mode annotation, expert-labeled across all five canonical batang, and evaluate four complementary input representations (mel spectrogram, F0 contour,… ▽ More Pansori is a traditional Korean vocal genre whose mode system (jo) is defined not by scale alone but by the entanglement of pitch collection, microtonal ornament (sigimsae), and vocal timbre. In this study, we introduce a 46-hour frame-level pansori mode annotation, expert-labeled across all five canonical batang, and evaluate four complementary input representations (mel spectrogram, F0 contour, MIDI piano roll, and a multi-cultural SSL encoder) under two split strategies designed to detect shortcut learning. Across the three well-represented modes, performance degrades by only 2.1--3.6 points of F1 when entire works are held out, indicating that the models learn mode-relevant features rather than memorizing repertoire. Per-class results further show that source separation removes the percussion cue on which changjo depends, and that generic multi-cultural pre-training fails specifically on the Ujo--Gyemyeonjo distinction. Qualitative analysis of cross-modal disagreement recovers musicologically documented phenomena and agrees with published score-based analyses of modern changjak pansori. △ Less

Submitted 6 August, 2026; originally announced August 2026.

Comments: Accepted to the 27th International Society for Music Information Retrieval (ISMIR) Conference, 2026

Journal ref: ISMIR 2026

Rethinking Personalized Reward Modeling for LLMs under Preference Heterogeneity via Group-Debiased Federated Learning

Authors: Seongyoon Kim, Boryeong Cho, Jihwan Oh, Seokhyun Chung, Se-Young Yun

Abstract: Large language models are increasingly aligned to human preferences via reward modeling, but user preference data are sensitive and often cannot be centralized. Federated learning keeps such data local while learning a shared initial reward model, which is later personalized for each client through local fine-tuning. Because users often assign opposite labels to the same pair of responses, existin… ▽ More Large language models are increasingly aligned to human preferences via reward modeling, but user preference data are sensitive and often cannot be centralized. Federated learning keeps such data local while learning a shared initial reward model, which is later personalized for each client through local fine-tuning. Because users often assign opposite labels to the same pair of responses, existing federated methods address preference heterogeneity by clustering similar clients and training one reward model per group, assuming that each group requires its own initialization. We show that this assumption is unnecessary. Under balanced preference groups, a single FedAvg model, despite starting at nearly random accuracy, surpasses reward models trained separately for each ground-truth group after only a few local optimization steps. We attribute this phenomenon to the flatness of the shared initialization: averaging across all clients learns richer shared representations that distinguish responses while canceling conflicting preference directions, leaving the model near a decision boundary that can be rapidly adapted. Group imbalance breaks this effect as the cancellation becomes asymmetric and leaves minority clients too far from the boundary to recover. Motivated by this observation, we propose FedGD (Federated Learning with Group Debiasing), which discovers latent preference groups during federated training and learns a single reward model using group-debiased client sampling. By counteracting the effect of group imbalance, FedGD learns an initialization that remains highly adaptable, enabling effective personalization without prior knowledge of the underlying groups. △ Less

Submitted 2 August, 2026; originally announced August 2026.

Out of Sight, Still in Mind: Token Compression for Omni-LLMs

Authors: Suho Yoo, Youngjoon Jang, Hyebin Cho, Joon Son Chung

Abstract: The goal of this paper is to reduce the input token cost of Omni-modal large language models (Omni-LLMs) at inference time. Omni-LLMs reason jointly over audio, video and text, but the cost of the three streams is highly unbalanced: visual tokens account for the vast majority of the input, and are highly redundant. In this paper, we propose ReMo, a training-free framework that compresses visual to… ▽ More The goal of this paper is to reduce the input token cost of Omni-modal large language models (Omni-LLMs) at inference time. Omni-LLMs reason jointly over audio, video and text, but the cost of the three streams is highly unbalanced: visual tokens account for the vast majority of the input, and are highly redundant. In this paper, we propose ReMo, a training-free framework that compresses visual tokens by redistributing their information across modalities: a visual token is kept only if its information appears nowhere else. ReMo achieves this in two ways: (i) it aligns audio and video in a common embedding space, and removes visual tokens already explained by the audio or by other visual tokens; and (ii) it replaces object-level visual tokens with compact text proxies, short descriptions of each object and its location, conveying the same content in far fewer tokens. On Qwen2.5-Omni at two model scales, ReMo removes 54% of the input tokens with no loss in accuracy. Indeed, it slightly exceeds the full-token model, reaching 101.2% and 101.3% of its average accuracy over five audio-visual benchmarks. △ Less

Submitted 23 July, 2026; originally announced July 2026.

Comments: Preprint

From Voting to Agent Collaboration: Answer-Type-Aware LLM Pipelines for BioASQ 14b

Authors: Taeyun Roh, Eunha Lee, Wonjune Jang, Sohyun Chung, Junha Jung, Jaewoo Kang

Abstract: Biomedical question answering requires not only accurate extraction of information from scientific literature but also reliable integration of evidence across multiple documents. This study presents a question-type-specific large language model (LLM) framework for BioASQ 14b Task B, designed to improve answer robustness and evidence grounding in biomedical question answering. Rather than applying… ▽ More Biomedical question answering requires not only accurate extraction of information from scientific literature but also reliable integration of evidence across multiple documents. This study presents a question-type-specific large language model (LLM) framework for BioASQ 14b Task B, designed to improve answer robustness and evidence grounding in biomedical question answering. Rather than applying a single prompting strategy to all questions, the framework selects different inference procedures for yes/no, factoid, and list questions according to their distinct reasoning and evaluation requirements. For yes/no questions, snippet shuffling and self-reflection are used to reduce sensitivity to evidence ordering and improve decision stability. For factoid questions, full-snippet input is combined with chain-of-thought-based in-context learning to support accurate biomedical entity identification. For list questions, a multi-agent architecture is employed, in which evidence extraction, candidate generation, answer verification, and final aggregation are handled collaboratively. Preliminary experiments on BioASQ 13b were used to identify effective inference strategies for each question type, and the resulting framework was subsequently evaluated in the official BioASQ 14b Task B challenge. In the official evaluation, our framework showed competitive performance across multiple batches and achieved first place in the factoid subtask of Batch 4. These results demonstrate the effectiveness of combining question-type-specific inference, ensemble prediction, and agent-based verification for reliable biomedical question answering. △ Less

Submitted 7 July, 2026; originally announced July 2026.

Comments: 15 pages

Sampling-Based Coordination-Informed Multi-Objective Multi-Robot Reinforcement Learning

Authors: Antonio Marino, Esteban Restrepo, Soon-jo Chung, Paolo Robuffo Giordano, Claudio Pacchierotti

Abstract: Multi-robot systems must simultaneously optimize competing objectives while maintaining coordinated behavior. Existing multi-agent reinforcement learning approaches often rely on fixed or centralized coordination, which limits adaptability and violates distributed constraints. This work introduces the Coordination-Informed Multi-Objective Reinforcement Learning (CIMORL) framework, integrating a di… ▽ More Multi-robot systems must simultaneously optimize competing objectives while maintaining coordinated behavior. Existing multi-agent reinforcement learning approaches often rely on fixed or centralized coordination, which limits adaptability and violates distributed constraints. This work introduces the Coordination-Informed Multi-Objective Reinforcement Learning (CIMORL) framework, integrating a distributed weight prediction mechanism, a privileged expert training strategy, and theoretical guarantees for Pareto-optimal solutions. We present the base CIMORL method alongside two sampling-based variants, CIMORL-TS (Tree Search) and CIMORL-MPPI (MPPI), which leverage privileged global information during training to enable fully decentralized deployment. Experimental validation in cooperative and adversarial scenarios demonstrates a $21.2\%$ hypervolume improvement and superior policy stability compared to state-of-the-art baselines. Real-world experiments with Crazyflie drones further validate the framework's robustness in resource allocation and multi-attacker multi-defend scenarios under partial observability. △ Less

Submitted 14 August, 2026; v1 submitted 29 June, 2026; originally announced June 2026.

Comments: 20 pages, 11 figures, 4 tables

See & Sniff: Learning Visuo-Olfactory Representations

Authors: Seongyu Kim, Seungwoo Lee, Hyeonggon Ryu, Joon Son Chung, Arda Senocak

Abstract: While modern multimodal models integrate vision with language, audio, or touch, olfaction remains largely unexplored due to the lack of paired visuo-olfactory data. We introduce SmellNet-V, a scalable visuo-olfactory dataset built on the insight that odor identity is largely invariant to visual transformations within a semantic category. This allows us to synthetically pair smell-only samples with… ▽ More While modern multimodal models integrate vision with language, audio, or touch, olfaction remains largely unexplored due to the lack of paired visuo-olfactory data. We introduce SmellNet-V, a scalable visuo-olfactory dataset built on the insight that odor identity is largely invariant to visual transformations within a semantic category. This allows us to synthetically pair smell-only samples with semantically aligned in-the-wild web images, converting a unimodal olfactory dataset into a cross-modal benchmark without costly co-collection. Building on this dataset, we propose See & Sniff, a self-supervised framework that learns joint visuo-olfactory representations via dense local alignment and naturally produces smell saliency maps for spatial grounding of odor sources. We further introduce pixel-level smell localization task and a benchmark for evaluation. Our method surpasses smell-only baselines by 7% in smell classification from smell alone and generalizes to cross-modal retrieval and smell localization, establishing visuo-olfactory learning as a new direction in multimodal perception. △ Less

Submitted 25 June, 2026; originally announced June 2026.

Comments: ECCV 2026. Project Page: https://mm.kaist.ac.kr/projects/SeeandSniff/

ProsoCodec: Prosody-Oriented Speech Codec for Voice Conversion

Authors: Jeongsoo Choi, Ji-Hoon Kim, Shujie Hu, Joon Son Chung

Abstract: Neural speech codecs efficiently compress speech and have become a foundation for speech generation, but they are typically learned as holistic representations that intertwine linguistic content, speaker identity, and prosody. While this design is effective for zero-shot voice cloning, it hinders downstream tasks that require prosody preservation or transfer, such as voice conversion. To address t… ▽ More Neural speech codecs efficiently compress speech and have become a foundation for speech generation, but they are typically learned as holistic representations that intertwine linguistic content, speaker identity, and prosody. While this design is effective for zero-shot voice cloning, it hinders downstream tasks that require prosody preservation or transfer, such as voice conversion. To address this, we introduce ProsoCodec, a prosody-oriented speech codec that models prosody as a conditional residual rather than as a disentangled stream. Specifically, by conditioning both the encoder and decoder on text and speaker embeddings as prefix tokens, the discrete bottleneck is encouraged to capture prosodic variation not explained by content and speaker. To further preserve prosody, we use the low-frequency mel band and train the model on paired same-speaker utterances. Experiments on voice conversion show improved prosody preservation and reduced source-timbre leakage. △ Less

Submitted 20 June, 2026; originally announced June 2026.

