arXiv subjects

Ning Liu

Publications and source records attributed to Ning Liu.

At least 19 recordsLinked to original sources

arXiv2026

MVFA: A Multi-View Text-Guided Multimodal Fusion LLM Adapter for Sentiment Analysis and Emotion Recognition

Multimodal sentiment analysis and emotion recognition in conversations demand effective modeling of heterogeneous interactions across textual, acoustic, and visual modalities. Although large language models (LLMs) offer powerful language understanding, adapting them to multimodal affective computing remains challenging: full-model fine-tuning is computationally prohibitive, while many existing lightweight adapters fail to preserve rich textual cues during cross-modal fusion. To address these limitations, we propose the multi-view text-guided multimodal fusion adapter (MVFA), a parameter-efficient framework that augments frozen LLMs with strong multimodal reasoning capability. MVFA first constructs complementary text views via max pooling, mean pooling, and attention pooling; these views then guide cross-modal interactions with audio and visual features. The fused multimodal representations are subsequently compressed into a compact set of learnable pseudo-tokens through an Enhanced Q-Former Fusion Module. Using ChatGLM3-6B-base as the primary backbone, we further validate MVFA on LLaMA2-7B and Qwen3-8B to examine its portability across multiple frozen LLM backbones. MVFA is evaluated on three challenging datasets: CH-SIMS V2.0, MELD, and CHERMA. Experimental results demonstrate that MVFA achieves state-of-the-art performance on key metrics while updating only a small fraction of parameters. Specifically, it attains 84.62\% Acc2 and 84.59\% F1 on CH-SIMS V2.0, 67.36\% Acc and 66.03\% WF1 on MELD, and 74.66\% Acc on CHERMA. These findings establish multi-view text-guided fusion as an effective and scalable paradigm for parameter-efficient multimodal LLM adaptation in affective computing. The code is publicly available at https://github.com/Overwhelm1208/MVFA.

cs.AI↗

arXiv2026

SafeAtlas-VL: Beyond Binary Multimodal Safety with Large-Scale Data and Guard Models

Multimodal safety moderation requires distinguishing risks arising from visual content, user intent, and assistant behavior. Existing safeguards, however, are typically trained for a single judgment target and reduce safety assessment to a binary decision. Consequently, risk becomes difficult to compare across a multimodal interaction, and ambiguous cases are obscured. We introduce SafeAtlas-VL, a dataset of 1.5M training instances that places image-, request-, and response-level judgments on a five-level ordered scale. We curate a broad collection of safety-relevant data from both real-world and synthetic sources and apply a disagreement-aware annotation procedure. The resulting dataset spans 15 harm categories and 55 fine-grained subcategories, covering a broad range of multimodal safety scenarios. We also construct SafeAtlas-Bench, a held-out set of 5,000 instances for evaluating five-level predictions and continuous risk scores. Upon this dataset, we train the SafeAtlas Guard series of models via target-conditioned tuning for multimodal safety detection. Our models not only perform five-way classification of safety levels but also map safety to continuous scores through a soft cumulative ordinal head. Experimental results demonstrate that guard models trained on our dataset exhibit strong generalization: even without using the training sets of other benchmarks, they achieve competitive performance on the corresponding test sets. Notably, our 8B model attains the overall best performance, outperforming the previous SOTA by approximately 4% in F1 score. Code, data, and models are released to support further research. Warning: this paper contains example data that may be offensive, harmful, graphic, or disturbing.

cs.AI↗

arXiv2026

H2R-Bench: Benchmarking Human-to-Robot Manipulation Video Generation in World Models

