Searcharxiv⌕ Search

arXiv subjects

Hongjun Liu

Publications and source records attributed to Hongjun Liu.

At least 19 recordsLinked to original sources

Cleaning an Agent's History Hurts When the World Still Holds the Error

Tool-using language-model agents accumulate histories of their own mistakes, and a system that compacts, repairs or hands off such a history decides what it says about those mistakes while their effects stay in the world. Prior work cleans the history at a fixed world, where showing a corrected step and saying it is done move together. We ask whether the same correct content helps or hurts depending on how the history states it and what the world holds. We freeze an agent mid-task with its exact world state, edit the history or the world, and let a receiver model continue. On AgentCE-Bench, after the agent commits a wrong value, we hold the world fixed and add the correct value by two natural edits: a truthful note beside the preserved error or the agent's own completed action. The note raises recovery, the share of remaining targets the receiver gets right, from 51% under a matched neutral note to 66%. The completed action lowers it below the unedited history, from 50% to 44%, so the value's presence alone does not predict the effect. On AppWorld, after a database write, we vary whether its receipt reports success or failure and whether its effect is present. A success receipt lowers whole-task success from 45% to 28% while the effect is absent and raises it from 38% to 49% once present. The two effects differ by 27 points pooled over two receivers, a difference positive in six of eight receivers at 97.5% and seven at 95%. A history edit's effect thus depends on the world it describes, and a history made to look clean while the world still holds the error can leave an agent worse off than no edit at all.

cs.AI↗

MedMMV: A Controllable Multimodal Multi-Agent Framework for Reliable and Verifiable Clinical Reasoning

Current multimodal large language models (MLLMs) suffer from unstable reasoning due to the lack of mechanisms for controlling reasoning trajectories, often leading to hallucinated conclusions. Our analysis reveals that early-stage uncertainty induces trajectory switching, significantly increasing hallucination risk. We propose MedMMV, an evidence-grounded trajectory control framework that treats reasoning as a search-and-selection process over candidate trajectories. By constructing, refining, and selecting among multiple reasoning paths, MedMMV improves reliability and produces structured supervision signals over trajectory quality. We further convert these signals into learnable objectives for post-training, including reject sampling and pairwise preference learning, enabling smaller models to learn trajectory selection as an efficient single-pass capability. Instead of merely imitating reasoning traces, the model is trained to assign higher likelihood to preferred trajectories. Experiments show that richer supervision leads to stronger transfer, with accuracy improving from 10.5% to 29.0%. Our results demonstrate that trajectory selection is the key transferable capability for reliable reasoning.

cs.AI↗

Receiver Design for Continuous-Time AFDM based on Pre-Filter De-Chirp

Affine frequency division multiplexing (AFDM) is a promising waveform for future high-mobility communications due to its Doppler robustness and diversity gain. However, existing continuous-time AFDM relies on chirp-dependent wideband receive filtering, which increases noise power and causes channel-energy leakage. In this paper, we propose an AFDM reception scheme based on pre-filter de-chirp for doubly selective channels. The coupling between de-chirp and the channel is first characterized, yielding an equivalent path representation with delay-coupled effective Doppler shifts. The discrete affine Fourier transform (DAFT)-domain equivalent channel is then derived under general transmit pulse shaping and linear time-invariant receive filtering, showing that the effective filtering kernel depends only on the relative sample offset and avoids the additional spreading caused by conventional chirp-dependent filtering. The receive-noise power is further analyzed, and an explicit expression for the signal-to-noise ratio (SNR) gap is obtained by the required receive-filter bandwidths. Simulations verify the theoretical SNR analysis and show that the proposed scheme significantly outperforms the conventional scheme while approaching ideal discrete-time AFDM.

eess.SP↗

NESA: Non-stationary Evidence-aware Semantic Alignment for Robust ECG--Text Representation Pretraining

ECG-text pretraining offers transferable electrocardiogram (ECG) representations for zero-shot diagnosis, but its supervision is inherently asymmetric: clinical reports contain sparse diagnostic phrases, while ECG signals provide rich and redundant evidence across leads and time. As a result, cross-modal learning can collapse into coarse prototype matching, relying on a few highly aligned ECG tokens while overlooking complementary evidence. Existing methods enrich text supervision or perturb ECG inputs, yet rarely model how cross-modal correspondence should adapt as the available ECG evidence changes. We propose NESA, a Non-stationary Evidence-aware Semantic Alignment framework that replaces static prototype matching with an evolving alignment trajectory. NESA estimates each ECG token's contribution to the current cross-modal alignment and progressively removes dominant tokens in latent space to expose residual evidence. To preserve semantic correspondence under evidence removal, it constructs adaptive alignment states from text-derived clinical semantics, full-view ECG context, and residual ECG evidence. This trajectory guides the ECG encoder to capture complementary diagnostic cues rather than rely on a small set of tokens. Experiments with MIMIC-IV-ECG pretraining and three downstream benchmarks show improvements in zero-shot diagnosis and label-efficient transfer. Dominant-token deletion analysis further indicates that NESA preserves cross-modal alignment after highly aligned ECG evidence is removed, supporting more distributed ECG-text grounding.

