Searcharxiv⌕ Search

arXiv subjects

Bo Wang

Publications and source records attributed to Bo Wang.

At least 55 records · Page 3Linked to original sources
arXiv2026

NestedKV: Nested Memory Routing for Long-Context KV Cache Compression

Long-context language models are limited by the memory footprint of the key-value (KV) cache. Existing training-free KV compression methods usually rank tokens by one importance signal -- attention, recency, layer-wise allocation, or key distinctiveness -- which becomes brittle when useful context is globally distinctive, locally episodic, or immediately relevant. We introduce NestedKV, a key-only KV cache compression method inspired by the Continuum Memory System in Nested Learning. NestedKV maintains global, block-level, and sliding-window key anchors, scores tokens by multi-time-scale cosine anomaly, and combines the resulting rankings with a training-free outer learner using head-adaptive mixing and surprise-gated token routing. The score is paired with adaptive per-head budgets and requires no training or LLM modification. Across RULER (4k--32k), LooGLE, LongBench, LongBench-E, InfiniteBench, and MMLU-Pro on Qwen3 and Llama-3.2 models, NestedKV is strongest when the retained cache is small. On Qwen3-4B, it improves over KeyDiff by up to 19.10 points on RULER and 19.29 on LongBench at $r=0.75$; at $r=0.95$, it retains 37.32 on LongBench versus 17.55 for KeyDiff.

cs.CL↗
arXiv2026

RSFusionDet: Underwater RGB-Sonar Multimodal Object Detection

Underwater unimodal object detection faces many challenges in sensor imaging, such as optical images limited by underwater noise and visible distance, and sonar images limited by less object structural information. While, optical images have rich object structural information, and sonar images are less affected by underwater noise and have a longer visible distance. Optical (RGB modality) and sonar (Sonar modality) images have complementary information underwater. In this paper, we create an RGB-Sonar multimodal object detection dataset, \textbf{R}GB-\textbf{S}onar \textbf{Fusion} (RSFusion) and propose evaluation metrics for the benchmark. And we propose the \textbf{R}GB-\textbf{S}onar \textbf{Fusion} \textbf{Det}ector (RSFusionDet) with a new RGB-Sonar multimodal object detection result expression for RGB-Sonar multimodal object detection. We analyze the features of RGB and Sonar modal information, and design a Cross-Attention Fusion (CAFusion) module to fuse RGB-Sonar spatial misalignment features and Object Matching Head (OMHead) with Loss (OMLoss) to match identical objects in RGB-Sonar modalities. Our RSFusionDet achieves 76.4/48.6 AP (RGB/Sonar) for object detection and 83.4 \(\text{F1-Score}_{match}\) for object matching, on RSFusion, which outperforms other object detection models. Compared with the DINO baseline, our method improves by 0.7/1.4 AP (RGB/Sonar) while simultaneously providing reliable cross-modal object matching. The code and datasets are publicly available at https://github.com/LEFTeyex/RSFusionDet.

cs.CV↗
arXiv2026

Cross-Subject Generalization in Decoding Perceived Speech from Non-Invasive Brain Recordings

Decoding perceived speech from non-invasive brain recordings has garnered significant attention in recent years due to its wide range of potential applications. However, existing methods face considerable challenges in cross-subject decoding, primarily due to limited generalizability and the absence of explicit mechanisms for extracting subject-consistent information. These limitations result in high training costs and suboptimal decoding performance. To address these challenges, we propose an innovative Cross-Subject Perceived Speech Decoding (CPSD) framework, which comprises two training stages: source model pre-training and personal specialization. In the source model pre-training stage, contrastive learning is employed to capture shared representations across multiple source subjects. Subsequently, personal specialization initializes the model for the target subject by extracting consistent components from the source model and fine-tuning it using target subject data. Additionally, we introduce the Positional Encoding-based Spatial Attention (PESA) module, which remaps MEG/EEG data into a standardized reference space, thereby enhancing cross-subject consistency and facilitating model training. We evaluate the proposed CPSD framework on three perceived speech neural datasets encompassing different modalities and languages. The results demonstrate that our framework outperforms baseline methods by more than 6.8%, 15.4%, and 15.8% in Top-10 accuracy on the Armeni 2022, PKUEEG 2025, and Broderick 2018 datasets, respectively. Further analyses confirm the effectiveness, efficiency, and robustness of the proposed approach.