Large-scale manipulation data is essential for robot learning, yet collecting robot demonstrations remains expensive and difficult to scale. Meanwhile, abundant egocentric human manipulation videos provide rich behavioral experiences, but transferring them across embodiments remains challenging due to differences between human hands and robotic end-effectors. Recent advances in video world models offer a promising pathway to synthesize robot-centric manipulation videos from human observations, while their cross-embodiment transfer capability remains largely unexplored. Therefore, we introduce H2R-Bench, a benchmark for evaluating cross-embodiment human-to-robot manipulation video generation, where models transform egocentric human demonstrations into robot manipulation videos under specified embodiments. Each benchmark instance contains a human demonstration video, target embodiment constraints, and source-grounded annotations covering task goals, action events, functional contacts, and object responses. H2R-Bench evaluates generated videos through five dimensions, including goal-state completion, action-event completion, functional contact transfer, embodiment correctness, and general video quality. We benchmark eleven state-of-the-art video generation models across six manipulation families and two robot embodiments. Our evaluation reveals that current video world models remain limited in human-to-robot manipulation transfer: even leading models often fail in embodiment consistency, functional interaction, and task execution. H2R-Bench provides a systematic diagnostic framework for evaluating whether video world models can bridge the human-to-robot embodiment gap and convert human manipulation observations into robot-centric training resources.

cs.RO↗

arXiv2026

AgenticVAU: Multi-Agent Explore-Verify Reasoning for Video Anomaly Understanding

Video anomaly understanding (VAU) focuses on comprehensively interpreting abnormal events in videos, requiring models to identify anomalous occurrences, discover their supporting evidence, and explain the underlying causes beyond simple anomaly detection. Existing VAU methods often rely on specialized training or limited observations, restricting generalization or evidence coverage. Although single-agent alternatives support adaptive video observation, they still integrate exploration, observation, and decision-making within a unified reasoning process, offering limited role specialization and structured evidence coordination. To address these limitations, we present AgenticVAU, a training-free multi-agent framework that casts VAU as an explore--verify process, where the system first discovers potential anomalies and then verifies them through targeted observations. To achieve this, four specialized agents are introduced to handle visual-rule construction, search planning, video observation, and final decision, respectively. These agents communicate through an anchor registry, a shared evidence memory that binds each observation. Guided by this agent framework, AgenticVAU interleaves broad temporal exploration, dense local verification, and cross-interval comparison until sufficient evidence is collected. We conduct extensive experiments on the ECVA, UCF-Crime, and MSAD subsets of VAU-Bench, the results show that AgenticVAU outperforms zero-shot inference and reinforcement learning-based baselines, demonstrating the value of multi-agent collaboration for video anomaly understanding.

cs.CV↗

arXiv2026

JoyAI-Talker: Full-Duplex Speech Interactive Large Model Built for Empathetic Voice Agents

We present JoyAI-Talker, a full-duplex speech dialogue system that delivers robust foundation model capabilities while empowering empathetic interaction and voice agent intelligence. JoyAI-Talker adopts a modular Thinker-Talker architecture and further implements a unified speech-text joint training pipeline to mitigate the common "cognitive degradation" bottleneck, thereby largely preserving the model's core textual reasoning, STEM, and logical capabilities while extending them to speech-based interaction. For expressive speech synthesis, the Talker module employs a text-controllable generation paradigm that enables natural-language instructions to flexibly control vocal attributes and localized paralinguistic events, such as laughter and sighs, supporting more expressive and fine-grained speech responses. To enhance conversational empathy, we introduce the Persona-Adaptive Empathetic Response (PAER) framework. PAER employs a hierarchical cognitive pipeline to extract non-verbal speaker cues, such as gender, age, and emotional state, from raw input audio, incorporate them into the Thinker's CoT reasoning, and generate context-adaptive responses that align semantically appropriate text with fine-grained control over utterance-level expressiveness and localized paralinguistic events, including sighs, speaking rate, and volume. We further integrate Joy-Duplex, a state-driven, plug-and-play full-duplex framework that functions as an efficient gating engine for real-time turn control. Extensive evaluations show that JoyAI-Talker achieves highly competitive performance on foundational T2T and S2T benchmarks. In full-duplex evaluation, the system reaches a high response rate of 0.88 under user interruptions while maintaining an extremely low false-trigger rate under background speech, demonstrating its readiness for fluid and natural speech dialogue.

cs.SD↗

arXiv2026

Direct Detection of Light Self-Interacting Dark Matter via Electronic Collective Excitations