Comments: Interspeech 2026

MixProLAP: Mixture-Induced Uncertainty Modeling for Probabilistic Language-Audio Pretraining

Authors: Yu Nakagome, Jaesong Lee, Soo-Whan Chung

Abstract: Acoustic environments often contain multiple overlapping sound events, and the same acoustic scene can be described using diverse textual expressions, making audio-text alignment inherently ambiguous. This paper proposes a probabilistic audio-language pretraining framework to model many-to-many correspondence ambiguity in audio-text alignment. Unlike conventional contrastive methods that learn det… ▽ More Acoustic environments often contain multiple overlapping sound events, and the same acoustic scene can be described using diverse textual expressions, making audio-text alignment inherently ambiguous. This paper proposes a probabilistic audio-language pretraining framework to model many-to-many correspondence ambiguity in audio-text alignment. Unlike conventional contrastive methods that learn deterministic point embeddings, our approach represents each modality as a distribution and learns uncertainty-aware cross-modal alignment. Rather than relying on masking-based uncertainty simulation, we mix audio-text pairs to create overlapping sounds that better reflect real acoustic mixtures and capture semantic inclusion relations among sound events. We further introduce a multi-level inclusion loss to enforce representations consistent with these relations. Experiments on audio-text retrieval benchmarks show that the proposed method outperforms deterministic baselines. △ Less

Submitted 18 June, 2026; originally announced June 2026.

Comments: Accepted to Interspeech 2026

Who Wins the Conflict? Mechanistic Interpretability of Text Bias in Audio LLMs

Authors: Hyebin Cho, Suho Yoo, Jaehyuk Jang, Changick Kim, Joon Son Chung

Abstract: While Audio Large Language Models (Audio LLMs) excel at multimodal understanding, they suffer from text dominance, a bias where models blindly favor text over acoustic evidence, causing hallucinations. However, the internal mechanisms underlying how these models behave when audio and textual inputs contradict each other remain unexplored. In this work, we present the first mechanistic analysis of… ▽ More While Audio Large Language Models (Audio LLMs) excel at multimodal understanding, they suffer from text dominance, a bias where models blindly favor text over acoustic evidence, causing hallucinations. However, the internal mechanisms underlying how these models behave when audio and textual inputs contradict each other remain unexplored. In this work, we present the first mechanistic analysis of this phenomenon by tracing the propagation of internal representations across layers. Our investigation reveals three key findings: (i) text dominance is systematically and empirically across models; (ii) while text and audio rely on functionally distinct pathways, they ultimately converge into a shared semantic space in late layers; and (iii) the text pathway does not erase audio information, but rather actively suppresses intact audio representations. Building on these insights, we leverage back-patching, a training-free intervention that routes late-layer audio activations back into earlier layers. This amplifies the audio representations, enabling them to overcome textual suppression. Our evaluation shows that back-patching consistently reduces text dominance, paving the way for mechanistic multimodal alignment under conflict. △ Less

Submitted 17 June, 2026; originally announced June 2026.

Comments: Preprint

Acoustic Prompting via Stage-wise Modulation for Few-Shot Learning in Audio Language Models

Authors: Hyebin Cho, Jaehyuk Jang, Changick Kim, Joon Son Chung

Abstract: Audio-Language Models (ALMs) have shown remarkable success in zero-shot audio classification by aligning audio waveforms with text. Recent efforts to improve downstream performance focus on learning optimal text prompts. However, previous approaches focus on the text encoder, leaving the potential of learnable prompts within the audio encoder unexplored. In this paper, we propose a novel framework… ▽ More Audio-Language Models (ALMs) have shown remarkable success in zero-shot audio classification by aligning audio waveforms with text. Recent efforts to improve downstream performance focus on learning optimal text prompts. However, previous approaches focus on the text encoder, leaving the potential of learnable prompts within the audio encoder unexplored. In this paper, we propose a novel framework that introduces trainable prompts into the audio encoder to capture task-specific acoustic features. We demonstrate that integrating audio-side prompt learning with existing text-side approaches enhances few-shot adaptation. Through extensive experiments across 11 datasets show that integrating our method as a plug-and-play module alongside existing text prompt tuning generally leads to performance improvements. These findings suggest that explicitly modulating the audio representation space effectively complements text-only prompting approaches. The code is available at https://github.com/hyebin-c/aspl. △ Less

Submitted 14 June, 2026; originally announced June 2026.

Comments: Accepted to INTERSPEECH 2026

PAI: Preserving Amplitude Information in Representation-Based Time-Series Anomaly Detection

Authors: Kang Zhang, Wei Jian Lau, Shoushou Ren, Dong Lin, Joon Son Chung, Chuanhao Sun

Abstract: Representation-based time-series anomaly detection algorithms significantly outperform other methods on diverse anomaly detection tasks. However, we notice that they suffer from a major limitation in our evaluation - their learned embeddings are often amplitude-agnostic. Losing amplitude information can degrade performance on amplitude related anomalies, and this failure is prevalent across all ex… ▽ More Representation-based time-series anomaly detection algorithms significantly outperform other methods on diverse anomaly detection tasks. However, we notice that they suffer from a major limitation in our evaluation - their learned embeddings are often amplitude-agnostic. Losing amplitude information can degrade performance on amplitude related anomalies, and this failure is prevalent across all existing representation-based methods. To address aforementioned issues, we propose a new anomaly scoring scheme named PAI. PAI consists of two complementary modules, a diagnostic module and a final score augmentation function. The diagnostic module compares cosine and Euclidean scoring on the same representation bank to test whether amplitude information is already captured in the learned representation. Then in final score augmentation function, PAI computes a point-wise median and MAD deviation score and a local mean-shift score-which are fused with the representation score to produce the final anomaly score. On the TSB-AD-U-Eva and TAB UV datasets, PAI improves all four evaluated representation-based methods across every reported metric, achieving average VUS-PR gains of 98.4% and 36.8%, respectively. Among all evaluated combinations, PaAno + PAI achieves the best performance, outperforming the state-of-the-art method by 15%. Further evaluation on bootstrap confidence intervals, anomaly-type breakdowns, and a TS2Vec input-normalization ablation further support the proposed scheme. These results suggest that explicitly retaining amplitude information is important for representation-based time-series anomaly detection, which has been underemphasized in existing scoring schemes. Code is available at: https://github.com/pantheon5100/PAI △ Less

Submitted 7 June, 2026; originally announced June 2026.

Comments: 15 pages

Inference-Time Scaling for Joint Audio-Video Generation

Authors: Jaemin Jung, Kyeongha Rho, Inkyu Shin, Joon Son Chung

Abstract: Joint audio-video generation aims to synthesize realistic audio-video pairs that are both semantically aligned with text prompts and precisely synchronized. While existing joint audio-video generation models often require substantial training resources to improve fidelity, Inference-Time Scaling (ITS) has recently emerged as a promising training-free alternative in single-modality domains. However… ▽ More Joint audio-video generation aims to synthesize realistic audio-video pairs that are both semantically aligned with text prompts and precisely synchronized. While existing joint audio-video generation models often require substantial training resources to improve fidelity, Inference-Time Scaling (ITS) has recently emerged as a promising training-free alternative in single-modality domains. However, extending ITS from a single modality to multimodal domains is non-trivial, as it requires balancing multiple heterogeneous objectives. In this paper, we present the first comprehensive study of ITS for joint audio-video generation. We first demonstrate that a multi-verifier framework is essential to address the limitations of single-objective guidance, including asymmetric performance trade-offs and verifier hacking. Through systematic analysis, we then identify an optimal multi-verifier combination that yields balanced improvements across all quality dimensions. Finally, to effectively aggregate diverse reward signals, we propose Adaptive Reward Weighting (ARW), a novel test-time optimization algorithm. ARW treats reward aggregation as an online optimization problem, utilizing learnable parameters to calibrate reward variances without requiring prior knowledge of reward distributions, thereby ensuring robust multi-objective selection. Experimental results on VGGSound and JavisBench-mini benchmarks demonstrate that our framework significantly enhances semantic alignment, perceptual quality, and audio-visual synchronization of generated outputs. Synthesized samples and code are available on the project page: https://jung-jaemin.github.io/ITS-AVGen-Proj. △ Less

Submitted 2 June, 2026; originally announced June 2026.

Comments: Accepted by Transactions on Machine Learning Research (TMLR). Project page: https://jung-jaemin.github.io/ITS-AVGen-Proj/

Task-Induced Representational Invariances Depend on Learning Objective in Deep RL

Authors: Manu Srinath Halvagal, Sebastian Lee, SueYeon Chung

Abstract: Reinforcement Learning (RL) has long served as a model for goal-directed animal behavior in neuroscience. Modern deep RL has shown remarkable success across many domains, further strengthening this connection. The ability to learn abstract representations of high-dimensional state spaces underlies much of this success. However, theoretical understanding of these learned representations remains lim… ▽ More Reinforcement Learning (RL) has long served as a model for goal-directed animal behavior in neuroscience. Modern deep RL has shown remarkable success across many domains, further strengthening this connection. The ability to learn abstract representations of high-dimensional state spaces underlies much of this success. However, theoretical understanding of these learned representations remains limited, hindering direct comparisons between models and animal learning. We address this gap by analyzing deep RL representations through the lens of MDP reduction theory. Investigating canonical RL algorithms in a navigation task, we find that even when performance is comparable, the value-based method (DQN) learns representations that are invariant to MDP homomorphism symmetries, while the policy-gradient method (PPO) learns representations invariant to action symmetries. These differences emerge consistently across domains, have downstream consequences for transfer learning, and appear in LLMs in a prompt-dependent manner. Our findings provide a principled approach to comparing learned representations across RL algorithms, with demonstrated practical implications and possible insights for neural coding in the brain. △ Less

Submitted 1 June, 2026; originally announced June 2026.