cs.SD↗
arXiv2026

Scaling-Based Reciprocal Control Barrier Functions for Nonholonomic Mobile Robots

This paper studies the construction of control barrier functions (CBFs) for force-controlled nonholonomic mobile robots subject to relative-degree-two safety constraints arising from position-level obstacle avoidance. A scaling-based reciprocal barrier construction is proposed, in which a positive motion-dependent scaling factor is placed in the numerator of a reciprocal barrier associated with the original physical safety function. The resulting barrier is defined exactly on the interior of the physical safe set and becomes singular on its boundary, thereby preserving the certified interior domain of the original safety constraint while recovering first-order control authority. For a force-controlled nonholonomic robot model, sufficient conditions are derived under which the proposed construction defines a reciprocal CBF, and the interior of the physical safe set is forward invariant under controllers satisfying the induced reciprocal-CBF condition. A scalar strict-feedback system is further used to provide a structural interpretation of the underlying higher-relative-degree cascade under explicit structural assumptions. Numerical simulations demonstrate the induced safe-set geometry and its integration with an optimization-based control framework for obstacle avoidance.

eess.SY↗
arXiv2026

LoRC: Detecting AI-Generated Images via Low-Rank Collapse in Semantic Residuals

Modern generators faithfully model macroscopic semantics, producing synthetic images that appear highly realistic. Consequently, decisive forensic cues reside in subtle non-semantic visual discrepancies. To reveal these cues, we revisit AIGI detection from a geometric perspective and identify an architecture-agnostic signature. Specifically, modern generators exhibit low-rank collapse (\textit{i.e.}, rank degeneracy) in the semantic-residual orthogonal subspace while largely preserving the dominant semantic direction. This structural flattening consistently emerges during the final decoding stage, forming a shared bottleneck across diverse generator architectures. Motivated by this signature, we propose \textbf{LoRC}, a framework that decouples semantic dominance to capture the collapsed residual geometry induced by the generative decoding bottleneck. Our method improves accuracy by an average of 7.0\% across multiple benchmarks and achieves 97.0\% accuracy on 39 unseen generators. These results demonstrate strong cross-model generalization and robustness, making LoRC a reliable approach for AIGI detection in complex real-world environments.

cs.CV↗
arXiv2026

EnterpriseRAG: Benchmarking LLM Instruction Adherence and Robustness under Non-Ideal Enterprise Retrieval

Enterprise RAG deployments face a critical reliability gap: while LLMs satisfy 80% of individual constraints, only 26.8% of responses meet all requirements simultaneously, revealing a 57-point orchestration gap. Existing benchmarks assume clean retrieval with simple queries, failing to capture production conditions where noisy documents and multi-dimensional constraints coexist. We introduce EnterpriseRAG, a benchmark of 983 expert-validated samples across six domains that systematically simulates three failure modes absent from prior work: retrieval noise, knowledge gaps, and factual conflicts, coupled with complex instructions. Evaluation of 13 state-of-the-art LLMs reveals a severe instruction adherence collapse, where high per-constraint satisfaction masks low holistic compliance. Critical findings expose deep barriers under knowledge gaps and factual conflicts, even with reasoning-enhanced inference, indicating production RAG requires explicit context-aware protocols and calibrated judgment. EnterpriseRAG provides a reproducible foundation for measuring and closing these gaps, directly informing deployment decisions for enterprise-scale RAG systems. We will release the benchmark and evaluation framework upon publication.

