arXiv subjects

Shanwen Tan

Publications and source records attributed to Shanwen Tan.

6 recordsLinked to original sources

arXiv2026

Living-Harness Is an Interactive-Agent Evolver

Large language model (LLM) agents may recover from a failure within an episode or after a retry, yet the same execution failure can recur in later tasks because post-episode feedback rarely revises the persistent harness that guides future interactions. Static harnesses improve reliability through fixed tools, context, memory, and workflow structures, but remain unchanged after deployment. We propose $\textbf{Living-Harness}$, a self-evolving agent harness that converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates. Guided by a domain-level $\textbf{Evolution-SOP}$ ($\textbf{S}$tandard $\textbf{O}$perating $\textbf{P}$rocedure), Living-Harness extracts an episode abstraction and structured update evidence, and writes two complementary forms of procedural knowledge: episodic memory that records trigger conditions, failure patterns, and recovery actions, and a state graph that records state nodes, repair edges, and transition rules. The updated harness state is retrieved to guide future interactions, while tools and base context remain frozen, allowing procedural repairs to accumulate across evolution cycles. On eight interactive environments derived from $\tau^2$-Bench and MultiWOZ-2.4, Living-Harness improves average Pass@1 over the strongest interactive baseline by 10.07 and 9.91 percentage points, respectively, and supports retrieval-only reuse of the evolved harness state across model backbones. Our code will be made publicly available soon at https://github.com/anotherbricki/Living-Harness.

cs.MA↗

arXiv2026

BRIDGE: Biological Evidence Refinement and Heterogeneous Dynamic Gating for Gene Regulatory Networks

Motivation: Gene regulatory network inference from single-cell RNA sequencing (scRNA-seq) data is important for uncovering cell-state-specific transcriptional programs. However, scRNA-seq measurements are sparse and noisy, and experimentally validated TF-target interactions remain limited, making reliable inference challenging. Although graph neural networks have advanced GRN prediction, existing methods often rely on biologically unconstrained graph augmentation, such as random edge perturbation, and insufficiently control information transfer between genes and cells. These limitations may distort regulatory structures and weaken robustness under noisy and weakly supervised settings. Results: To address these issues, we propose an innovative framework named Biological Evidence Refinement and Heterogeneous Dynamic Gating for Gene Regulatory Networks (BRIDGE). BRIDGE extracts gene and cell representations from the expression matrix and its matrix dual, and performs contrastive learning in the gene space and cell space between self and neighbors across the co-expression-refined regulatory view and the original graph. It then applies heterogeneous gated encoding to adaptively regulate information transfer between genes and cells, enabling robust transcription factor-to-target gene prediction. Experiments on benchmark datasets spanning three network types and seven cell types show that BRIDGE achieves state-of-the-art AUROC and AUPRC in most settings. In particular, on Specific networks, BRIDGE improves average AUPRC by 5% over the second-best baseline, GCLink. In cross-cell-type few-shot transfer, BRIDGE consistently outperforms GCLink and GENELink across all six target cell types. A case study on hESC further supports the biological relevance of the predictions, with 9 of the top 10 and 46 of the top 100 novel TF-target interactions validated by ChIPBase.

q-bio.MN↗

arXiv2026

SWIFT: Prompt-Adaptive Memory for Efficient Interactive Long Video Generation

Streaming long-video generation faces a central challenge in continuous semantic switching, requiring adaptive memory to preserve coherent visual evolution. Current approaches rely on cache rebuilding at prompt boundaries or fixed memory budgets, but they introduce redundant computation and limit flexible semantic adaptation. This limitation arises from a mismatch between cached video history and prompt updates, as memory preserves visual continuity while prompt switches demand rapid semantic adaptation. Motivated by this observation, we present SWIFT, Semantic Windowing and Injection for Flexible Transitions, a training-free framework for multi-prompt long-video generation that enables efficient semantic switching while preserving temporal coherence in causal video diffusion models. SWIFT introduces a lightweight Semantic Injection Cache that augments cached video memory rather than reconstructing it from scratch at every prompt boundary. To avoid uniformly perturbing all attention channels, we further perform head-wise semantic injection, so that each attention head receives a prompt update proportional to its alignment with the current video state. In addition, we introduce an Adaptive Dynamic Window that allocates temporal memory according to prompt phase, using larger local context near switching boundaries and smaller windows during stable segments to reduce average inference cost. To preserve long-range semantic consistency under compressed local attention, we further maintain segment-level semantic anchors that summarize prompt-conditioned video history and reintroduce it as compact memory tokens. Compared with current state-of-the-art methods, SWIFT preserves generation quality while achieving 22.6 FPS on a single H100 GPU, establishing a substantially more efficient solution for multi-prompt long-video generation. Our code is available at https://github.com/ShanwenTan/SWIFT.