Universal Quantum Transformer

Authors: Sungyong Chung, Alireza Talebpour

Abstract: Classical continuous-space neural networks fundamentally struggle to lock into exact formal rules, whether mathematical, such as modular arithmetic and non-Abelian group algebra, or linguistic, such as systematic compositional generalization. To approximate these discrete logical rules, they often rely on massive parameter scaling, resulting in stochastic instability even after delayed generalizat… ▽ More Classical continuous-space neural networks fundamentally struggle to lock into exact formal rules, whether mathematical, such as modular arithmetic and non-Abelian group algebra, or linguistic, such as systematic compositional generalization. To approximate these discrete logical rules, they often rely on massive parameter scaling, resulting in stochastic instability even after delayed generalization phenomena known as grokking. Here, we introduce the Universal Quantum Transformer (UQT), a novel, quantum-native computing architecture that uses the physical properties of multi-qubit systems as a universal inductive bias for exact algebraic and compositional reasoning. Rather than translating classical neural mechanisms, our framework relies entirely on parameterized geometric phase embedding and $SU(2)$ wave-interference. We demonstrate that an identical quantum attention circuit, operating on a highly compact 5 or 6 qubit substrate with only 551 to 1,650 trainable parameters, exactly learns three highly distinct formal classes: cyclic modular arithmetic ($\mathbb{Z}_{11}$), non-Abelian algebra (the $S_4$ permutation group), and systematic linguistic compositionality (the SCAN language). While standard classical models, including multi-layer perceptrons (MLPs) and Transformers, exhibit stochastic instability at convergence, the UQT achieves mathematically exact, deterministic generalization. We define this stricter regime as crystallization: a step beyond the well-known phenomenon of grokking. Finally, we deploy the UQT on noisy intermediate-scale quantum (NISQ) hardware, achieving 97.5% accuracy on IBM Quantum computers. These results demonstrate that the UQT provides a structurally suited inductive bias for exact formal reasoning that standard classical continuous-space architectures do not natively provide. △ Less

Submitted 27 July, 2026; v1 submitted 29 April, 2026; originally announced June 2026.

KVoiceBench, KOpenAudioBench, and KMMAU: Agent-Driven Korean Speech Benchmarks for Evaluating SpeechLMs

Authors: Haechan Kim, Seungjun Chung, Inkyu Park, Jihoo Lee, Jonghyun Lee

Abstract: Speech language models (SpeechLMs) have achieved substantial progress by extending large language models (LLMs) to the speech modality. However, SpeechLM evaluation remains heavily centered on English, limiting reliable assessment of multilingual speech capabilities. Straightforward benchmark transfer through ASR, translation, normalization, and TTS can corrupt language-specific instructions, answ… ▽ More Speech language models (SpeechLMs) have achieved substantial progress by extending large language models (LLMs) to the speech modality. However, SpeechLM evaluation remains heavily centered on English, limiting reliable assessment of multilingual speech capabilities. Straightforward benchmark transfer through ASR, translation, normalization, and TTS can corrupt language-specific instructions, answer constraints, and spoken forms; for audio understanding, transferring source-language audio also fails to preserve target-language speaker attributes, accents, and paralinguistic properties. To address these limitations, we propose two human-agent benchmark-construction frameworks: one transfers source-language SpokenQA benchmarks into target-language SpokenQA benchmarks, and the other converts target-language ASR corpora into audio understanding benchmarks using transcriptions and speaker metadata. Using these frameworks, we construct and publicly release three Korean speech benchmarks: KVoiceBench and KOpenAudioBench for Korean SpokenQA, and KMMAU for Korean audio understanding, comprising 12,345 samples in total. We evaluate eight recent SpeechLMs and find that English-Korean performance gaps vary substantially across models and task families, and that SpokenQA and audio understanding rankings diverge, revealing complementary weaknesses invisible to English-only evaluation. △ Less

Submitted 27 May, 2026; originally announced May 2026.

Comments: 16 pages, 4 figures

Two Speeds of Learning: A Representation-Readout Decomposition of Grokking and Double Descent

Authors: Chi-Ning Chou, Oscar Uzdelewicz, Neng-Chun Chiu, Yao-Yuan Yang, SueYeon Chung

Abstract: Training loss and accuracy are the standard signals used to monitor generalization during deep neural network training. Two well-documented phenomena complicate this picture: in grokking, train loss falls rapidly while test performance improves abruptly only after a long delay; in epoch-wise double descent, train loss decreases monotonically while test loss or error rises and falls. Existing accou… ▽ More Training loss and accuracy are the standard signals used to monitor generalization during deep neural network training. Two well-documented phenomena complicate this picture: in grokking, train loss falls rapidly while test performance improves abruptly only after a long delay; in epoch-wise double descent, train loss decreases monotonically while test loss or error rises and falls. Existing accounts are often task-specific, and a task-agnostic analysis framework for diagnosing and explaining these phenomena across realistic tasks and architectures is missing. We address this challenge by analyzing two competing processes that underlie learning dynamics: representation learning in the encoder and readout calibration in the final classifier. Using tools from representational geometry, neural tangent kernels, and linear probing, we show that both processes are active throughout training, with the fluctuations of their relative speed giving rise to seemingly anomalous generalization dynamics. Applying the representation-readout decomposition to grokking across a wide range of tasks and architectures, we find that the readout is train-biased before grokking onset, and representation learning is gradual but not absent, contrary to the lazy-to-rich account. The framework further provides diagnostic signatures distinguishing spurious from genuine generalization: in a previously reported MNIST grokking example and an epoch-wise double descent example, apparent delayed or non-monotone generalization is shown to arise from representation degradation and readout misalignment induced by non-standard training recipes. Together, these results establish the representation-readout decomposition as a top-down framework for understanding learning dynamics and revealing underlying algorithms for interpretability research. △ Less

Submitted 28 May, 2026; v1 submitted 26 May, 2026; originally announced May 2026.

Raon-Speech Technical Report

Authors: Beomsoo Kim, Changho Choi, Dohyun Kim, Dongki Lee, Ethan Ewer, Eunchong Kim, Gyeongman Kim, Haechan Kim, Hyeonghwan Kim, Inkyu Park, Jihun Yun, Jihwan Moon, Jiyun Kim, Joonghyun Bae, Junhyuck Kim, Minkyu Kim, Sehun Lee, Seungjun Chung, Sungwoo Cho, Dongmin Park, Dongwon Kim, Hara Kang, Jonghyun Lee, Keon Lee, Kangwook Lee , et al. (1 additional authors not shown)

Abstract: We present Raon-Speech, a top-performing 9B-parameter speech language model (SpeechLM) for English and Korean speech understanding, answering, and generation, and Raon-SpeechChat, a high-performing full-duplex extension for natural real-time conversation. Raon-Speech successfully transforms a pre-trained LLM into a SpeechLM that both understands and generates speech while preserving strong text ca… ▽ More We present Raon-Speech, a top-performing 9B-parameter speech language model (SpeechLM) for English and Korean speech understanding, answering, and generation, and Raon-SpeechChat, a high-performing full-duplex extension for natural real-time conversation. Raon-Speech successfully transforms a pre-trained LLM into a SpeechLM that both understands and generates speech while preserving strong text capabilities. It trains on 1.38M hours of highly curated English and Korean speech and text datasets with the following training stages: (1) speech modules alignment, (2) end-to-end SpeechLM pre-training with knowledge distillation, and (3) multi-task preference optimization-based post-training. Across 42 English and Korean speech and text benchmarks, Raon-Speech establishes the strongest overall profile on speech-centric tasks in our comparison against eight similarly sized recent audio foundation models, including Qwen2.5-Omni and Fun-Audio-Chat, while preserving strong text question answering performance. Building upon it, Raon-SpeechChat enables natural full-duplex conversation by continual training on 119K hours of time-aligned real and synthetic dialogue data. It proceeds through three complementary training stages: (1) causal encoder adaptation, (2) full-duplex pre-training, (3) full-duplex fine-tuning for voice and role-control. On multiple full-duplex benchmarks, Raon-SpeechChat shows its clearest strengths on the turn-taking and interruption-sensitive behaviors covered by FDB v1.0, and remains competitive across the broader full-duplex evaluation suite. We open-source all model checkpoints, the training and inference pipeline, and an interactive demo. △ Less

Submitted 8 April, 2026; originally announced May 2026.

AtelierEval: Agentic Evaluation of Humans & LLMs as Text-to-Image Prompters

Authors: Hanjun Luo, Zhimu Huang, Sylvia Chung, Yiran Wang, Yingbin Jin, Jialin Li, Jiang Li, Xinfeng Li, Hanan Salam

Abstract: Text-to-image (T2I) systems increasingly rely on upstream prompters, either humans or multimodal large language models (MLLMs), to translate user intent into detailed prompts. Yet current benchmarks fix the prompt and only evaluate T2I models, leaving the prompting proficiency of this upstream component entirely unmeasured. We introduce AtelierEval, the first unified benchmark that quantifies prom… ▽ More Text-to-image (T2I) systems increasingly rely on upstream prompters, either humans or multimodal large language models (MLLMs), to translate user intent into detailed prompts. Yet current benchmarks fix the prompt and only evaluate T2I models, leaving the prompting proficiency of this upstream component entirely unmeasured. We introduce AtelierEval, the first unified benchmark that quantifies prompting proficiency across 360 expert-crafted tasks. Grounded in a cognitive view, it spans three task categories and instantiates tasks using a taxonomy of real-world challenges, with a dual interface for both humans and MLLMs. To enable scalable and reliable evaluation, we propose AtelierJudge, a skill-based, memory-augmented agentic evaluator. It produces subjective and objective scores for prompt-image pairs, achieving a Spearman correlation of 0.79 with human experts, approaching human performance. Extensive experiments benchmark 8 MLLMs against 48 human users across 4 T2I backends, validate AtelierEval as a robust diagnostic tool, and reveal the superiority of mimicry over planning, advocating for an image-augmented direction for future prompters. Our work is released to support future research. △ Less

Submitted 21 May, 2026; originally announced May 2026.

