arXiv subjects

Feng Chen

Publications and source records attributed to Feng Chen.

At least 73 records · Page 4Linked to original sources

arXiv2026

DeFacto: Counterfactual Thinking with Images for Enforcing Evidence-Grounded and Faithful Reasoning

Recent advances in multimodal language models (MLLMs) have made thinking with images a dominant paradigm for multimodal reasoning. However, existing methods still fail to ensure evidence-answer consistency, where correct answers must be supported by correct visual evidence. To address this issue, we propose DeFacto, a counterfactual reasoning framework that explicitly aligns visual evidence with final answers. Our approach integrates three complementary training paradigms: positive, counterfactual, and random-masking. We further develop a language-guided evidence construction pipeline that automatically localizes question-relevant regions and generates counterfactual variants, resulting in DeFacto-100K. Building on this dataset, we train MLLMs with GRPO-based reinforcement learning and design three complementary rewards to promote correct answering, structured reasoning, and consistent evidence selection. Moreover, we introduce DeFacto-1.5K, a human-annotated benchmark for systematically evaluating evidence-grounded consistency beyond answer accuracy. Experiments on diverse benchmarks demonstrate that DeFacto substantially improves both answer accuracy and evidence-answer consistency over strong baselines.

cs.AI↗

arXiv2026

Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations

Agentic memory systems enable large language model (LLM) agents to maintain state across long interactions, supporting long-horizon reasoning and personalization beyond fixed context windows. Despite rapid architectural development, the empirical foundations of these systems remain fragile: existing benchmarks are often underscaled, evaluation metrics are misaligned with semantic utility, performance varies significantly across backbone models, and system-level costs are frequently overlooked. This survey presents a structured analysis of agentic memory from both architectural and system perspectives. We first introduce a concise taxonomy of MAG systems based on four memory structures. Then, we analyze key pain points limiting current systems, including benchmark saturation effects, metric validity and judge sensitivity, backbone-dependent accuracy, and the latency and throughput overhead introduced by memory maintenance. By connecting the memory structure to empirical limitations, this survey clarifies why current agentic memory systems often underperform their theoretical promise and outlines directions for more reliable evaluation and scalable system design.

cs.CL↗

arXiv2026

DeTrack: A Benchmark and Altitude-Aware Dual World Model for Drone-embodied Tracking

Aerial object tracking has broad applications in public safety, emergency rescue, wildlife monitoring, and related fields. However, existing aerial tracking benchmarks are mainly based on passive 2D video sequences captured from fixed camera locations or predefined flight paths, where drones are treated as passive cameras rather than embodied agents that actively perceive, interact, and control their motion in dynamic 3D scenes. In this paper, we define a new drone-embodied tracking task, termed DeTrack, which requires a drone to track a target in interactive 3D environments using online egocentric observations and active flight control in a closed loop. We build a large-scale benchmark containing 11,368 target trajectories across diverse scenes, rendering conditions, semantic regions, and moving distractors, together with evaluation metrics for target visibility, tracking accuracy, and trajectory success. We further propose AaDWorlds, an altitude-aware dual world model framework for drone-embodied tracking. AaDWorlds consists of an altitude-aware perception module and dual world models that imagine future states under both high- and low-altitude regimes. By combining pseudo altitude-aware observations and imagined future states, AaDWorlds alleviates the intrinsic altitude-mediated contradiction between target visibility and flight safety. Experiments on the DeTrack benchmark demonstrate that AaDWorlds improves closed-loop tracking performance across all evaluation metrics.

cs.CV↗

arXiv2026

SocialLM: Social Signal Processing of Patient-Provider Communication using LLMs and Contextual Aggregation

Effective patient-provider communication is difficult to assess at scale. We examine whether large language models (LLMs) can track 20 social behaviors from clinical transcripts without fine-tuning. Across three model families and multiple prompting strategies, LLMs reliably detect social signals, though performance varies by patient race and visit segment. To address this variability under query-only API constraints, we introduce an agreement-weighted ensemble using group-level agreement patterns. This approach improves both accuracy and stability over the best individual model, demonstrating a practical pathway for scalable social signal tracking in clinical conversations.

cs.CL↗

arXiv2026

Burst-Mode Ultrafast Laser Welding of Sapphire and Invar Alloy Across Large Interfacial Gaps up to 10 $μ$m