cs.MM↗

Rethinking EEG Spatial Super-Resolution as Dynamic State Decomposition and Filtering

EEG spatial super-resolution aims to recover high-density neural signals from sparse electrode observations. Existing methods have achieved substantial progress through spatial modeling, generative reconstruction, and temporal representation learning. However, the current EEG state contains two different sources of information: predictable evolution inherited from preceding states and newly emerging changes revealed by the current observation. Most approaches typically do not model these two components explicitly. In this work, we formulate streaming EEG spatial super-resolution as a sequential state filtering problem and propose Dynamic State Filtering (DSF). Rather than relying on a single temporal reconstruction pathway, DSF combines a history-aware dynamical branch with an innovation-filtering branch that explicitly decomposes waveform-state estimation into prediction and observation-driven correction. To infer how spatially incomplete innovations should affect missing locations, DSF routes the observed innovations across electrodes through a mask-constrained, topology-aware transition operator and incorporates them through uncertainty-aware state correction. The two complementary estimates are finally fused to reconstruct the dense EEG signal. Experiments on SEED, SEED-IV and BCI2000 demonstrate that DSF improves reconstruction fidelity and downstream utility while maintaining robustness under dynamically changing and unreliable sparse observations.

cs.AI↗

Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict

When retrieved evidence contradicts an agent's prior beliefs, does it revise its answer, acknowledge uncertainty, or persist with an incorrect conclusion? Existing evaluations of agentic systems focus primarily on task success, offering limited insight into how agents handle such conflicts. We propose to evaluate agents on epistemic humility (EH): the agent's willingness to recognize, act on, and communicate uncertainty during task execution. We operationalize EH through three trajectory-level behavioral dimensions: Identify, Solve, and Escalate (ISE). Through knowledge conflict, situations where the backbone language model's parametric knowledge contradicts the evidence it encounters, or where two contextual sources disagree, we evaluate two conflict settings: (1) controlled conflict and (2) naturally occurring conflict during multi-step agentic execution, each paired with matched no-conflict controls. Evaluating four agents, we find that higher task accuracy does not necessarily correspond to greater epistemic humility: some high-accuracy configurations recognize conflicts during execution but do not communicate unresolved uncertainty in their incorrect final answers. Trajectory-level analysis further reveals that agents frequently detect conflicts in early steps of execution but fail to maintain or resolve them in later steps. Finally, we show that model-level interventions can improve EH, but often at the cost of task accuracy, suggesting that epistemic humility emerges from the interaction among the backbone model, agent harness, and evaluation environment.

cs.AI↗

Harnessing LLM Agents with Skill Programs

Equipping LLM agents with reusable skills derived from past experience has become a popular and successful approach for tackling complex and long-horizon tasks. However, such lessons are often encoded as textual guidance that remains largely advisory, lacking explicit mechanisms for when and how to intervene in the agent loop. To bridge the gap, we introduce HASP(Harnessing LLM Agents with Skill Programs), a new framework that upgrades skills into executable Program Functions (PFs). Rather than offering passive advice, PFs act as executable guardrails that activate on failure-prone states and modify the next action or inject corrective context. HASP is highly modular: it can be applied at inference time for direct agent-loop intervention, during post-training to provide structured supervision, or for self-improvement by evolving validated, teacher-reviewed PFs. Empirically, HASP drives substantial gains compared to both training-free and training-based methods on web-search, math reasoning, and coding tasks. For example, on web-search reasoning, inference-time PFs alone improve the average performance by 25% compared to (multi-loop) ReAct Agent, while post-training and controlled evolution achieve a 30.4% gain over Search-R1. To provide deeper insights into HASP, our mechanism analysis reveals how PFs trigger and intervene, how skills are internalized, and the requirement for stable skill library evolution.