cs.CV↗

arXiv2026

LLM-MRD: LLM-Guided Multi-View Reasoning Distillation for Fake News Detection

Multimodal fake news detection is crucial for mitigating societal disinformation. Existing approaches attempt to address this by fusing multimodal features or leveraging Large Language Models (LLMs) for advanced reasoning. However, these methods suffer from serious limitations, including a lack of comprehensive multi-view judgment and fusion, and prohibitive reasoning inefficiency due to the high computational costs of LLMs. To address these issues, we propose \textbf{LLM}-Guided \textbf{M}ulti-View \textbf{R}easoning \textbf{D}istillation for Fake News Detection ( \textbf{LLM-MRD}), a novel teacher-student framework. The Student Multi-view Reasoning module first constructs a comprehensive foundation from textual, visual, and cross-modal perspectives. Then, the Teacher Multi-view Reasoning module generates deep reasoning chains as rich supervision signals. Our core Calibration Distillation mechanism efficiently distills this complex reasoning-derived knowledge into the efficient student model. Experiments show LLM-MRD significantly outperforms state-of-the-art baselines. Notably, it demonstrates a comprehensive average improvement of 5.19\% in ACC and 6.33\% in F1-Fake when evaluated across all competing methods and datasets. Our code is available at https://github.com/Nasuro55/LLM-MRD

cs.CL↗

arXiv2023

Turbulence calculation based on the extended Navier-Stokes equations

In this study, we propose a computational method for solving the turbulence problem of incompressible viscous Newtonian fluids based on the extended Navier-Stokes (N-S) equations. With some phenomenological observations and H. J. Kreuer's interpretation of the origin of viscosity, we make a hypothesis in the fluid flow that the shear process is accompanied by eddy formation. Considering the influence of eddy on convection and diffusion, the classical N-S equations have been improved to obtain the extended N-S equations. The extended equations are closed and the source of velocity fluctuations is explicitly considered as additional convection and diffusion. The extended equations are compatible with the classical N-S equations and are able to describe the laminar and turbulent flow in a unified way. In fluid flow simulations, the equations describing the mean flow quantities can be obtained directly from the extended N-S equations without any additional turbulence model. The flow over a cube placed in a channel was numerically investigated to verify the extended equation, these simulation results were in good agreement with the Large Eddy Simulation (LES) and experimental results.

physics.flu-dyn↗

arXiv2023

A new turbulence model based on scale decomposition

Based on the characteristics of the multi-scale and similarity at different scales in turbulent flow, we propose a scale decomposition for solving the turbulence problem of incompressible Newtonian fluid. The solution domain is decomposed into two-level scales, the large scale component represents mean flow and large scale eddies, and the small scale one represents the turbulent fluctuations. The problem is solved in large scale by the equations of motion and the effect of the turbulent fluctuations on the mean flow is evaluated approximately by using equivalent eddy. Furthermore, the effect of equivalent eddy is decomposed into two parts including convective effect and diffusion effect, which is expressed as a function of mean quantities in large scale. The modified Naiver-Stokes equations are established, there ensures the closure of the equations in large scale. Finally, the modified Naiver-Stokes equations is verified by the numerical simulation. Flow around cylinder is numerically investigated and able to obtain flow behavior from low to high Reynolds numbers. A general-purpose turbulence model is established in this study, which is worthy of engineering application.

physics.flu-dyn↗