cs.AI↗
arXiv2026

Intent Speaks Louder: Controllable User Simulation Beyond Response Imitation

User simulators are widely used as scalable environments for training and evaluating interactive assistants. Generating the next user turn is inherently one-to-many: the same profile and dialogue context may support multiple plausible continuations with different local interaction intents. A fluent response may therefore advance the dialogue through an inappropriate intent, such as acceptance rather than repair. Our key insight is that controllable user simulation should separate which local interaction intent the next user turn should realize from how that intent is expressed in language. We introduce UserIDA (User Intent-Directive Alignment), which exposes interaction intent as an explicit per-turn directive. UserIDA defines a six-way intent interface, learns directive-conditioned generation through supervised fine-tuning, and uses intent-calibrated policy optimization during group-based reinforcement learning. The reward preserves composite response quality while ensuring that intent-violating candidates rank below compliant alternatives in mixed groups. On LMSYS-USP, UserIDA achieves 86.6\% intent accuracy, outperforming the strongest dedicated user-simulator baseline by 24.3 percentage points while improving semantic and stylistic similarity. In within-context interventions, it realizes at least four of the six target intents in 91.7\% of evaluated dialogue states, compared with 22.9\% for the strongest external baseline. These results establish per-turn intent control as a complementary dimension to response fidelity in user simulation.

cs.CL↗
arXiv2026

VideoVIBE: A Video-Grounded Diagnostic Benchmark for One-Shot Interactive Website Generation

Natural-language-driven "vibe coding" enables the one-shot generation of visually rich and interactive web applications, yet reliable assessment of their quality has not kept pace. Existing evaluations often score isolated artifacts or final task outcomes, offering limited evidence about which failures occur and why. We introduce VideoVIBE, a video-grounded benchmark that transforms human-operated webpage recordings into fine-grained diagnostic tasks. It contains approximately 1.7K diagnostic Video QA instances derived from 6,338 verified failures across generated webpages, spanning semantic-logical, visual-motion, structural-temporal, and functional failures. Diagnoses are grounded primarily in recorded presentation and behavior, with webpage source code used as complementary context. We further propose V2Lens, a training-free, evidence-grounded multi-agent system that challenges and selectively refines initial video-based diagnoses through targeted visual and source-code verification. Across thirteen closed-source and open-weight Video MLLMs, Gemini-2.5-Flash is the strongest standalone model with a score of 64.54, while V2Lens reaches 71.72, an improvement of 7.18 points. Together, our results show that video-grounded evaluation can move beyond isolated artifacts and aggregate outcomes toward a behaviorally faithful and diagnostically informative account of generated application quality.

cs.CV↗
arXiv2026

Teacher-Feature Drifting: One-Step Diffusion Distillation with Pretrained Diffusion Representations

Sampling from pretrained diffusion and flow-matching models typically requires many forward passes to generate diverse and high-fidelity images. Existing distillation methods often rely on multiple auxiliary networks, carefully designed training stages, or complex optimization pipelines. In this work, we revisit the recently proposed Drifting Model objective and show that a single drifting loss can be directly used to simplify one step distillation. A key observation is that the pretrained diffusion teacher itself already provides a strong representation space. Unlike the original Drifting Model, which relies on an additional pretrained feature extractor, we use intermediate hidden states of the pretrained teacher model as the feature representation. This removes the need for training or introducing an extra representation network while preserving a semantically meaningful feature geometry for drifting. Furthermore, we introduce a lightweight mode coverage loss to mitigate mode collapse during distillation and encourage the student generator to cover diverse teacher-supported regions. Extensive experiments on ImageNet and SDXL demonstrate that our method achieves efficient one step generation with competitive image quality and diversity, achieving FID scores of 1.58 on ImageNet-64$\times$64 and 18.4 on SDXL, while substantially simplifying the overall distillation framework.

cs.CV↗
arXiv2026

Beyond the Dirac Delta: Mitigating Diversity Collapse in Reinforcement Fine-Tuning for Versatile Image Generation