Achieving reliable joining between transparent materials and metals under non-optical-contact conditions remains challenging due to limited energy coupling and uncontrolled interfacial reaction across $μ$m-scale gaps. Burst-mode ultrafast lasers provide a potential solution for large-gap welding through temporally distributed energy deposition. However, the underlying interaction mechanisms and achievable joining limits remain unclear. In this study, burst-mode ultrafast laser welding of sapphire to Invar alloy was investigated under controlled interfacial gaps from 3 to 10 $μ$m. Cross-sectional microscopy, elemental mapping, white-light interferometry, and shear testing were employed to analyze joint morphology, elemental distribution, fracture behavior, and mechanical performance.After optimization of the processing parameters for burst-mode ultrafast laser welding, the interfacial morphological evolution and joint strength under different gap conditions were systematically investigated. At a 3 $μ$m gap, cyclic thermal stresses induced by burst pulses generate transverse micro-crack networks in sapphire, accompanied by a reduction in joint strength with increasing sub-pulse numbers. Notably, at a 10 $μ$m gap, where single-pulse welding fails, burst-mode ultrafast laser welding enables interfacial bridging with a maximum shear strength of 6.3 MPa, representing the highest level among published studies.These results indicate a gap-dependent evolution in burst-mode welding behavior governed by crack formation and energy accumulation. This study provides an important theoretical basis and practical guidance for achieving high-performance joining of dissimilar materials under large gap conditions.

physics.optics↗

arXiv2026

Data-driven Radiative Magnetohydrodynamics Simulations with the MURaM Code: Coronal Heating and Dynamics in an Emerging Active Region

We present the application of the data-driven branch of the MURaM code, which follows the evolution of the active region 11640 over 4 days starting from 2012 December 30 at 12:00 UT and reproduces many key coronal extreme-ultraviolet (EUV) emission features seen in remote sensing observations. Radiative magnetohydrodynamic (MHD) simulations that account for sophisticated energy transport processes, such as those in the real corona, have been extended with the ability to use observations as time-dependent boundaries such that the models follow the evolution of actual active regions. This opens the possibility of a one-to-one model of a target region over an extensive time period. We use a hybrid strategy that combines fast-evolving idealized zero-$β$ models that capture the evolution of the large-scale active region magnetic field over a long time period and sophisticated radiative MHD models for a shorter time period of interest. The synthesized EUV images illustrate the formation of coronal loops that connect the two sunspots or fan out to the domain boundary. The model reveals in three-dimensional space fine structure in the coronal heating and plasma properties, which are usually concealed behind the EUV observables. The volumetric heating rate in bright coronal loops is proportional to $\mathbf{B}^{2}$. The emission-measure-weighted line-of-sight velocity, which represents the Doppler shift of a spectral line forming in a certain temperature range, reveals vigorous dynamics in plasma at different temperatures and ubiquitous MHD waves, as expected in the real solar corona.

astro-ph.SR↗

arXiv2026

LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models

Large reasoning models, such as OpenAI o1 and DeepSeek-R1, tend to become increasingly verbose as their reasoning capabilities improve. These inflated Chain-of-Thought (CoT) trajectories often exceed what the underlying problems require, wasting compute, latency, and context budgets. While introducing length-based efficiency rewards during reinforcement learning offers a natural remedy, existing methods struggle with two fundamental challenges: the optimal balance between correctness and efficiency is non-stationary throughout training, and intrinsic reasoning budgets vary drastically across problems. Relying on static reward weights and global length constraints inevitably forces a compromise between degraded accuracy and unrealized compression. To overcome these limitations, we propose LEAD (Length-Efficient Adaptive and Dynamic reasoning), a method that replaces static heuristics with online, self-adaptive mechanisms. LEAD dynamically calibrates the correctness-efficiency trade-off at each step using a Potential-Scaled Instability, directing optimization capacity to the most informative learning signal. Furthermore, it estimates an adaptive per-problem target length online based on the model's own correct rollouts, applying a symmetric efficiency reward that penalizes both overthinking and over-compression. Evaluated on five mathematical reasoning benchmarks, LEAD achieves the highest accuracy and Accuracy-Efficiency Score among RL-trained efficient-reasoning methods while producing substantially shorter outputs than the base model.