Models of light dark matter often invoke a light mediator to facilitate interactions with the Standard Model. If sufficiently light, this mediator can induce long-range self-interactions among dark matter particles, offering a compelling resolution to small-scale structure anomalies. However, direct detection of light self-interacting dark matter (SIDM) remains challenging for conventional detectors. In this work, we investigate the sensitivity of searches for light SIDM accelerated by high-energy cosmic rays in silicon detectors. Leveraging the electronic collective excitations, we derive 90\% C.L. exclusion limits using public SENSEI and DAMIC-M ionization data. Our constraints can cover a portion of the light SIDM parameter space favored by galactic small-scale anomalies.

hep-ph↗

arXiv2026

Stateful Cooperative Agents Safeguarding LLMs Against Evolving Multi-Turn Attacks

As LLMs become increasingly integrated into complex applications, their vulnerability to adversarial attacks has raised significant concerns. However, existing defenses remain reactive in nature. This limitation makes it difficult for them to counter sophisticated threats, as adversaries continuously adjust their strategies across multi-turn interactions. In this paper, we present a proactive defense framework for securing LLMs against evolving multi-turn adversarial attacks that combines disruption, misdirection, and adaptation across successive interaction turns. In particular, it employs a cooperative multi-agent architecture in which specialized agents execute complementary defense strategies. These strategies include controlled response pacing to increase attack costs, strategically ambiguous outputs to mislead adversaries into ineffective strategies, and forensic analysis of interaction logs to identify attack patterns and refine defenses. These agents are coordinated by an adaptive mechanism that dynamically adjusts the defense strategy in response to escalating threats. To facilitate comprehensive evaluation, we present the EMRA dataset designed to simulate evolving strategies across multi-turn attacks, including 5,200 adversarial samples across eight attack types. Experimental results on EMRA across multiple LLM backbones show that the proposed framework reduces ASR by 69% on average relative to evaluated state-of-the-art baselines. Beyond suppressing harmful outputs, it sustains deceptive engagement, achieving an average DR more than six times that of the strongest baselines and increasing attacker-token consumption by 198.83% on average relative to evaluated baselines. Code and dataset are available at https://github.com/SiyuanLi00/CoopGuard.

cs.CR↗

arXiv2026

Strong imposition of Dirichlet boundary velocities in structure-preserving discretizations of elastodynamics

The imposition of boundary velocities in finite element models of port-Hamiltonian elastodynamics typically relies on Lagrange multipliers, yielding Differential-Algebraic Equations (DAEs). Alternatively, weak imposition methods that maintain an Ordinary Differential Equation (ODE) structure often exhibit poor accuracy at Dirichlet boundaries. To address these limitations, this paper introduces an additive kinematic decomposition at the continuous level, splitting the displacement and velocity fields into a relative dynamic component that vanishes on the boundary and a prescribed lifting function extending into the interior domain. This decomposition induces a distributed port that maps the effects of the boundary actuation inside the domain. By incorporating this mapping into suitable virtual power principles, we derive lifted port-Hamiltonian system (PHS) models that, upon finite element discretization, reduce to ODE systems in which Dirichlet boundary velocities are strongly imposed. The framework is applied to derive 2-field and 4-field formulations suited to distinct PHS geometric representations. Furthermore, we show that under specific shape functions, standard FEM schemes are recovered, demonstrating that the lifting framework in the discrete models is equivalent to the classic algebraic matrix partitioning in computational mechanics practice. The energy-balance properties and computational performance of the proposed methodology are verified through numerical simulations.

cs.CE↗

arXiv2026

Bringing GRACE to Recommendation: Fine-Tuning for Sustainable and Accurate Personalization