Comments: Accepted by ICML 2026

CAdam: Context-Adaptive Moment Estimation for 3D Gaussian Densification in Generative Distillation

Authors: SeungJeh Chung, Geonho Park, Misong Kim, HyeongYeop Kang

Abstract: Adaptive densification is the engine of 3D Gaussian Splatting (3DGS). However, when transposed to the optimization-based Generative Distillation paradigm, this reconstruction-native mechanism reveals fundamental limitations, resulting in inefficient representations cluttered with redundant primitives. We diagnose this failure as a Densification Dilemma stemming from the stochastic nature of genera… ▽ More Adaptive densification is the engine of 3D Gaussian Splatting (3DGS). However, when transposed to the optimization-based Generative Distillation paradigm, this reconstruction-native mechanism reveals fundamental limitations, resulting in inefficient representations cluttered with redundant primitives. We diagnose this failure as a Densification Dilemma stemming from the stochastic nature of generative guidance: the standard magnitude-based accumulation indiscriminately aggregates transient noise alongside geometric signals, making it difficult to strike a balance between over-densification and under-fitting. To resolve this, we introduce Context-Adaptive Moment Estimation (CAdam), a novel framework that reinterprets densification as a statistically grounded signal verification problem. CAdam leverages the first moment of gradients to exploit the interference principle, where stochastic fluctuations cancel out via destructive interference while consistent geometric drifts accumulate via constructive interference, effectively disentangling the underlying signal from the generative noise floor. This is further augmented by a quantile-based context awareness and an intrinsic Signal-to-Noise Ratio (SNR) gating mechanism, which ensure robust adaptation across optimization stages and enable the soft termination of densification. Extensive experiments across diverse objectives (SDS, ISM, VFDS) and strong generative 3DGS backbones show that CAdam reduces Gaussian count by 85%-97% relative to standard densification while preserving overall comparable perceptual quality. These results highlight signal-aware density control as a practical way to improve memory efficiency in optimization-based generative distillation. △ Less

Submitted 20 May, 2026; originally announced May 2026.

Comments: Accepted to SIGGRAPH 2026 Conference Papers. 12 pages, 8 figures

SpeakerLLM: A Speaker-Specialized Audio-LLM for Speaker Understanding and Verification Reasoning

Authors: KiHyun Nam, Jungwoo Heo, Siu Bae, Ha-Jin Yu, Joon Son Chung

Abstract: As audio-first agents become increasingly common in physical AI, conversational robots, and screenless wearables, audio large language models (audio-LLMs) must integrate speaker-specific understanding to support user authorization, personalization, and context-aware interaction. This requires modeling who is speaking, how the voice sounds, and how recording conditions affect speaker cues. Conventi… ▽ More As audio-first agents become increasingly common in physical AI, conversational robots, and screenless wearables, audio large language models (audio-LLMs) must integrate speaker-specific understanding to support user authorization, personalization, and context-aware interaction. This requires modeling who is speaking, how the voice sounds, and how recording conditions affect speaker cues. Conventional speaker verification systems provide strong scalar scores but little linguistic evidence, while current audio-LLMs and speaker-aware language models have limited ability to organize speaker information beyond binary labels or descriptive profiles. We present SpeakerLLM, a speaker-specialized audio-LLM framework that unifies single-utterance speaker profiling, recording-condition understanding, utterance-pair speaker comparison, and evidence-organized verification reasoning within a natural-language interface. We construct verification-reasoning targets and a decision-composition policy that separate profile-level evidence from the final same-or-different decision and organize recording condition, profile evidence, and the decision into a structured trace. At its core, SpeakerLLM uses a hierarchical speaker tokenizer designed to capture multiple granularities of speaker evidence. Utterance-level speaker embeddings summarize identity and profile-level cues, whereas frame-level speaker features preserve fine-grained acoustic descriptors. Experiments show that SpeakerLLM-Base improves speaker-profile and recording-condition understanding over general audio-LLMs, while SpeakerLLM-VR preserves strong generated-verdict accuracy and produces decision traces grounded in the supervised verification reasoning schema. We will release the metadata-enriched supervision dataset and target-construction code for reproducibility. △ Less

Submitted 14 May, 2026; originally announced May 2026.

FiTS: Interpretable Spiking Neurons via Frequency Selectivity and Temporal Shaping

Authors: Jongmin Choi, Joon Son Chung

Abstract: Spiking Neural Networks (SNNs) are a promising framework for event-driven temporal processing. Prior work has improved temporal modeling through richer neuron dynamics and network-level mechanisms such as recurrence and delays, but it remains unclear how individual spiking neurons should specialize within a network. In this work, we introduce FiTS, a spiking neuron that factorizes temporal computa… ▽ More Spiking Neural Networks (SNNs) are a promising framework for event-driven temporal processing. Prior work has improved temporal modeling through richer neuron dynamics and network-level mechanisms such as recurrence and delays, but it remains unclear how individual spiking neurons should specialize within a network. In this work, we introduce FiTS, a spiking neuron that factorizes temporal computation within each neuron into Frequency Selectivity (FS) and Temporal Shaping (TS). The FS module parameterizes each neuron's target frequency as the maximizer of its subthreshold magnitude response, while the TS module reshapes when frequency components contribute to membrane voltage accumulation through group-delay modulation. On auditory benchmarks where frequency selectivity and timing are central to the input structure, FiTS consistently improves over a plain Leaky Integrate-and-Fire (LIF) baseline in simple feedforward SNNs without recurrence or network-level delays, while remaining competitive with strong temporal SNN baselines. Beyond accuracy, the learned target frequencies and group-delay shifts provide interpretable neuron-level summaries of the frequency and timing organization learned within the network. △ Less

Submitted 13 May, 2026; originally announced May 2026.

Comments: 23 pages, 7 figures

Keep What Audio Cannot Say: Context-Preserving Token Pruning for Omni-LLMs

Authors: Chaeyoung Jung, Kyeongha Rho, Joon Son Chung

Abstract: Omnimodal Large Language Models (Omni-LLMs) incur substantial computational overhead due to the large number of multimodal input tokens they process, making token reduction essential for real-world deployment. Existing Omni-LLM pruning methods typically reduce this cost by selecting tokens that are important for the current query or strongly aligned with cross-modal cues. However, such strategies… ▽ More Omnimodal Large Language Models (Omni-LLMs) incur substantial computational overhead due to the large number of multimodal input tokens they process, making token reduction essential for real-world deployment. Existing Omni-LLM pruning methods typically reduce this cost by selecting tokens that are important for the current query or strongly aligned with cross-modal cues. However, such strategies can discard evidence that falls outside these criteria, even when needed for different questions or for understanding context beyond aligned audio-visual cues. To address this limitation, we reframe Omni-LLM token reduction as preserving broad audio-visual context while removing cross-modal redundancy. We propose ContextGuard, an inference-time token pruning framework built on this principle. ContextGuard predicts coarse visual semantics from audio and prunes video tokens whose coarse semantics are likely recoverable from audio, while retaining additional video tokens to preserve localized visual details that audio alone cannot specify. For further compression, our method merges temporally similar video tokens. The framework requires no downstream LLM fine-tuning and uses only an independently trained lightweight predictor. On Qwen2.5-Omni and Video-SALMONN2+ at 3B and 7B scales across six audio-visual benchmarks, ContextGuard outperforms prior inference-time pruning methods while pruning more tokens. Notably, on Qwen2.5-Omni 7B, ContextGuard achieves full-token-level performance on five of six benchmarks while pruning 55% of input tokens. △ Less

Submitted 12 May, 2026; originally announced May 2026.

Probing Cross-modal Information Hubs in Audio-Visual LLMs

Authors: Jihoo Jung, Chaeyoung Jung, Ji-Hoon Kim, Joon Son Chung

Abstract: Audio-visual large language models (AVLLMs) have recently emerged as a powerful architecture capable of jointly reasoning over audio, visual, and textual modalities. In AVLLMs, the bidirectional interaction between audio and video modalities introduces intricate processing dynamics, necessitating a deeper understanding of their internal mechanisms. However, unlike extensively studied text-only or… ▽ More Audio-visual large language models (AVLLMs) have recently emerged as a powerful architecture capable of jointly reasoning over audio, visual, and textual modalities. In AVLLMs, the bidirectional interaction between audio and video modalities introduces intricate processing dynamics, necessitating a deeper understanding of their internal mechanisms. However, unlike extensively studied text-only or large vision language models, the internal workings of AVLLMs remain largely unexplored. In this paper, we focus on cross-modal information flow between audio and visual modalities in AVLLMs, investigating where information derived from one modality is encoded within the token representations of the other modality. Through an analysis of multiple recent AVLLMs, we uncover two common findings. First, AVLLMs primarily encode integrated audio-visual information in sink tokens. Second, sink tokens do not uniformly hold cross-modal information. Instead, a distinct subset of sink tokens, which we term cross-modal sink tokens, specializes in storing such information. Based on these findings, we further propose a simple training-free hallucination mitigation method by encouraging reliance on integrated cross-modal information within cross-modal sink tokens. Our code is available at https://github.com/kaistmm/crossmodal-hub. △ Less

Submitted 11 May, 2026; v1 submitted 11 May, 2026; originally announced May 2026.

Comments: Accepted by ICML 2026

LRS-VoxMM: A benchmark for in-the-wild audio-visual speech recognition

Authors: Doyeop Kwak, Jeongsoo Choi, Suyeon Lee, Joon Son Chung

Abstract: We introduce LRS-VoxMM, an in-the-wild benchmark for audio-visual speech recognition (AVSR). The benchmark is derived from VoxMM, a dataset of diverse real-world spoken conversations with human-annotated transcriptions. We select AVSR-suitable samples and preprocess them in an LRS-style format for direct use in existing AVSR pipelines. Compared with commonly used benchmarks, LRS-VoxMM covers a mor… ▽ More We introduce LRS-VoxMM, an in-the-wild benchmark for audio-visual speech recognition (AVSR). The benchmark is derived from VoxMM, a dataset of diverse real-world spoken conversations with human-annotated transcriptions. We select AVSR-suitable samples and preprocess them in an LRS-style format for direct use in existing AVSR pipelines. Compared with commonly used benchmarks, LRS-VoxMM covers a more diverse range of scenarios and acoustic conditions. We also release distorted evaluation sets with additive noise, reverberation, and bandwidth limitation to support evaluation under severe acoustic degradation. Experimental results show that LRS-VoxMM is considerably harder than LRS3 and that the contribution of visual information becomes more evident as the audio signal degrades. LRS-VoxMM supports more realistic AVSR benchmarking and encourages further research on the role of visual information in challenging real-world conditions. △ Less

Submitted 30 April, 2026; originally announced April 2026.