cs.AI↗

Memory Is a Derivation: The Distributed-Evidence Paradox in Long-Term Agents

Long-running LLM agents compress past interactions into persistent memories that may be reused as premises for later tasks. This creates a distinct derivation problem: whether the memory actually follows from what the interaction history supports. Relevant evidence may be scattered across earlier interactions, while compression can introduce relations or event status that the history never established. A valid memory may therefore appear unsupported because its citations omit relevant evidence, while individually supported facts may be composed into a stronger statement the history never established. We characterize this problem through three coupled requirements: (1) Evidence scope; (2) Compositional validity; (3) Admission reliability. We therefore ask whether the interaction history available at write time supports what enters persistent memory. We introduce DerivAudit, a framework for auditing whether a memory is actually supported by the history available when it was written. The audit separates three questions: whether supporting evidence lies beyond writer-provided citations, whether the composed memory introduces unsupported meaning, and how write-time admission decisions affect later memory use. Across two natural memory corpora, audits using broader pre-write history recover support for nearly 60% of memories that appear unsupported from citations alone, while 17-21% remain unsupported after expansion. Yet broader evidence does not by itself make admission reliable: unsupported memories are still frequently admitted across verification models, and evidence expansion alone worsens it on two backbones.

cs.AI↗

BioKD: Selective Physiology-to-Video Knowledge Distillation via Reliability Gate for Emotion Recognition

To address the limitations of video-based emotion recognition under ambiguous or socially masked behavioral cues, as well as the poor deployability of physiological signals, this paper proposes a reliability-aware physiology-to-video knowledge distillation framework, termed BioKD. The proposed framework leverages physiological signals as privileged information during training to guide a video-based student model in learning deep affective representations, while relying solely on non-intrusive video inputs at inference time. To cope with the high noise and instability of physiological teacher supervision caused by inter-subject variability, signal artifacts, and temporal inconsistency, BioKD incorporates a sample-wise reliability-aware gating mechanism together with a progressive distillation strategy. By adaptively regulating the strength of knowledge transfer, the framework suppresses negative transfer induced by unreliable physiological supervision and enables more stable cross-modal distillation. Experiments on DEAP and AMIGOS show that BioKD consistently outperforms representative baselines under both trial-wise and subject-wise evaluation protocols for valence and arousal recognition. For example, BioKD achieves 68.01\% on DEAP (trial-wise arousal) and 65.29\% under the more challenging subject-wise setting, demonstrating improved performance under a subject-independent evaluation setting. Further analyses show that BioKD effectively mitigates overconfident teacher errors and outperforms an entropy-only weighting strategy, confirming the importance of explicitly modeling supervision reliability. In addition, BioKD introduces no additional inference-time overhead relative to the same video student architecture and removes the need for physiological sensing and multimodal synchronization.

cs.LG↗

Coordinate-Queryable Neural Field Reconstruction for EEG Spatial Super-Resolution with Unseen-Electrode Generation

EEG spatial super-resolution (EEGSR) in real deployments is challenged by random channel missingness, unstable electrode quality, and changing visible-channel patterns caused by bad contacts or device variability. Most existing EEGSR methods learn a fixed low-to-high channel mapping under pre-defined input-output layouts, which makes them brittle when missing channels vary at test time. In this paper, we reformulate EEGSR as learning a shared conditional scalp field from partially observed support channels. Specifically, a position-guided encoder summarizes the observed EEG channels and their coordinates into a latent condition, and a conditional implicit neural representation decoder reconstructs target EEG signals by querying this condition at desired electrode coordinates. During inference, the model directly reconstructs unseen electrode signals from the available EEG support and the queried coordinates. To strengthen the constraint of the encoded latent representation on the decoder and thereby construct a more stable scalp field consistent with the observed channels, we further introduce a fidelity-preserving channel corruption training strategy under mixed electrode states. Extensive experiments across multiple EEG datasets demonstrate the effectiveness of our framework for both random missing-channel reconstruction and strict unseen-electrode signal generation. Notably, under the strict held-out-electrode setting on AAD, our method reduces NMSE by 37.5\% and improves SNR by 2.12 dB over the strongest baseline, showing its ability to synthesize signals at electrode locations never exposed during training.

eess.SP↗

Chirp Parameter Optimization and Distributed Detection for Cooperative RSMA-AFDM Systems