Growing concern about environmental sustainability (e.g., reducing carbon emissions and resource use) and public health has motivated ``green'' recommender systems that steer users toward more eco-friendly and healthier choices. However, many existing green recommendation approaches require training new models from scratch, incurring substantial computational and energy costs. Reranking-based methods, meanwhile, introduce an additional sorting stage at inference, increasing latency and computational cost. In this work, we propose GRACE (Green Recommendation via Adaptive Conflict-rEsolution), a fine-tuning framework that integrates item-level sustainability signals (e.g., eco-scores or health indices) into pretrained recommendation models. Since these green values are usually discrete and non-differentiable, existing methods often rely on pairwise comparisons to promote greener items. GRACE instead introduces a differentiable approximation that enables direct optimization of the green criterion. To balance sustainability and personalization quality, GRACE further employs a gradient projection mechanism to mitigate conflicts between the green objective and the accuracy objective during fine-tuning. Experiments on real-world datasets demonstrate that GRACE improves sustainability-oriented recommendation outcomes while generally preserving recommendation accuracy through a controllable preference-anchored update mechanism.

cs.IR↗

arXiv2026

Track, Rank, Crack: Epistemic Working Memory Scales Multi-Hop Reasoning in Language Agents

Language agents that interleave reasoning and tool use degrade sharply as reasoning chains lengthen, even when each individual step is easy. We trace this to context dilution: an agent's investigative state (what it has confirmed, what it suspects, and what it still needs) lives only implicitly in a growing context window, where early discoveries are buried under later retrievals. We introduce SLEUTH, which makes this state explicit and actionable through a structured epistemic working memory: the agent maintains Confirmed Facts grounded to sources, Active Hypotheses ranked by evidence, and Open Questions that directly drive its next action. Across five multi-hop benchmarks and five established baselines, SLEUTH's advantage grows with difficulty, from +5 points on HotpotQA to +11 on 4-hop chains, surpassing Reflexion without multiple episodes. Analyzing where the remaining gap lies, we identify the evidence sufficiency problem: agents often find the answer but fail to commit, exhausting their budget on needless verification. A lightweight commitment trigger fixes this, but only when the agent already maintains structured state: the identical trigger applied to an unstructured agent yields no improvement, isolating organized epistemic state as the necessary condition for effective commitment. Finally, enforcing protocol adherence on a weaker model recovers up to +19 points on the hardest problems, showing that how an agent organizes its reasoning, not raw model capability, is the active ingredient for scaling multi-hop reasoning.

cs.LG↗

arXiv2026

LLMs as a Jury: Cross-Model Consensus Can Outperform Process Reward Models for LLM Reasoning

Selecting the correct answer from a pool of candidate reasoning chains is the engine of test-time scaling, yet the standard selectors each carry a cost: self-consistency inherits the errors of the single model it resamples, and trained reward models need labeled data and transfer poorly off-distribution. We study a third signal, free at inference time: cross-model consensus, the degree to which independently trained models, each solving the problem once, agree on a final answer. We treat the panel as an LLM-jury, in which the verification signal is the structure of agreement itself, with no model scoring another's work. Across seven benchmarks it selects correct answers better than self-consistency and far better than a model scoring its own candidates: on competition math it closes the entire gap to an oracle selector, while self-scoring closes almost none. The mechanism is error decorrelation: independently trained models err differently, so their wrong answers scatter while the correct one accumulates agreement. We make this precise with a parameter-free law, derived in closed form, that predicts consensus accuracy from three measured panel statistics to a mean absolute error of $0.03$ and exposes the method's ceiling: a shared-error floor where models share a misconception, near zero on math but non-trivial on science. Against four trained verifiers spanning discriminative, outcome, and generative reward models, the free LLM-jury matches the strongest inside their math training domain and is the top selector outside it. Cross-model consensus is thus a verifier we can characterize in advance: a law that says when to trust it, and a floor that marks where it cannot.

cs.LG↗

arXiv2026

Tool-Making and Self-Evolving LLM Agents in Low-Latency Systems

Production LLM agents often waste latency and reliability by regenerating code for the same procedural steps on every request. We replace this inference-time coding loop with an agentic tool-making pipeline that compiles repeated SOP steps into validated, versioned tools before deployment. The tool-maker grounds synthesis in the live environment as it collects execution traces, observes backend schemas and values, generates candidate tools, and repairs them against labeled cases. At runtime, the production agent calls these tools directly and falls back to code generation only when needed. We deploy the approach in a Fulfillment Center alarm-triage system, where an agent diagnoses alarms against a 44-node SOP over heterogeneous metric backends. In production, tool calls reduce p50 latency by 42%. On 1,500 historical alarms, they reduce end-to-end error rate by up to 53% by suppressing run-to-run variance in repeated steps. Because tools return compact structured verdicts, they also enable a simpler direct-call architecture, reducing p50 latency by a further 62% in a controlled ablation. Versioned tools also improve auditability and expose specification gaps and upstream data drift. Our results show that agents that build and maintain their own tool libraries, a key element of self-evolving agents, can make industrial LLM systems faster, more reliable, and easier to operate.