Comments: Technical report for the LRS-VoxMM dataset release. Project page: https://mm.kaist.ac.kr/projects/voxmm

ICPR 2026 Competition on Low-Resolution License Plate Recognition

Authors: Rayson Laroca, Valfride Nascimento, Donggun Kim, Sanghyeok Chung, Subin Bae, Uihwan Seo, Seungsang Oh, Chi M. Phung, Minh G. Vo, Xingsong Ye, Yongkun Du, Yuchen Su, Zhineng Chen, Sunhee Heo, Hyangwoo Lee, Kihyun Na, Khanh V. Vu Nguyen, Sang T. Pham, Duc N. N. Phung, Trong P. Le, Vy N. Vo Tran, David Menotti

Abstract: Low-Resolution License Plate Recognition (LRLPR) remains a challenging problem in real-world surveillance scenarios, where long capture distances, compression artifacts, and adverse imaging conditions can severely degrade license plate legibility. To promote progress in this area, we organized the ICPR 2026 Competition on Low-Resolution License Plate Recognition, the first competition specifically… ▽ More Low-Resolution License Plate Recognition (LRLPR) remains a challenging problem in real-world surveillance scenarios, where long capture distances, compression artifacts, and adverse imaging conditions can severely degrade license plate legibility. To promote progress in this area, we organized the ICPR 2026 Competition on Low-Resolution License Plate Recognition, the first competition specifically dedicated to LRLPR using real low-quality data collected under operationally relevant conditions. The competition was based on the LRLPR-26 dataset, which comprises 20,000 training tracks and 3,000 test tracks; each training track contains five low-resolution and five high-resolution images of the same license plate. Notably, a total of 269 teams from 41 countries registered for the competition, and 99 teams submitted valid entries in the Blind Test Phase. The winning team achieved a Recognition Rate of 82.13%, and four teams surpassed the 80% mark, highlighting both the high level of competition at the top of the leaderboard and the continued difficulty of the task. In addition to presenting the competition design, evaluation protocol, and main results, this paper summarizes the methods adopted by the top-5 teams and discusses current trends and promising directions for future research on LRLPR. The competition webpage is available at https://icpr26lrlpr.github.io/ △ Less

Submitted 24 April, 2026; originally announced April 2026.

Comments: Accepted for presentation at the International Conference on Pattern Recognition (ICPR) 2026

Planetary Exploration 3.0: A Roadmap for Software-Defined, Radically Adaptive Space Systems

Authors: Masahiro Ono, Daniel Selva, Morgan L. Cable, Marie Ethvignot, Margaret Hansen, Andreas M. Hein, Elena-Sorina Lupu, Zachary Manchester, David Murrow, Chad Pozarycki, Pascal Spino, Amanda Stockton, Mathieu Choukroun, Soon-Jo Chung, John Day, Alexander Demagall, Anthony Freeman, Chloe Gentgen, Michel D. Ingham, Charity M. Phillips-Lander, Richard Rieber, Alejandro Salado, Maria Sakovsky, Lori R. Shiraishi, Yisong Yue , et al. (1 additional authors not shown)

Abstract: The surface and subsurface of worlds beyond Mars remain largely unexplored. Yet these worlds hold keys to fundamental questions in planetary science - from potentially habitable subsurface oceans on icy moons to ancient records preserved in Kuiper Belt objects. NASA's success in Mars exploration was achieved through incrementalism: 22 progressively sophisticated missions over decades. This paradig… ▽ More The surface and subsurface of worlds beyond Mars remain largely unexplored. Yet these worlds hold keys to fundamental questions in planetary science - from potentially habitable subsurface oceans on icy moons to ancient records preserved in Kuiper Belt objects. NASA's success in Mars exploration was achieved through incrementalism: 22 progressively sophisticated missions over decades. This paradigm, which we call Planetary Exploration 2.0 (PE 2.0), is untenable for the outer Solar System, where cruise times of a decade or more make iterative missions infeasible. We propose Planetary Exploration 3.0 (PE 3.0): a paradigm in which unvisited worlds are explored by a single or a few missions with radically adaptive space systems. A PE 3.0 mission conducts both initial exploratory science and follow-on hypothesis-driven science based on its own in situ data returns, evolving spacecraft capabilities to work resiliently in previously unseen environments. The key enabler of PE 3.0 is software-defined space systems (SDSSs) - systems that can adapt their functions at all levels through software updates. This paper presents findings from a Keck Institute for Space Studies (KISS) workshop on PE 3.0, covering: (1) PE 3.0 systems engineering including science definition, architecture, design methods, and verification & validation; (2) software-defined space system technologies including reconfigurable hardware, multi-functionality, and modularity; (3) onboard intelligence including autonomous science, navigation, controls, and embodied AI; and (4) three PE 3.0 mission concepts: a Neptune/Triton smart flyby, an ocean world explorer, and an Oort cloud reconnaissance mission. △ Less

Submitted 21 April, 2026; originally announced April 2026.

Journal ref: AIAA ASCEND 2026

Evolution of Lane-Changing Behavior in Mixed Traffic: A Quantum Game Theory Approach

Authors: Sungyong Chung, Tina Radvand, Alireza Talebpour

Abstract: As automated vehicles (AVs) enter mixed traffic, proactively anticipating the evolution of human driving behavior during critical interactions, such as lane changes, is essential. However, classical Evolutionary Game Theory (EGT) fails to capture the complexity of human decision-making during lane changes. Specifically, by strictly assuming independence between agents, classical models calibrated… ▽ More As automated vehicles (AVs) enter mixed traffic, proactively anticipating the evolution of human driving behavior during critical interactions, such as lane changes, is essential. However, classical Evolutionary Game Theory (EGT) fails to capture the complexity of human decision-making during lane changes. Specifically, by strictly assuming independence between agents, classical models calibrated on empirical payoffs predict a convergence to unrealistic full cooperation, contradicting the stable 42% cooperation rate observed in real-world data. To resolve this discrepancy, this study introduces a Quantum Game Theory (QGT) framework. We analyze 7,636 lane-changing interactions from the Waymo Open Motion Dataset (WOMD) to derive empirical payoff matrices via a Quantal Response Equilibrium (QRE) model. Utilizing the Marinatto-Weber (MW) quantization scheme, we introduce an entanglement parameter to mathematically embed latent correlations directly into the payoff structure of a single interaction. Our results identify a human entanglement parameter of $|b|^2_{HDV} \approx 0.52$ that accurately reproduces the observed mixed equilibrium. Furthermore, simulations of three AV deployment strategies (classical, entangled, and inverted) reveal that human adaptation depends critically on the underlying AV algorithm: while cooperative classical AVs maximize system-wide cooperation at high market penetration rates, defective inverted AVs paradoxically yield higher overall cooperation at low penetration rates by prompting more cooperative behaviors from human drivers. Consequently, rather than waiting for large scale deployment to observe these effects, stakeholders can utilize this framework to simulate repeated interactions and proactively anticipate how human driver behavior will evolve in response to specific AV software designs. △ Less

Submitted 16 April, 2026; originally announced April 2026.

Low-Stack HAETAE for Memory-Constrained Microcontrollers

Authors: Gustavo Banegas, Kim Youngbeom, Seo Seog Chung, Vredendaal Christine Van

Abstract: We present a low-stack implementation of the module-lattice signature scheme HAETAE, targeting microcontrollers with 8 kB-16 kB of available SRAM. On such devices, peak stack usage is often the binding constraint, and HAETAE's hyperball-based sampler, large transient polynomial vectors, and variable-length signature payloads (hint and high-bits arrays) pose a particular challenge. To address this… ▽ More We present a low-stack implementation of the module-lattice signature scheme HAETAE, targeting microcontrollers with 8 kB-16 kB of available SRAM. On such devices, peak stack usage is often the binding constraint, and HAETAE's hyperball-based sampler, large transient polynomial vectors, and variable-length signature payloads (hint and high-bits arrays) pose a particular challenge. To address this we introduce (i) Rejection-aware pass decomposition, which isolates encoding to the post-acceptance path; (ii) Component-level early rejection, which short-circuits the response computation when a partial norm already exceeds the bound; and (iii) Reverse-order streaming entropy coding using range Asymmetric Numeral Systems (rANS), which eliminates full hint and high-bits staging buffers. Combined with streamed matrix generation, a two-pass hyperball sampler with streaming Gaussian backend, and row-streamed verification, these techniques bring Signing stack from 71 kB-141 kB in the reference implementation down to 5.8 kB-6.0 kB, key generation to 4.7 kB-5.7 kB, and verification to 4.7 kB-4.8 kB across all three security levels. Our pure C implementation covers all three security levels (HAETAE-2/3/5), whose optimization paths differ due to the public-key domain (d>0 vs. d=0) and rejection structure. We implement our optimization on a Nucleo-L4R5ZI and compare to the reference pqm4 (for HAETAE-2 and -3) and a recently published memory-optimized implementation (targeting HAETAE-5 only). We reduce HAETAE-2, -3, and -5 stack by respectively 75, 86 and 8 % for key generation, 92, 95 and 24 % for signature generation, and 85, 91 and 22 % for verification. Depending on the parameter set, this impacts performance by at most a factor 1.8 and 3.4 for key and signature generation respectively, while even offering a performance improvement up to 18 % for verification. Verification at all security levels fits within 8 kB of RAM (signature buffer + stack) and is 2.34-3.34x faster than ML-DSA m4fstack at each comparable security level. We additionally validate portability under RIOT-OS on ARM Cortex-M4 and RISC-V targets. △ Less

Submitted 17 April, 2026; originally announced April 2026.