Reinforcement learning (RL) has emerged as a powerful paradigm for fine-tuning large-scale generative models, such as diffusion and flow models, to align with complex human preferences and user-specified tasks. A fundamental limitation remains \textit{the curse of diversity collapse}, where the objective formulation and optimization landscape inherently collapse the policy to a Dirac delta distribution. To address this challenge, we propose \textbf{DRIFT} (\textbf{D}ive\textbf{R}sity-\textbf{I}ncentivized Reinforcement \textbf{F}ine-\textbf{T}uning for Versatile Image Generation), an innovative framework that systematically incentivizes output diversity throughout the on-policy fine-tuning process, reconciling strong task alignment with high generation diversity to enhance versatility essential for applications that demand diverse candidate generations. We approach the problem across three representative perspectives: i) \textbf{sampling} a reward-concentrated subset that filters out reward outliers to prevent premature collapse; ii) \textbf{prompting} with stochastic variations to expand the conditioning space, and iii) \textbf{optimization} of the intra-group diversity with a potential-based reward shaping mechanism. Experimental results show that DRIFT achieves superior Pareto dominance regarding task alignment and generation diversity, yielding a $ 9.08\%\!\sim\! 43.46\%$ increase in diversity at equivalent alignment levels and a $ 59.65\% \!\sim\! 65.86\%$ increase in alignment at equivalent levels of diversity.

cs.LG↗
arXiv2026

ContextWeave: A Real-World Workflow Benchmark

Memory is essential as language agents move from isolated tasks to long-horizon, stateful workflows, yet existing evaluations often reduce it to retrieval or question answering. We introduce ContextWeave, a longitudinal benchmark that evaluates whether recalled experience improves downstream agent performance in realistic office-work streams. ContextWeave reconstructs privacy-preserved, multi-month workflows of 14 participants into 1,005 executable tasks, including 568 core evaluation tasks, with instructions, containerized environments, trajectories, and task-specific rubrics. It measures workspace quality and alignment with participant-specific preferences, complemented by diagnostics of relevance, continuity, solvability, and robustness to misleading recall. Across six memory components under a fixed model, the strongest configuration raises Workspace Score from 68.08 to 78.20 and Preference Score from 41.50 to 70.60. With a fixed memory component, recall improves both outcomes for all five tested base models, although gains vary substantially. Our analysis shows that actionable, experience-rich memory supports workflow continuation and reduces redundant exploration more effectively than compact summaries, while it can also be more susceptible to misleading recall. These findings motivate memory systems that optimize not only retrieval relevance but also reliable use during execution.

cs.AI↗
arXiv2026

SeqLLM: Augmenting LLMs with Behavioral-Sequence Modeling for High-Stakes Decisions at WeChat Pay

Merchant risk control at large payment platforms screens tens of millions of merchants daily, where false positives harm legitimate merchants and false negatives leave harmful activity undetected. The hardest cases require jointly understanding a merchant's textual profile and long behavioral sequence. Large language models (LLMs) excel at text but cannot natively model such sequences, while adapting them often causes catastrophic forgetting. We present SeqLLM, a framework that adds behavioral-sequence modeling to a pretrained LLM while preserving its language ability. SeqLLM combines three components: a compact discrete vocabulary that represents behavioral events as native tokens; a lightweight projector, trained with a two-stage alignment curriculum, that grounds these tokens in the LLM's semantic space; and prefix-guided capability injection, which acquires sequence-modeling ability through task-prefixed supervised fine-tuning rather than continual pre-training. SeqLLM is deployed at WeChat Pay, screening millions of merchants daily. Against the production DeepSeek-based LLM baseline, it raises screening precision from 92.0% to 97.5%. Its pretrained behavior-token embeddings also improve Precision@Top-0.01% by 26.8 percentage points in a production fraud detector serving billion-scale transaction traffic. Beyond payments, SeqLLM achieves state-of-the-art results on public recommendation benchmarks. On MovieLens and Amazon, it surpasses the strong User-LLM baseline by up to 32% relative Recall@5 while retaining markedly stronger language ability. On RecIF, it improves Pass@32 by 14.2% over the full OneRec-8B pipeline using only one-fifth of its GPU-days.

cs.CL↗
arXiv2026

The double white dwarf merger model for the progenitors of SN 2021yfj-like events