cs.CL↗

arXiv2026

Eluna: An Agentic LLM System for Automating Warehouse Operations with Reasoning and Task Execution

Warehouse operations are governed by Standard Operating Procedures (SOPs) that encode complex, multi-system decision logic, which must be executed reliably under strict time constraints, yet LLM agents lack mechanisms to enforce procedural compliance and degrade under the context overload full SOP specifications introduce. We present Eluna, a production-deployed agentic system for reliable SOP execution. Eluna is a graph-guided, multi-agent framework that encodes SOPs as directed acyclic graphs with progressive disclosure and delegates independent tasks to parallel sub-agents, each with persistent code execution and live data access. To meet production latency and accuracy needs, we use asymmetric episodic distillation where a strong teacher is improved through episodic error memories, then a smaller student is fine-tuned on the corrected trajectories with memory stripped, internalizing corrections without inference-time overhead. On a 13-task benchmark and two production applications, our fine-tuned models match or exceed their teacher, beat all larger off-the-shelf baselines, and reach 94% expert agreement on the ticket processing application.

cs.LG↗

arXiv2026

A skew Murnaghan--Nakayama rule for Hopf dual pairs

We develop a uniform skew Murnaghan--Nakayama theory for graded Hopf dual pairs equipped with a nondegenerate Hopf pairing. Using the completed Cauchy element, its grouplike factorization, and the resulting partial contraction operators, we establish a general skew Cauchy identity together with an abstract skew Murnaghan--Nakayama rule. Specializing this framework recovers and extends the classical skew Murnaghan--Nakayama rule for symmetric functions, and yields new skew Murnaghan--Nakayama formulas in several settings, including the dual pairs $(\mathrm{NSym}, \mathrm{QSym}) $ and $(\Lambda^{(k)}, \Lambda_{(k)}) $ arising in $k$-Schur theory, as well as the type $C$ affine Grassmannian context. As applications, we obtain generating functions for irreducible characters of Ariki--Koike algebras, including their type $A$ and type $B$ specializations, as well as Hecke--Clifford algebras and $\mathfrak q$-rook monoid algebras. We also give ribbon-tableau expansions for skew $(q,t)$-Kostka polynomials and for the entries of the inverse transition matrix, thereby answering a question of Carbonara (1998). Finally, by specializing the auxiliary alphabet $Y$ to sums of powers of primitive roots of unity, we derive a skew plethystic Murnaghan--Nakayama formula together with a Schur expansion for skew modular Schur functions; as a further consequence, we confirm Walker's conjecture (1994) by showing that if the transition from the modular Schur functions to the Schur basis is trivial in the row indexed by $\lambda$, then $\lambda$ must be a $k$-core.

math.CO↗

arXiv2026

Testing lepton-flavor-violating decay of doubly charged Higgs bosons in type-II seesaw via photon fusion at the high-energy LHC

Tiny neutrino masses can be explained by the type-II seesaw mechanism, where a triplet scalar under $SU(2)_{L}$ is predicted. Collider searches for this exotic scalar have been extensively conducted, especially for its doubly charged component $\Delta^{\pm\pm}$. Utilizing the forward detectors at the Large Hadron Collider (LHC), we study the probing sensitivity for the elastic photon fusion production of the scalars $pp\to p(\gamma\gamma\to\Delta^{++}\Delta^{--})p$ followed by the lepton-flavor-violating (LFV) decay channels $\Delta^{\pm\pm}\to e^{\pm}\mu^{\pm}$. With a high center-of-mass energy of 100 TeV and several luminosity scenarios, we can extensively broaden the exclusion bounds in the parametric space of Br$(\Delta^{\pm\pm}\to e^{\pm}\mu^{\pm})$ versus the triplet scalar mass $m_{\Delta}$. Specifically, at the 100 TeV LHC with an integrated luminosity of 3 ab$^{-1}$, the mass exclusion limit at 95\% C.L. can reach around 1150 GeV with the assumption of inverted neutrino mass hierarchy.