Seeing Through Touch: Tactile-Driven Visual Localization of Material Regions

Authors: Seongyu Kim, Seungwoo Lee, Hyeonggon Ryu, Joon Son Chung, Arda Senocak

Abstract: We address the problem of tactile localization, where the goal is to identify image regions that share the same material properties as a tactile input. Existing visuo-tactile methods rely on global alignment and thus fail to capture the fine-grained local correspondences required for this task. The challenge is amplified by existing datasets, which predominantly contain close-up, low-diversity ima… ▽ More We address the problem of tactile localization, where the goal is to identify image regions that share the same material properties as a tactile input. Existing visuo-tactile methods rely on global alignment and thus fail to capture the fine-grained local correspondences required for this task. The challenge is amplified by existing datasets, which predominantly contain close-up, low-diversity images. We propose a model that learns local visuo-tactile alignment via dense cross-modal feature interactions, producing tactile saliency maps for touch-conditioned material segmentation. To overcome dataset constraints, we introduce: (i) in-the-wild multi-material scene images that expand visual diversity, and (ii) a material-diversity pairing strategy that aligns each tactile sample with visually varied yet tactilely consistent images, improving contextual localization and robustness to weak signals. We also construct two new tactile-grounded material segmentation datasets for quantitative evaluation. Experiments on both new and existing benchmarks show that our approach substantially outperforms prior visuo-tactile methods in tactile localization. △ Less

Submitted 13 April, 2026; originally announced April 2026.

Comments: CVPR 2026. Project page: https://mm.kaist.ac.kr/projects/SeeingThroughTouch/

Data is All You Need: Markov Chain Car-Following (MC-CF) Model

Authors: Sungyong Chung, Yanlin Zhang, Nachuan Li, Dana Monzer, Alireza Talebpour

Abstract: Car-following behavior is fundamental to traffic flow theory, yet traditional models often fail to capture the stochasticity of naturalistic driving. This paper introduces a new car-following modeling category called the empirical probabilistic paradigm, which bypasses conventional parametric assumptions. Within this paradigm, we propose the Markov Chain Car-Following (MC-CF) model, which represen… ▽ More Car-following behavior is fundamental to traffic flow theory, yet traditional models often fail to capture the stochasticity of naturalistic driving. This paper introduces a new car-following modeling category called the empirical probabilistic paradigm, which bypasses conventional parametric assumptions. Within this paradigm, we propose the Markov Chain Car-Following (MC-CF) model, which represents state transitions as a Markov process and predicts behavior by randomly sampling accelerations from empirical distributions within discretized state bins. Evaluation of the MC-CF model trained on the Waymo Open Motion Dataset (WOMD) demonstrates that its variants significantly outperform physics-based models including IDM, Gipps, FVDM, and SIDM in both one-step and open-loop trajectory prediction accuracy. Statistical analysis of transition probabilities confirms that the model-generated trajectories are indistinguishable from real-world behavior, successfully reproducing the probabilistic structure of naturalistic driving across all interaction types. Zero-shot generalization on the Naturalistic Phoenix (PHX) dataset further confirms the model's robustness. Finally, microscopic ring road simulations validate the framework's scalability. By incrementally integrating unconstrained free-flow trajectories and high-speed freeway data (TGSIM) alongside a conservative inference strategy, the model drastically reduces collisions, achieving zero crashes in multiple equilibrium and shockwave scenarios, while successfully reproducing naturalistic and stochastic shockwave propagation. Overall, the proposed MC-CF model provides a robust, scalable, and calibration-free foundation for high-fidelity stochastic traffic modeling, uniquely suited for the data-rich future of intelligent transportation. △ Less

Submitted 29 March, 2026; originally announced March 2026.

Cinematic Audio Source Separation Using Visual Cues

Authors: Kang Zhang, Suyeon Lee, Arda Senocak, Joon Son Chung

Abstract: Cinematic Audio Source Separation (CASS) aims to decompose mixed film audio into speech, music, and sound effects, enabling applications like dubbing and remastering. Existing CASS approaches are audio-only, overlooking the inherent audio-visual nature of films, where sounds often align with visual cues. We present the first framework for audio-visual CASS (AV-CASS), leveraging visual context to e… ▽ More Cinematic Audio Source Separation (CASS) aims to decompose mixed film audio into speech, music, and sound effects, enabling applications like dubbing and remastering. Existing CASS approaches are audio-only, overlooking the inherent audio-visual nature of films, where sounds often align with visual cues. We present the first framework for audio-visual CASS (AV-CASS), leveraging visual context to enhance separation quality. Our method formulates CASS as a conditional generative modeling problem using conditional flow matching, enabling multimodal audio source separation. To address the lack of cinematic datasets with isolated sound tracks, we introduce a training data synthesis pipeline that pairs in-the-wild audio and video streams (e.g., facial videos for speech, scene videos for effects) and design a dedicated visual encoder for this dual-stream setup. Trained entirely on synthetic data, our model generalizes effectively to real-world cinematic content and achieves strong performance on synthetic, real-world, and audio-only CASS benchmarks. Code and demo are available at \url{https://cass-flowmatching.github.io}. △ Less

Submitted 27 March, 2026; originally announced March 2026.

Comments: CVPR 2026. Project page: https://cass-flowmatching.github.io

Plug-and-Steer: Decoupling Separation and Selection in Audio-Visual Target Speaker Extraction

Authors: Doyeop Kwak, Suyeon Lee, Joon Son Chung

Abstract: The goal of this paper is to provide a new perspective on audio-visual target speaker extraction (AV-TSE) by decoupling separation and target selection. Conventional AV-TSE systems typically integrate audio and visual features deeply to re-learn the entire separation process, which can act as a fidelity ceiling due to the noisy nature of in-the-wild audio-visual datasets. To address this, we propo… ▽ More The goal of this paper is to provide a new perspective on audio-visual target speaker extraction (AV-TSE) by decoupling separation and target selection. Conventional AV-TSE systems typically integrate audio and visual features deeply to re-learn the entire separation process, which can act as a fidelity ceiling due to the noisy nature of in-the-wild audio-visual datasets. To address this, we propose Plug-and-Steer, which assigns high-fidelity separation to a frozen audio-only backbone and limits the role of the visual modality strictly to target selection. We introduce the Latent Steering Matrix (LSM), a minimalist linear transformation that re-routes latent features within the backbone to anchor the target speaker to a designated channel. Experiments across four representative architectures show that our method effectively preserves the acoustic priors of diverse backbones, achieving perceptual quality comparable to that of the original backbones. Audio samples are available at: https://plugandsteer.github.io △ Less

Submitted 16 June, 2026; v1 submitted 20 March, 2026; originally announced March 2026.

Comments: Accepted by Interspeech 2026; demo available https://plugandsteer.github.io

ContractionPPO: Certified Reinforcement Learning via Differentiable Contraction Layers

Authors: Vrushabh Zinage, Narek Harutyunyan, Eric Verheyden, Fred Y. Hadaegh, Soon-Jo Chung

Abstract: Legged locomotion in unstructured environments demands not only high-performance control policies but also formal guarantees to ensure robustness under perturbations. Control methods often require carefully designed reference trajectories, which are challenging to construct in high-dimensional, contact-rich systems such as quadruped robots. In contrast, Reinforcement Learning (RL) directly learns… ▽ More Legged locomotion in unstructured environments demands not only high-performance control policies but also formal guarantees to ensure robustness under perturbations. Control methods often require carefully designed reference trajectories, which are challenging to construct in high-dimensional, contact-rich systems such as quadruped robots. In contrast, Reinforcement Learning (RL) directly learns policies that implicitly generate motion, and uniquely benefits from access to privileged information, such as full state and dynamics during training, that is not available at deployment. We present ContractionPPO, a framework for certified robust planning and control of legged robots by augmenting Proximal Policy Optimization (PPO) RL with a state-dependent contraction metric layer. This approach enables the policy to maximize performance while simultaneously producing a contraction metric that certifies incremental exponential stability of the simulated closed-loop system. The metric is parameterized as a Lipschitz neural network and trained jointly with the policy, either in parallel or as an auxiliary head of the PPO backbone. While the contraction metric is not deployed during real-world execution, we derive upper bounds on the worst-case contraction rate and show that these bounds ensure the learned contraction metric generalizes from simulation to real-world deployment. Our hardware experiments on quadruped locomotion demonstrate that ContractionPPO enables robust, certifiably stable control even under strong external perturbations. △ Less

Submitted 20 March, 2026; originally announced March 2026.

Comments: Accepted to RA-L journal

Localizing and Editing Knowledge in Large Audio-Language Models

Authors: Sung Kyun Chung, Jiaheng Dong, Qiuchi Hu, Gongping Huang, Hong Jia, Ting Dang

Abstract: Large Audio-Language Models (LALMs) have shown strong performance in speech understanding, making speech a natural interface for accessing factual information. Yet they are trained on static corpora and may encode incorrect facts. Existing model editing methods localize and update facts in text-only LLMs, but do not account for continuous speech representations, or where knowledge is stored across… ▽ More Large Audio-Language Models (LALMs) have shown strong performance in speech understanding, making speech a natural interface for accessing factual information. Yet they are trained on static corpora and may encode incorrect facts. Existing model editing methods localize and update facts in text-only LLMs, but do not account for continuous speech representations, or where knowledge is stored across acoustic or language modules, or their cross-modal module. We construct the first audio benchmark for knowledge localization and editing in LALMs and propose a speech-driven locate-then-edit framework. First, we use speech-aware causal tracing to localize layers and modules that support factual retrieval and then apply editing at identified sites. Experiments show that factual knowledge is jointly encoded in audio and text modules, and that audio editing yields more effective updates than text editing or fine-tuning, enabling fine-grained knowledge control in speech AI systems. △ Less

Submitted 19 June, 2026; v1 submitted 15 March, 2026; originally announced March 2026.

Comments: Paper was accepted by Interspeech 2026

On the Nature of Attention Sink that Shapes Decoding Strategy in Omni-LLMs

Authors: Suho Yoo, Youngjoon Jang, Joon Son Chung

Abstract: The goal of this paper is to strengthen the reasoning of Omnimodal Large Language Models (Omni-LLMs) at inference time, without additional training. These models jointly process video, audio, and text, and given the large number of tokens they consume, how attention is routed across them is central to their behaviour. We focus specifically on attention sinks, tokens that absorb a disproportionate… ▽ More The goal of this paper is to strengthen the reasoning of Omnimodal Large Language Models (Omni-LLMs) at inference time, without additional training. These models jointly process video, audio, and text, and given the large number of tokens they consume, how attention is routed across them is central to their behaviour. We focus specifically on attention sinks, tokens that absorb a disproportionate share of attention mass regardless of their semantic content, to understand how this routing unfolds. To this end, we conduct a systematic analysis of sink behaviour in Omni-LLMs. Our analysis yields two key findings: (i) high sink attention does not solely indicate head redundancy, suggesting that sink value representations play additional functional roles; (ii) the sink value vector acts as a shared bias added to every token's output, serving as a global signal that organises the representation as a whole. Building on this, we propose OutRo, which correspondingly aligns non-sink token representations with the sink in feature space, and relaxes the causal mask for sink tokens at an early layer to sharpen this bias before the rest of decoding proceeds. This design enhances the reasoning process without requiring additional forward passes or access to attention maps. Based on extensive experiments, OutRo consistently improves performance on seven video QA benchmarks and demonstrates strong generalisation, while incurring only a 1.1x decoding overhead. △ Less

Submitted 7 May, 2026; v1 submitted 15 March, 2026; originally announced March 2026.