Affine frequency division multiplexing (AFDM) exhibits excellent Doppler robustness and the ability to characterize doubly selective channels. However, its signal dispersion characteristics make it challenging to directly adopt traditional time-frequency multiple access schemes. To address this issue, we introduce cooperative rate splitting multiple access (RSMA) for AFDM systems. The flexible configuration of AFDM chirp parameters can reduce the correlation between users' equivalent channels, which decreases the interference from RSMA private streams. We conduct a theoretical analysis of the cooperative RSMA-AFDM system and demonstrate that minimizing the overlap in the channel column spaces among users can effectively enhance the system performance. Guided by this analysis, we design a chirp parameter optimization scheme that reduces multi-user interference and maximizes diversity gain. To fully exploit the diversity gain brought by the proposed chirp parameter optimization, two expectation propagation (EP)-based distributed cooperative detection schemes are proposed. First, a decision-fusion-based method is developed, where local information and cooperative information are fused by maximum ratio combining, achieving a globally consistent estimate of the common stream. Second, we develop a belief-consensus EP-based detection scheme. In each iteration, user nodes exchange and fuse the first- and second-order statistics of the common stream, and the resulting beliefs gradually converge to a consistent global decision, which significantly improves the overall reliability.

eess.SP↗

Reading, Not Thinking: Understanding and Bridging the Modality Gap When Text Becomes Pixels in Multimodal LLMs

Multimodal large language models (MLLMs) can process text presented as images, yet they often perform worse than when the same content is provided as textual tokens. We systematically diagnose this "modality gap" by evaluating seven MLLMs across seven benchmarks in five input modes, spanning both synthetically rendered text and realistic document images from arXiv PDFs to Wikipedia pages. We find that the gap is highly sensitive to rendering choices such as font and resolution, and that natural document images often exhibit much smaller gaps, suggesting the performance difference partly reflects evaluation artifacts rather than fundamental limitations. Through a grounded-theory error analysis of over 4,000 examples, we identify the primary cause: image input alone suppresses reasoning effort, with models producing 5--19x shorter outputs that skip step-by-step computation or reasoning. The reluctance to reason, not a failure of perception or knowledge retrieval, drives the performance gap, particularly on tasks requiring multi-step reasoning. We show that a simple, lightweight on-policy self-distillation method by fine-tuning models on their own text-mode reasoning traces paired with image inputs closes this gap, raising image-mode accuracy to match or exceed text-mode performance with over 50\% improvement, and the gains transfer to unseen benchmarks without catastrophic forgetting. Overall, our results and analyses provide a systematic understanding of the modality gap and suggest a practical path toward improving visual text understanding in multimodal language models.

cs.CL↗

Analysis and Compensation of Tx and Rx IQ Imbalances in AFDM System

Affine frequency division multiplexing (AFDM) is a recently proposed multicarrier waveform whose bit error rate (BER) performance in doubly selective channels is comparable to that of orthogonal time-frequency space (OTFS) and superior to that of orthogonal frequency division multiplexing (OFDM). In this paper, the impacts of joint transmitter (Tx) and receiver (Rx) in-phase and quadrature imbalance (IQI) on AFDM signals are investigated, where we show that AFDM suffers more severe IQI than OFDM and OTFS due to the inherent feature of complicated chirp-assisted modulation. We further derive analytical expressions for the pairwise and average bit error probability as a function of the IQI parameters. These indicate that such distortions significantly limit the achievable operating signal-to-noise ratio at the receiver side and data rates. To this end, we propose a cascade compensation scheme to mitigate these effects. Specifically, we first compensate for Rx IQI to convert the improper Gaussian noise into additive white Gaussian noise, and then apply a judicious design to eliminate the Tx IQI. Both analytical and simulation results reveal that joint Tx and Rx IQI introduce an error floor in the BER performance of AFDM systems, whereas the proposed approach effectively compensates such impairments.

eess.SP↗

Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation

To enhance group robustness to spurious correlations, prior work often relies on auxiliary group annotations and assumes identical sets of groups across training and test domains. To overcome these limitations, we propose to leverage superclasses -- categories that lie higher in the semantic hierarchy than the task's actual labels -- as a more intrinsic signal than group labels for discerning spurious correlations. Our model incorporates superclass guidance from a pretrained vision-language model via gradient-based attention alignment, and then integrates feature disentanglement with a theoretically supported minimax-optimal feature-usage strategy. As a result, our approach attains robustness to more complex group structures and spurious correlations, without the need to annotate any training samples. Experiments across diverse domain generalization tasks show that our method significantly outperforms strong baselines and goes well beyond the vision-language model's guidance, with clear improvements in both quantitative metrics and qualitative visualizations.