cs.LG↗

arXiv2026

The Distributionally Robust Cyclic Inventory Routing Problem

We study the cyclic inventory routing problem that involves joint decisions on vehicle routing and inventory replenishment on an infinite, cyclic horizon. It considers a single warehouse and a set of geographically dispersed retailers. We model retailer demand as random variables with uncertain distributions belonging to a moment-based ambiguity set. We develop a distributionally robust optimization formulation that minimizes the worst-case expected cost over the ambiguity set, while ensuring service reliability through a distributionally robust chance constraint. Our main results are that we prove that the worst-case expected inventory cost is attained under a multi-point distribution, which can be identified a posteriori via linear programming, and that the distributionally robust chance constraint can be reformulated into near-equivalent deterministic forms. This yields a deterministic reformulation of the original problem. To solve it, we design a nested branch-and-price framework, in which the first level partitions retailers into clusters, and the second level concerns routing and replenishment decisions within each cluster. Computational experiments on both synthetic instances and real-world data from SAIC Volkswagen Automobile Co., Ltd. demonstrate the effectiveness and efficiency of the proposed approach.

math.OC↗

arXiv2026

Fin-PRM: A Domain-Specialized Process Reward Model for Financial Reasoning in Large Language Models

Process Reward Models (PRMs) supervise intermediate reasoning steps in large language models (LLMs), but existing PRMs are mainly trained on general-domain data and struggle with the structured, symbolic, and fact-sensitive nature of financial reasoning. Financial tasks require not only correct final answers but also verifiable intermediate steps grounded in domain knowledge. In this paper, we propose Fin-PRM, a domain-specialized, trajectory-aware PRM for financial reasoning that jointly models step-level correctness and trajectory-level coherence, producing binary supervision signals for both local and global reasoning quality. To support reliable supervision, we construct a high-quality financial reasoning dataset of 3K trajectories, where step- and trajectory-level labels are automatically derived from multi-source reward signals, including Monte Carlo rollouts, LLM-based evaluation, and explicit financial knowledge verification. Fin-PRM defines a unified ranking score that integrates step- and trajectory-level rewards, enabling consistent use across multiple settings. We evaluate Fin-PRM in three scenarios: (1) offline trajectory selection for supervised fine-tuning, (2) reward-guided Best-of-$N$ inference for test-time scaling, and (3) process-aware reward shaping for reinforcement learning. Experiments on financial reasoning benchmarks, including CFLUE and FinQA, show that Fin-PRM consistently outperforms general-purpose PRMs and strong baselines. Our project resources will be available at https://github.com/aliyun/qwen-dianjin.

cs.CL↗

arXiv2026

Self-DACE++: Robust Low-Light Enhancement via Efficient Adaptive Curve Estimation

In this paper, we present Self-DACE++, an improved unsupervised and lightweight framework for Low-Light Image Enhancement (LLIE), building upon our previous Self-Reference Deep Adaptive Curve Estimation (Self-DACE). To better address the trade-off between computational efficiency and restoration quality, Self-DACE++ introduces enhanced Adaptive Adjustment Curves (AACs). These curves, governed by minimal trainable parameters, flexibly adjust the dynamic range while preserving the color fidelity, structural integrity, and naturalness of the enhanced images. To achieve an extremely lightweight architecture without sacrificing performance, we propose a randomized order training strategy coupled with a network fusion mechanism, which compresses the model into an efficient iterative inference structure. Furthermore, we formulate a physics-grounded objective function based on Retinex theory and incorporate a dedicated denoising module to effectively estimate and suppress latent noise in dark regions. Extensive qualitative and quantitative evaluations on multiple real-world benchmark datasets demonstrate that Self-DACE++ outperforms existing state-of-the-art methods, delivering superior enhancement quality with real-time inference capability. The code is available at https://github.com/John-Wendell/Self-DACE.

cs.CV↗

arXiv2026

Knowledge-Data Dually Driven Paradigm for Accurate Landslide Susceptibility Prediction under Data-Scarce Conditions Using Geomorphic Priors and Tabular Foundation Model