Comments: Preprint

Diagnosing Generalization Failures from Representational Geometry Markers

Authors: Chi-Ning Chou, Artem Kirsanov, Yao-Yuan Yang, SueYeon Chung

Abstract: Generalization, the ability to perform well beyond the training context, is a hallmark of biological and artificial intelligence, yet anticipating unseen failures remains a central challenge. Conventional approaches often take a ``bottom-up'' mechanistic route by reverse-engineering interpretable features or circuits to build explanatory models. While insightful, these methods often struggle to pr… ▽ More Generalization, the ability to perform well beyond the training context, is a hallmark of biological and artificial intelligence, yet anticipating unseen failures remains a central challenge. Conventional approaches often take a ``bottom-up'' mechanistic route by reverse-engineering interpretable features or circuits to build explanatory models. While insightful, these methods often struggle to provide the high-level, predictive signals for anticipating failure in real-world deployment. Here, we propose using a ``top-down'' approach to studying generalization failures inspired by medical biomarkers: identifying system-level measurements that serve as robust indicators of a model's future performance. Rather than mapping out detailed internal mechanisms, we systematically design and test network markers to probe structure, function links, identify prognostic indicators, and validate predictions in real-world settings. In image classification, we find that task-relevant geometric properties of in-distribution (ID) object manifolds consistently forecast poor out-of-distribution (OOD) generalization. In particular, reductions in two geometric measures, effective manifold dimensionality and utility, predict weaker OOD performance across diverse architectures, optimizers, and datasets. We apply this finding to transfer learning with ImageNet-pretrained models. We consistently find that the same geometric patterns predict OOD transfer performance more reliably than ID accuracy. This work demonstrates that representational geometry can expose hidden vulnerabilities, offering more robust guidance for model selection and AI interpretability. △ Less

Submitted 2 March, 2026; originally announced March 2026.

Comments: Published in the International Conference on Learning Representations (ICLR), 2026

Machine Learning on Heterogeneous, Edge, and Quantum Hardware for Particle Physics (ML-HEQUPP)

Authors: Julia Gonski, Jenni Ott, Shiva Abbaszadeh, Sagar Addepalli, Matteo Cremonesi, Jennet Dickinson, Giuseppe Di Guglielmo, Erdem Yigit Ertorer, Lindsey Gray, Ryan Herbst, Christian Herwig, Tae Min Hong, Benedikt Maier, Maryam Bayat Makou, David Miller, Mark S. Neubauer, Cristián Peña, Dylan Rankin, Seon-Hee, Seo, Giordon Stark, Alexander Tapper, Audrey Corbeil Therrien, Ioannis Xiotidis, Keisuke Yoshihara , et al. (99 additional authors not shown)

Abstract: The next generation of particle physics experiments will face a new era of challenges in data acquisition, due to unprecedented data rates and volumes along with extreme environments and operational constraints. Harnessing this data for scientific discovery demands real-time inference and decision-making, intelligent data reduction, and efficient processing architectures beyond current capabilitie… ▽ More The next generation of particle physics experiments will face a new era of challenges in data acquisition, due to unprecedented data rates and volumes along with extreme environments and operational constraints. Harnessing this data for scientific discovery demands real-time inference and decision-making, intelligent data reduction, and efficient processing architectures beyond current capabilities. Crucial to the success of this experimental paradigm are several emerging technologies, such as artificial intelligence and machine learning (AI/ML), silicon microelectronics, and the advent of quantum algorithms and processing. Their intersection includes areas of research such as low-power and low-latency devices for edge computing, heterogeneous accelerator systems, reconfigurable hardware, novel codesign and synthesis strategies, readout for cryogenic or high-radiation environments, and analog computing. This white paper presents a community-driven vision to identify and prioritize research and development opportunities in hardware-based ML systems and corresponding physics applications, contributing towards a successful transition to the new data frontier of fundamental science. △ Less

Submitted 24 July, 2026; v1 submitted 24 February, 2026; originally announced February 2026.

Comments: 123 pages, 53 figures

Environment-Aware Learning of Smooth GNSS Covariance Dynamics for Autonomous Racing

Authors: Y. Deemo Chen, Arion Zimmermann, Thomas A. Berrueta, Soon-Jo Chung

Abstract: Ensuring accurate and stable state estimation is a challenging task crucial to safety-critical domains such as high-speed autonomous racing, where measurement uncertainty must be both adaptive to the environment and temporally smooth for control. In this work, we develop a learning-based framework, LACE, capable of directly modeling the temporal dynamics of GNSS measurement covariance. We model th… ▽ More Ensuring accurate and stable state estimation is a challenging task crucial to safety-critical domains such as high-speed autonomous racing, where measurement uncertainty must be both adaptive to the environment and temporally smooth for control. In this work, we develop a learning-based framework, LACE, capable of directly modeling the temporal dynamics of GNSS measurement covariance. We model the covariance evolution as an exponentially stable dynamical system where a deep neural network (DNN) learns to predict the system's process noise from environmental features through an attention mechanism. By using contraction-based stability and systematically imposing spectral constraints, we formally provide guarantees of exponential stability and smoothness for the resulting covariance dynamics. We validate our approach on an AV-24 autonomous racecar, demonstrating improved localization performance and smoother covariance estimates in challenging, GNSS-degraded environments. Our results highlight the promise of dynamically modeling the perceived uncertainty in state estimation problems that are tightly coupled with control sensitivity. △ Less

Submitted 5 March, 2026; v1 submitted 24 February, 2026; originally announced February 2026.

Comments: 8 pages, Accepted to IEEE International Conference on Robotics and Automation (ICRA) 2026

Emergent Manifold Separability during Reasoning in Large Language Models

Authors: Chanwoo Chun, Alexandre Polo, SueYeon Chung

Abstract: Chain-of-Thought (CoT) prompting significantly improves reasoning in Large Language Models, yet the temporal dynamics of the underlying representation geometry remain poorly understood. We investigate these dynamics by applying Manifold Capacity Theory (MCT) to two compositional reasoning tasks: a controlled Boolean logic tree that supports deep mechanistic analysis, and a natural-language eligibi… ▽ More Chain-of-Thought (CoT) prompting significantly improves reasoning in Large Language Models, yet the temporal dynamics of the underlying representation geometry remain poorly understood. We investigate these dynamics by applying Manifold Capacity Theory (MCT) to two compositional reasoning tasks: a controlled Boolean logic tree that supports deep mechanistic analysis, and a natural-language eligibility task in which the model has to extract attributes from prose, compare them to thresholds, and compose the local decisions through a fixed evaluation tree. MCT lets us quantify the linear separability of latent representations without the confounding factors of probe training. On both tasks, and across several open-weight models, reasoning manifests as a transient geometric pulse: concept manifolds are untangled into linearly separable subspaces immediately prior to computation and rapidly compressed thereafter. This behavior diverges from standard linear probe accuracy, which remains high long after computation, suggesting a fundamental distinction between information that is merely retrievable and information that is geometrically prepared for processing. We interpret this phenomenon as Dynamic Manifold Management, a mechanism where the model dynamically modulates representational capacity to optimize the bandwidth of the residual stream throughout the reasoning chain. △ Less

Submitted 7 May, 2026; v1 submitted 23 February, 2026; originally announced February 2026.

Comments: Alexandre Polo and Chanwoo Chun contributed equally to this work

MAD: Modality-Adaptive Decoding for Mitigating Cross-Modal Hallucinations in Multimodal Large Language Models

Authors: Sangyun Chung, Se Yeon Kim, Youngchae Chee, Yong Man Ro

Abstract: Multimodal Large Language Models (MLLMs) suffer from cross-modal hallucinations, where one modality inappropriately influences generation about another, leading to fabricated output. This exposes a more fundamental deficiency in modality-interaction control. To address this, we propose Modality-Adaptive Decoding (MAD), a training-free method that adaptively weights modality-specific decoding branc… ▽ More Multimodal Large Language Models (MLLMs) suffer from cross-modal hallucinations, where one modality inappropriately influences generation about another, leading to fabricated output. This exposes a more fundamental deficiency in modality-interaction control. To address this, we propose Modality-Adaptive Decoding (MAD), a training-free method that adaptively weights modality-specific decoding branches based on task requirements. MAD leverages the model's inherent ability to self-assess modality relevance by querying which modalities are needed for each task. The extracted modality probabilities are then used to adaptively weight contrastive decoding branches, enabling the model to focus on relevant information while suppressing cross-modal interference. Extensive experiments on CMM and AVHBench demonstrate that MAD significantly reduces cross-modal hallucinations across multiple audio-visual language models (7.8\% and 2.0\% improvements for VideoLLaMA2-AV, 8.7\% and 4.7\% improvements for Qwen2.5-Omni). Our approach demonstrates that explicit modality awareness through self-assessment is crucial for robust multimodal reasoning, offering a principled extension to existing contrastive decoding methods. Our code is available at \href{https://github.com/top-yun/MAD}{https://github.com/top-yun/MAD} △ Less

Submitted 28 January, 2026; originally announced January 2026.