cs.CV↗

Step-Aware Residual-Guided Diffusion for EEG Spatial Super-Resolution

For real-world BCI applications, lightweight Electroencephalography (EEG) systems offer the best cost-deployment balance. However, such spatial sparsity of EEG limits spatial fidelity, hurting learning and introducing bias. EEG spatial super-resolution methods aim to recover high-density EEG signals from sparse measurements, yet is often hindered by distribution shift and signal distortion and thus reducing fidelity and usability for EEG analysis and visualization. To overcome these challenges, we introduce SRGDiff, a step-aware residual-guided diffusion model that formulates EEG spatial super-resolution as dynamic conditional generation. Our key idea is to learn a dynamic residual condition from the low-density input that predicts the step-wise temporal and spatial details to add and uses the evolving cue to steer the denoising process toward high density reconstructions. At each denoising step, the proposed residual condition is additively fused with the previous denoiser feature maps, then a step-dependent affine modulation scales and shifts the activation to produce the current features. This iterative procedure dynamically extracts step-wise temporal rhythms and spatial-topographic cues to steer high-density recovery and maintain a fidelity-consistency balance. We adopt a comprehensive evaluation protocol spanning signal-, feature-, and downstream-level metrics across SEED, SEED-IV, and Localize-MI and multiple upsampling scales. SRGDiff achieves consistent gains of up to 40% over strong baselines, proving its superiority in the task of EEG spatial super-resolution. Moreover, topographic visualizations comparison and substantial EEG-FID gains jointly indicate that our SR EEG mitigates the spatial-spectral shift between low- and high-density recordings. Our code is available at https://github.com/DhrLhj/ICLR2026SRGDiff.

cs.MM↗

CAFE: Channel-Autoregressive Factorized Encoding for Robust Biosignal Spatial Super-Resolution

High-density biosignal recordings are critical for neural decoding and clinical monitoring, yet real-world deployments often rely on low-density (LD) montages due to hardware and operational constraints. This motivates spatial super-resolution from LD observations, but heterogeneous dependencies under sparse and noisy measurements often lead to artifact propagation and false non-local correlations. To address this, we propose CAFE, a plug-and-play rollout generation scheme that reconstructs the full montage in geometry-aligned stages. Starting from the LD channels, CAFE first recovers nearby channels and then progressively expands to more distal regions, exploiting reliable local structure before introducing non-local interactions. During training, step-wise supervision is applied over channel groups and teacher forcing with epoch-level scheduled sampling along the group dimension is utilized to reduce exposure bias, enabling parallel computation across steps. At test time, CAFE performs an autoregressive rollout across groups, while remaining plug-and-play by reusing any temporal backbone as the shared predictor. Evaluated on $4$ modalities and $6$ datasets, CAFE demonstrates plug-and-play generality across $3$ backbones (MLP, Conv, Transformer) and achieves consistently better reconstruction than $5$ representative baselines.

cs.MM↗

SwarmSys: Decentralized Swarm-Inspired Agents for Scalable and Adaptive Reasoning

Large language model (LLM) agents have shown remarkable reasoning abilities. However, existing multi-agent frameworks often rely on fixed roles or centralized control, limiting scalability and adaptability in long-horizon reasoning. We introduce SwarmSys, a closed-loop framework for distributed multi-agent reasoning inspired by swarm intelligence. Coordination in SwarmSys emerges through iterative interactions among three specialized roles, Explorers, Workers, and Validators, that continuously cycle through exploration, exploitation, and validation. To enable scalable and adaptive collaboration, we integrate adaptive agent and event profiles, embedding-based probabilistic matching, and a pheromone-inspired reinforcement mechanism, supporting dynamic task allocation and self-organizing convergence without global supervision. Across symbolic reasoning, research synthesis, and scientific programming tasks, SwarmSys consistently outperforms baselines, improving both accuracy and reasoning stability. These findings highlight swarm-inspired coordination as a promising paradigm for scalable, robust, and adaptive multi-agent reasoning, suggesting that coordination scaling may rival model scaling in advancing LLM intelligence.

cs.AI↗

PuzzlePlex: Benchmarking Foundation Models on Reasoning and Planning with Puzzles

This work investigates the reasoning and planning capabilities of foundation models and their scalability in complex, dynamic environments. We introduce PuzzlePlex, a benchmark designed to assess these capabilities through a diverse set of puzzles. PuzzlePlex consists of 15 types of puzzles, including deterministic and stochastic games of varying difficulty, as well as single-player and two-player scenarios. The PuzzlePlex framework provides a comprehensive environment for each game, and supports extensibility to generate more challenging instances as foundation models evolve. Additionally, we implement customized game-playing strategies for comparison. Building on this benchmark, we develop fine-grained metrics to measure performance and conduct an in-depth analysis of frontier foundation models across two settings: instruction-based and code-based. Furthermore, we systematically investigate their scaling limits. Our findings show that reasoning models outperform others in instruction-based settings, while code-based execution presents greater challenges but offers a scalable and efficient alternative. PuzzlePlex enables targeted evaluation and guides future improvements in reasoning, planning, and generalization for foundation models.

cs.AI↗