Landslide susceptibility prediction is critical for geohazard risk assessment and mitigation. Conventional data-driven paradigm achieves high predictive accuracy but require sufficient conditioning factors and large-scale landslide inventories. However, in practical engineering applications across mountainous and plateau regions, data-scarce conditions are commonly observed, where such data requirements are rarely satisfied, rendering conventional data-driven paradigm inapplicable. To address this issue, we propose a knowledge-data dually driven paradigm for accurate landslide susceptibility prediction under data-scarce conditions. The essential idea behind the proposed novel paradigm is the integration of the geomorphic prior knowledge with scarce landslide data. To validate the proposed paradigm, we first applied it to a data-rich region in central Italy, where a conventional data-driven paradigm trained on the full dataset served as the baseline. By utilizing only 30% of the available landslide data, the proposed paradigm achieved comparable predictive accuracy to the baseline, demonstrating its effectiveness under data-scarce conditions. The paradigm was further evaluated in a genuinely data-scarce environment for application, the Qilian Permafrost Region of the Tibetan Plateau, where it also yielded reliable susceptibility predictions, confirming its applicability under data-scarce conditions.

cs.LG↗

arXiv2026

On-Chip Neodymium-Doped Lithium Niobate Microdisk Laser with Self-Induced Pulsing

Rare-earth-doped materials constitute the foundation of conventional solid-state lasers, but their bulk-crystal form is inherently incompatible with photonic integration, making it challenging to realize compact, high performance nanoscale laser sources. Lithium niobate on insulator (LNOI), with its exceptional electro-optic and nonlinear optical properties, has emerged as one of the most promising platforms for integrated photonics. Combining Nd3+ doping with LNOI offers the unique possibility of uniting the efficient gain provided by Nd3+ ions with the excellent characteristics of LNOI. However, on-chip laser emission from Nd:LNOI has not been demonstrated previously. In this work, we report the first realization of an integrated Nd:LNOI microdisk laser, demonstrating lasing at 1094.17 nm under 785.10 nm pumping with a low threshold of 146 uW and a slope efficiency of 1.962*10^(-5). Beyond continuous-wave operation, we further observe self-induced laser pulsing on the hundred-microsecond scale, with a laser-pulse duration down to 500 us and an oscillation period of 6.45 ms, arising from nonlinear thermo-optic-photorefractive dynamics. We demonstrate stable continuous wave lasing and self-induced pulsed emission within a monolithically integrated Nd:LNOI cavity. Our results expand the operational degrees of freedom for LNOI-based lasers and open a new direction toward deeply integrated gain with intrinsic nonlinear dynamical processes.

physics.optics↗

arXiv2026

Modeling Multiple Support Strategies within a Single Turn for Emotional Support Conversations

Emotional Support Conversation (ESC) aims to assist individuals experiencing distress by generating empathetic and supportive dialogue. While prior work typically assumes that each supporter turn corresponds to a single strategy, real-world supportive communication often involves multiple strategies within a single utterance. In this paper, we revisit the ESC task by formulating it as multi-strategy utterance generation, where each utterance may contain one or more strategy-response pairs. We propose two generation methods: All-in-One, which predicts all strategy-response pairs in a single decoding step, and One-by-One, which iteratively generates strategy-response pairs until completion. Both methods are further enhanced with cognitive reasoning guided by reinforcement learning to improve strategy selection and response composition. We evaluate our models on the ESConv dataset under both utterance-level and dialogue-level settings. Experimental results show that our methods effectively model multi-strategy utterances and lead to improved supportive quality and dialogue success. To our knowledge, this work provides the first systematic empirical evidence that allowing multiple support strategies within a single utterance is both feasible and beneficial for emotional support conversations. All code and data will be publicly available at https://github.com/aliyun/qwen-dianjin.

cs.CL↗

arXiv2026

Chain of Event-Centric Causal Thought for Physically Plausible Video Generation

Physically Plausible Video Generation (PPVG) has emerged as a promising avenue for modeling real-world physical phenomena. PPVG requires an understanding of commonsense knowledge, which remains a challenge for video diffusion models. Current approaches leverage commonsense reasoning capability of large language models to embed physical concepts into prompts. However, generation models often render physical phenomena as a single moment defined by prompts, due to the lack of conditioning mechanisms for modeling causal progression. In this paper, we view PPVG as generating a sequence of causally connected and dynamically evolving events. To realize this paradigm, we design two key modules: (1) Physics-driven Event Chain Reasoning. This module decomposes the physical phenomena described in prompts into multiple elementary event units, leveraging chain-of-thought reasoning. To mitigate causal ambiguity, we embed physical formulas as constraints to impose deterministic causal dependencies during reasoning. (2) Transition-aware Cross-modal Prompting (TCP). To maintain continuity between events, this module transforms causal event units into temporally aligned vision-language prompts. It summarizes discrete event descriptions to obtain causally consistent narratives, while progressively synthesizing visual keyframes of individual events by interactive editing. Comprehensive experiments on PhyGenBench and VideoPhy benchmarks demonstrate that our framework achieves superior performance in generating physically plausible videos across diverse physical domains. Code is available at https://github.com/ZixuanWang0525/CoECT.