ATTNSOM: Learning Cross-Isoform Attention for Cytochrome P450 Site-of-Metabolism

Authors: Hajung Kim, Eunha Lee, Sohyun Chung, Jueon Park, Seungheun Baek, Jaewoo Kang

Abstract: Identifying metabolic sites where cytochrome P450 enzymes metabolize small-molecule drugs is essential for drug discovery. Although existing computational approaches have been proposed for site-of-metabolism prediction, they typically ignore cytochrome P450 isoform identity or model isoforms independently, thereby failing to fully capture inherent cross-isoform metabolic patterns. In addition, pri… ▽ More Identifying metabolic sites where cytochrome P450 enzymes metabolize small-molecule drugs is essential for drug discovery. Although existing computational approaches have been proposed for site-of-metabolism prediction, they typically ignore cytochrome P450 isoform identity or model isoforms independently, thereby failing to fully capture inherent cross-isoform metabolic patterns. In addition, prior evaluations often rely on top-k metrics, where false positive atoms may be included among the top predictions, underscoring the need for complementary metrics that more directly assess binary atom-level discrimination under severe class imbalance. We propose ATTNSOM, an atom-level site-of-metabolism prediction framework that integrates intrinsic molecular reactivity with cross-isoform relationships. The model combines a shared graph encoder, molecule-conditioned atom representations, and a cross-attention mechanism to capture correlated metabolic patterns across cytochrome P450 isoforms. The model is evaluated on two benchmark datasets annotated with site-of-metabolism labels at atom resolution. Across these benchmarks, the model achieves consistently strong top-k performance across multiple cytochrome P450 isoforms. Relative to ablated variants, the model yields higher Matthews correlation coefficient, indicating improved discrimination of true metabolic sites. These results support the importance of explicitly modeling cross-isoform relationships for site-of-metabolism prediction. The code and datasets are available at https://github.com/dmis-lab/ATTNSOM. △ Less

Submitted 28 January, 2026; originally announced January 2026.

Comments: 14 pages

FastAV: Efficient Token Pruning for Audio-Visual Large Language Model Inference

Authors: Chaeyoung Jung, Youngjoon Jang, Seungwoo Lee, Joon Son Chung

Abstract: In this work, we present FastAV, the first token pruning framework tailored for audio-visual large language models (AV-LLMs). While token pruning has been actively explored in standard large language models (LLMs) and vision-language models (LVLMs), its application to AV-LLMs has received little attention, even though multimodal integration substantially increases their token demands. To address t… ▽ More In this work, we present FastAV, the first token pruning framework tailored for audio-visual large language models (AV-LLMs). While token pruning has been actively explored in standard large language models (LLMs) and vision-language models (LVLMs), its application to AV-LLMs has received little attention, even though multimodal integration substantially increases their token demands. To address this gap, we introduce a pruning strategy that utilizes attention weights to identify tokens emphasized at different stages and estimates their importance. Building on this analysis, FastAV applies a two-stage pruning strategy: (1) global pruning in intermediate layers to remove broadly less influential tokens, and (2) fine pruning in later layers considering the impact on next token generation. Notably, our method does not rely on full attention maps, which makes it fully compatible with efficient attention mechanisms such as FlashAttention. Extensive experiments demonstrate that FastAV reduces FLOPs by more than 40% on two representative AV-LLMs, while preserving or even improving model performance. △ Less

Submitted 19 January, 2026; originally announced January 2026.

UNMIXX: Untangling Highly Correlated Singing Voices Mixtures

Authors: Jihoo Jung, Ji-Hoon Kim, Doyeop Kwak, Junwon Lee, Juhan Nam, Joon Son Chung

Abstract: We introduce UNMIXX, a novel framework for multiple singing voices separation (MSVS). While related to speech separation, MSVS faces unique challenges: data scarcity and the highly correlated nature of singing voices mixture. To address these issues, we propose UNMIXX with three key components: (1) musically informed mixing strategy to construct highly correlated, music-like mixtures, (2) cross-so… ▽ More We introduce UNMIXX, a novel framework for multiple singing voices separation (MSVS). While related to speech separation, MSVS faces unique challenges: data scarcity and the highly correlated nature of singing voices mixture. To address these issues, we propose UNMIXX with three key components: (1) musically informed mixing strategy to construct highly correlated, music-like mixtures, (2) cross-source attention that drives representations of two singers apart via reverse attention, and (3) magnitude penalty loss penalizing erroneously assigned interfering energy. UNMIXX not only addresses data scarcity by simulating realistic training data, but also excels at separating highly correlated mixtures through cross-source interactions at both the architectural and loss levels. Our extensive experiments demonstrate that UNMIXX greatly enhances performance, with SDRi gains exceeding 2.2 dB over prior work. △ Less

Submitted 19 January, 2026; originally announced January 2026.

Comments: Accepted by ICASSP 2026

LAMB: LLM-based Audio Captioning with Modality Gap Bridging via Cauchy-Schwarz Divergence

Authors: Hyeongkeun Lee, Jongmin Choi, KiHyun Nam, Joon Son Chung

Abstract: Automated Audio Captioning aims to describe the semantic content of input audio. Recent works have employed large language models (LLMs) as a text decoder to leverage their reasoning capabilities. However, prior approaches that project audio features into the LLM embedding space without considering cross-modal alignment fail to fully utilize these capabilities. To address this, we propose LAMB, an… ▽ More Automated Audio Captioning aims to describe the semantic content of input audio. Recent works have employed large language models (LLMs) as a text decoder to leverage their reasoning capabilities. However, prior approaches that project audio features into the LLM embedding space without considering cross-modal alignment fail to fully utilize these capabilities. To address this, we propose LAMB, an LLM-based audio captioning framework that bridges the modality gap between audio embeddings and the LLM text embedding space. LAMB incorporates a Cross-Modal Aligner that minimizes Cauchy-Schwarz divergence while maximizing mutual information, yielding tighter alignment between audio and text at both global and token levels. We further design a Two-Stream Adapter that extracts semantically enriched audio embeddings, thereby delivering richer information to the Cross-Modal Aligner. Finally, leveraging the aligned audio embeddings, a proposed Token Guide directly computes scores within the LLM text embedding space to steer the output logits of generated captions. Experimental results confirm that our framework strengthens the reasoning capabilities of the LLM decoder, achieving state-of-the-art performance on AudioCaps. △ Less

Submitted 16 March, 2026; v1 submitted 8 January, 2026; originally announced January 2026.

Comments: 5 pages, 2 figures; Accepted to ICASSP 2026

Latent Space Element Method

Authors: Seung Whan Chung, Youngsoo Choi, Christopher Miller, H. Keo Springer, Kyle T. Sullivan

Abstract: How can we build surrogate solvers that train on small domains but scale to larger ones without intrusive access to PDE operators? Inspired by the Data-Driven Finite Element Method (DD-FEM) framework for modular data-driven solvers, we propose the Latent Space Element Method (LSEM), an element-based latent surrogate assembly approach in which a learned subdomain ("element") model can be tiled and… ▽ More How can we build surrogate solvers that train on small domains but scale to larger ones without intrusive access to PDE operators? Inspired by the Data-Driven Finite Element Method (DD-FEM) framework for modular data-driven solvers, we propose the Latent Space Element Method (LSEM), an element-based latent surrogate assembly approach in which a learned subdomain ("element") model can be tiled and coupled to form a larger computational domain. Each element is a LaSDI latent ODE surrogate trained from snapshots on a local patch, and neighboring elements are coupled through learned directional interaction terms in latent space, avoiding Schwarz iterations and interface residual evaluations. A smooth window-based blending reconstructs a global field from overlapping element predictions, yielding a scalable assembled latent dynamical system. Experiments on the 1D Burgers and Korteweg-de Vries equations show that LSEM maintains predictive accuracy while scaling to spatial domains larger than those seen in training. LSEM offers an interpretable and extensible route toward foundation-model surrogate solvers built from reusable local models. △ Less

Submitted 4 January, 2026; originally announced January 2026.

Comments: 17 pages, 10 figures

Report number: LLNL-JRNL-2014444

TAVID: Text-Driven Audio-Visual Interactive Dialogue Generation

Authors: Ji-Hoon Kim, Junseok Ahn, Doyeop Kwak, Joon Son Chung, Shinji Watanabe

Abstract: The objective of this paper is to jointly synthesize interactive videos and conversational speech from text and reference images. With the ultimate goal of building human-like conversational systems, recent studies have explored talking or listening head generation as well as conversational speech generation. However, these works are typically studied in isolation, overlooking the multimodal natur… ▽ More The objective of this paper is to jointly synthesize interactive videos and conversational speech from text and reference images. With the ultimate goal of building human-like conversational systems, recent studies have explored talking or listening head generation as well as conversational speech generation. However, these works are typically studied in isolation, overlooking the multimodal nature of human conversation, which involves tightly coupled audio-visual interactions. In this paper, we introduce TAVID, a unified framework that generates both interactive faces and conversational speech in a synchronized manner. TAVID integrates face and speech generation pipelines through two cross-modal mappers (i.e., a motion mapper and a speaker mapper), which enable bidirectional exchange of complementary information between the audio and visual modalities. We evaluate our system across four dimensions: talking face realism, listening head responsiveness, dyadic interaction fluency, and speech quality. Extensive experiments demonstrate the effectiveness of our approach across all these aspects. △ Less

Submitted 23 December, 2025; originally announced December 2025.

Comments: Project page: https://mm.kaist.ac.kr/projects/TAVID

BEAT2AASIST model with layer fusion for ESDD 2026 Challenge

Authors: Sanghyeok Chung, Eujin Kim, Donggun Kim, Gaeun Heo, Jeongbin You, Nahyun Lee, Sunmook Choi, Soyul Han, Seungsang Oh, Il-Youp Kwak

Abstract: Recent advances in audio generation have increased the risk of realistic environmental sound manipulation, motivating the ESDD 2026 Challenge as the first large-scale benchmark for Environmental Sound Deepfake Detection (ESDD). We propose BEAT2AASIST which extends BEATs-AASIST by splitting BEATs-derived representations along frequency or channel dimension and processing them with dual AASIST branc… ▽ More Recent advances in audio generation have increased the risk of realistic environmental sound manipulation, motivating the ESDD 2026 Challenge as the first large-scale benchmark for Environmental Sound Deepfake Detection (ESDD). We propose BEAT2AASIST which extends BEATs-AASIST by splitting BEATs-derived representations along frequency or channel dimension and processing them with dual AASIST branches. To enrich feature representations, we incorporate top-k transformer layer fusion using concatenation, CNN-gated, and SE-gated strategies. In addition, vocoder-based data augmentation is applied to improve robustness against unseen spoofing methods. Experimental results on the official test sets demonstrate that the proposed approach achieves competitive performance across the challenge tracks. △ Less

Submitted 17 December, 2025; originally announced December 2025.

Comments: 3 pages, 1 figure, challenge paper