hep-ph↗

arXiv2026

Long-time behaviour of two-dimensional Navier-Stokes equations in the presence of Couette flow on the half plane

In this paper, we study the long-time behavior of solutions to the two-dimensional Navier-Stokes equations in the presence of Couette flow on the half plane with Navier-slip boundary conditions. We prove that the total vorticity will approach \begin{align*} -1+\frac{M_2(\omega_{0})}{\nu^{3/2}(1+t)^{5/2}} \bar{\Omega}\left( \frac{x}{\sqrt{\nu(1+t)^3}}, \frac{y}{\sqrt{\nu(1+t)}} \right), \end{align*} where $-1$ is the vorticity of the Couette flow and $\bar{\Omega}$ is the kernel of a Fokker-Planck type operator $\mathcal{L}=\partial_Y^2+\frac32 X\partial_X+\frac12 Y\partial_Y+\frac52-Y\partial_X$. In the proof, we introduce a new idea of studying the spectrum of such type operators with boundary.

math.AP↗

arXiv2026

FedSDR: Federated Self-Distillation with Rectification

Federated fine-tuning of Large Language Models faces severe statistical heterogeneity. However, existing model-level defenses often overlook the root cause: intrinsic data distribution mismatches. In this work, we first establish Federated Self-Distillation (FedSD) as a fundamental and potent strategy. By projecting client representations into a smoothed ``model-understanding space,'' FedSD alone serves as a universal booster, demonstrating superior performance over conventional algorithms. Despite its success, we identify a subtle trade-off termed the Rewrite Paradox -- unconstrained self-distillation can inadvertently increase hallucinations and redundancy. To refine this paradigm, we further propose FedSDR (Federated Self-Distillation with Rectification), the ultimate reinforced framework. It augments FedSD with a dual-stream mechanism: a local LoRA-S (Smoothing) branch to implicitly absorb heterogeneity via distilled data, and a parallel global LoRA-R (Rectification) branch anchored to raw data to enforce factual correctness. By selectively aggregating only LoRA-R, FedSDR yields a globally aligned and faithful model. Extensive experiments verify its superior performance.

cs.LG↗

arXiv2026

Beyond Pairs: Your Language Model is Secretly Optimizing a Preference Graph

Direct Preference Optimization (DPO) aligns language models using pairwise preference comparisons, offering a simple and effective alternative to Reinforcement Learning (RL) from human feedback. However, in many practical settings, training data consists of multiple rollouts per prompt, inducing rich preference structure that pairwise DPO fails to exploit. Collapsing such data into independent pairs discards transitivity, introduces redundant or conflicting supervision, and can lead to unstable optimization. We propose Graph Direct Preference Optimization (GraphDPO), a principled generalization of DPO that operates over directed acyclic preference graphs induced by rollout rankings. GraphDPO encodes dominance relations as edges and optimizes a graph-structured Plackett--Luce-inspired objective that aggregates supervision over graph neighborhoods, enforcing transitivity while recovering standard DPO as a special case. To handle discrete or sparse signals, we introduce an equivalence-class construction where responses with identical preferences form graph layers, and intra-layer edges contribute zero loss, preventing spurious gradients. Despite leveraging full graph structure, GraphDPO maintains linear per-prompt complexity via efficient log-sum-exp aggregation. We further incorporate optional ground-truth anchoring by inserting verified solutions as dominant nodes and applying an annealed schedule that stabilizes early training while gradually relaxing oracle supervision. Experiments on reasoning and program synthesis tasks demonstrate superior performance, suggesting that graph-structured preference modeling is a scalable and robust alternative to pairwise and listwise alignment objectives.

cs.LG↗