cs.CV↗

arXiv2026

Training-free Motion Factorization for Compositional Video Generation

Compositional video generation aims to synthesize multiple instances with diverse appearance and motion. However, current approaches mainly focus on binding semantics, neglecting to understand diverse motion categories specified in prompts. In this paper, we propose a motion factorization framework that decomposes complex motion into three primary categories: motionlessness, rigid motion, and non-rigid motion. Specifically, our framework follows a planning before generation paradigm. (1) During planning, we reason about motion laws on the motion graph to obtain frame-wise changes in the shape and position of each instance. This alleviates semantic ambiguities in the user prompt by organizing it into a structured representation of instances and their interactions. (2) During generation, we modulate the synthesis of distinct motion categories in a disentangled manner. Conditioned on the motion cues, guidance branches stabilize appearance in motionless regions, preserve rigid-body geometry, and regularize local non-rigid deformations. Crucially, our two modules are model-agnostic, which can be seamlessly incorporated into various diffusion model architectures. Extensive experiments demonstrate that our framework achieves impressive performance in motion synthesis on real-world benchmarks. Code is available at https://github.com/ZixuanWang0525/MF-CVG.

cs.CV↗

arXiv2026

FinMCP-Bench: Benchmarking LLM Agents for Real-World Financial Tool Use under the Model Context Protocol

This paper introduces \textbf{FinMCP-Bench}, a novel benchmark for evaluating large language models (LLMs) in solving real-world financial problems through tool invocation of financial model context protocols. FinMCP-Bench contains 613 samples spanning 10 main scenarios and 33 sub-scenarios, featuring both real and synthetic user queries to ensure diversity and authenticity. It incorporates 65 real financial MCPs and three types of samples, single tool, multi-tool, and multi-turn, allowing evaluation of models across different levels of task complexity. Using this benchmark, we systematically assess a range of mainstream LLMs and propose metrics that explicitly measure tool invocation accuracy and reasoning capabilities. FinMCP-Bench provides a standardized, practical, and challenging testbed for advancing research on financial LLM agents.

cs.AI↗

arXiv2026

Stochastic maximum principle for time-changed forward-backward stochastic control problem with Lévy noise

This paper establishes a stochastic maximum principle for optimal control problems governed by time-changed forward-backward stochastic differential equations with Lévy noise. The system incorporates a random, non-decreasing operational time (the inverse of an $α$-stable subordinator) to model phenomena like trapping events and subdiffusion. Using a duality transformation and the convex variational method, we derive necessary and sufficient conditions for optimality, expressed through a novel set of adjoint equations. Finally, the theoretical results are applied to solve an explicit cash management problem under stochastic recursive utility.

math.OC↗

arXiv2026

LiveWorld: Simulating Out-of-Sight Dynamics in Generative Video World Models

Recent generative video world models aim to simulate visual environment evolution, allowing an observer to interactively explore the scene via camera control. However, they implicitly assume that the world only evolves within the observer's field of view. Once an object leaves the observer's view, its state is "frozen" in memory, and revisiting the same region later often fails to reflect events that should have occurred in the meantime. In this work, we identify and formalize this overlooked limitation as the "out-of-sight dynamics" problem, which impedes video world models from representing a continuously evolving world. To address this issue, we propose LiveWorld, a novel framework that extends video world models to support persistent world evolution. Instead of treating the world as static observational memory, LiveWorld models a persistent global state composed of a static 3D background and dynamic entities that continue evolving even when unobserved. To maintain these unseen dynamics, LiveWorld introduces a monitor-based mechanism that autonomously simulates the temporal progression of active entities and synchronizes their evolved states upon revisiting, ensuring spatially coherent rendering. For evaluation, we further introduce LiveBench, a dedicated benchmark for the task of maintaining out-of-sight dynamics. Extensive experiments show that LiveWorld enables persistent event evolution and long-term scene consistency, bridging the gap between existing 2D observation-based memory and true 4D dynamic world simulation. The baseline and benchmark will be publicly available at https://zichengduan.github.io/LiveWorld/index.html.

cs.CV↗