Context. Recently, a new class of supernovae with prominent narrow emission features of Si, S and Ar has been reported, i.e. SN 2021yfj-like events (SNe Ien). Their progenitor origin is still unknown. It has been suggested that a massive CO WD may evolve into a Si/S-rich WD when accreting He-rich matter at a high mass-transfer rate. If the He companion subsequently evolves into another more massive WD, the merger of this double WD system can generate Si/S-rich CSM through tidal stripping. Following the SN explosion, the interaction between the SN ejecta and the Si/S-rich CSM could produce the Si, S, and Ar emission lines characteristic of SN 2021yfj-like events. Aims. In this work, we aim to determine the initial parameter space of WD+He star systems that can lead to SN 2021yfj-like events via the double WD merger scenario, and to quantify their theoretical Galactic birthrate and delay-time distributions. Methods. We perform detailed binary evolution simulations of a large number of semidetached WD+He star systems to obtain the parameter space that leads to the formation of Si/S-rich WDs and subsequent double WD mergers. We then use binary population synthesis to calculate the Galactic birthrate and delay-time distribution of such events. Results. For the formation of SN 2021yfj-like events, we found that the initial CO WD and He companion masses must lie in the ranges of 1.0-1.2 M_sun and 2.2-2.5 M_sun, respectively. The derived merger rate for this scenario is ~(2.6-4.4)x10^-5 yr^-1 with delay times longer than 450 Myr, which is about 1% of the observed SN Ia rate. Conclusions. We suggest that the double WD merger scenario involving a Si/S-rich WD originating from a CO WD+He star system, represents a significant and competitive model for SN 2021yfj-like events, underscoring the need for further observations of similar events coupled with relevant theoretical investigations.

astro-ph.SR↗
arXiv2026

From Micro-Cognition to Self-Construction: A Four-Layer Integrative Review of Psychological Theories in HCI

Human-computer interaction (HCI) is undergoing a paradigm shift from "tool use" toward "partnership" and even "mind symbiosis," with psychology evolving from a supplementary explanatory tool to a core pillar shaping interaction paradigms and long-term relationships. This paper systematically reviews relevant research and proposes a four-layer integrative framework comprising the Micro-cognitive, Meso-affective, Macro-social, and Self-constructive layers. The framework reveals that: the cognitive layer constitutes the foundation of interaction, the affective layer drives relational engagement, the social layer regulates trust and behavior through norms, and the self-constructive layer points to the ultimate goal of human-machine symbiosis. These four layers follow a progressive logic of "foundation--mediation--context--goal." The review further argues that psychology has shifted from "post-hoc evaluation" to a "proactive design paradigm," with direct implications for emerging scenarios such as generative AI and embodied agents. Concurrently, this paper identifies core challenges including the nature of affect, agency erosion, and ethical risks, and proposes future directions such as bidirectional theory of mind and personalized co-evolution. This framework offers a systematic theoretical perspective for understanding complex human-machine relationships.

cs.CY↗
arXiv2026

Seen, Said, or Forgotten? A Causal Audit of Visual KV Memory Across Dialog Turns

Stateful multimodal assistants encode an image once but may answer questions about it many turns later. Attention-guided visual-KV eviction assumes that evidence irrelevant now will remain dispensable, although future questions are unknown. We ask when a visual fact is actually safe to forget and introduce the Causal Visual Memory Audit (CVMA), a paired single-prefill framework that tests what later answers lose when a visual region, the whole image, or prior assistant text becomes unavailable. On VisDial and ConvBench, current attention can rank future-useful regions worse than random even though a diagnostic marginal-utility control shows substantial selection headroom. Aggregate scores hide this failure when later turns do not need vision; controlled and stock-generated histories reveal a second escape route, in which assistant-text KV replaces image KV for facts already stated but not reliably for unstated facts. In the tested stacks, safe forgetting is supported by low future visual dependence or fact-specific verbalization---not by low current attention.

cs.CV↗
arXiv2026

GoQuant: Geometric Orthogonal Residual Projection for Multiplier-Free Power-of-Two Transformer Quantization

The deployment of Large Language Models (LLMs) and Vision Transformers (ViTs) on edge devices is significantly constrained by memory capacity and the critical timing bottlenecks introduced by dense Multiply--Accumulate (MAC) arrays. In the ultra-low-bit regime, logarithmic Power-of-Two (PoT) quantization provides a hardware-efficient alternative by replacing general multiplications in the dominant dot-product computation with bit-shift operations. However, its non-uniform exponential lattice inherently suffers from a \textbf{Low Angular Resolution Regime}, a structural limitation that becomes particularly pronounced below 4-bit precision and can substantially degrade the representation of high-dimensional feature manifolds. To address this geometric limitation, we propose Geometric Orthogonal Residual Projection Quantization (GoQuant), an algorithm--hardware co-design framework for multiplier-reduced low-bit inference. By formulating quantization as a dual-basis geometric projection, GoQuant constructs a higher-resolution residual lattice while retaining a shift-and-add inner-product structure. Its analytical solver further avoids computationally intensive gradient-based or iterative search procedures. The data-free Geometric-Only (GEO) mode quantizes LLaMA-2-7B in only 0.47 minutes, while the Activation-Refined (REF) mode completes full-model quantization in approximately \textbf{4.4 minutes}.

cs.LG↗
arXiv2026

EviDAG: Auditable Causal DAG Authoring with Biomedical Literature

Constructing causal directed acyclic graphs (DAGs) is a core step in biomedical causal analysis, yet it remains a largely manual process. Analysts must connect study variables to prior literature, evaluate uncertain causal claims, and preserve sufficient provenance for expert review. We present EviDAG, a browser-based system for authoring causal DAGs as auditable, evidence-linked artifacts from biomedical literature. Given free-text descriptions of study concepts, EviDAG creates a reproducible literature snapshot, uses an LLM-based reasoning module to generate structured pairwise causal judgments, links literature-supported judgments to verbatim evidence excerpts, and assembles the judgments into a constraint-checked graph. Each proposed edge includes confidence estimates, provenance, and a reviewable rationale. The interface supports study specification, progress monitoring, evidence review, graph comparison, adjustment-set computation, and export. In evaluations against both compact benchmark DAGs and reference DAGs derived from published literature, EviDAG achieves high edge recall on the literature-based cohort while retaining verifiable evidence trails absent from LLM-only baselines. EviDAG thus reduces the burden of causal DAG curation while making the resulting assumptions auditable, supporting the design, analysis, and interpretation of biomedical studies.

cs.AI↗
arXiv2026

SN 2022erq: A Superluminous Thermonuclear Supernova with Escalating Preexplosion Mass Loss

We present a photometric and spectroscopic study of the superluminous Type Ia supernova SN 2022erq. Its early spectra, dominated by iron-group elements with weak intermediate-mass features, might indicate highly efficient nuclear burning, broadly similar to that inferred for some overluminous SNe Ia. The rapid emergence and persistence of narrow Balmer emission lines superposed on this iron-rich spectrum provide clear evidence of long-lived interaction with a hydrogen-rich circumstellar medium (CSM), establishing SN 2022erq as a member of the rare Ia-CSM class. SN 2022erq reached a peak bolometric luminosity of about 8 x 10^43 erg/s and exhibited an exceptionally slow post-peak decline, indicating that its light curve is dominated by long-duration ejecta-CSM interaction. By combining H-alpha diagnostics with bolometric light-curve modeling, we reconstruct the pre-explosion mass-loss history of the progenitor. The mass-loss rate escalated by one order of magnitude over the final decades, rising from about 0.04 to about 0.6 solar masses per year. This surge produced a massive, extended CSM shell of about 3 solar masses out to about 3.5 x 10^16 cm. The young stellar environment (about 100 Myr) together with this substantial, extensive CSM points to a progenitor system consisting of a white dwarf and an intermediate-mass companion that underwent increasing mass loss prior to explosion.

astro-ph.HE↗