arXiv subjects

Yansong Li

Publications and source records attributed to Yansong Li.

At least 19 recordsLinked to original sources

arXiv2026

LensWalk: Agentic Video Understanding by Planning How You See in Videos

The dense, temporal nature of video presents a profound challenge for automated analysis. Despite the use of powerful Vision-Language Models, prevailing methods for video understanding are limited by the inherent disconnect between reasoning and perception: they rely on static, pre-processed information and cannot actively seek raw evidence from video as their understanding evolves. To address this, we introduce LensWalk, a flexible agentic framework that empowers a Large Language Model reasoner to control its own visual observation actively. LensWalk establishes a tight reason-plan-observe loop where the agent dynamically specifies, at each step, the temporal scope and sampling density of the video it observes. Using a suite of versatile, Vision-Language Model based tools parameterized by these specifications, the agent can perform broad scans for cues, focus on specific segments for fact extraction, and stitch evidence from multiple moments for holistic verification. This design allows for progressive, on-demand evidence gathering that directly serves the agent's evolving chain of thought. Without requiring any model fine-tuning, LensWalk delivers substantial, plug-and-play performance gains on multiple model recipes, boosting their accuracy by over 5\% on challenging long-video benchmarks like LVBench and Video-MME. Our analysis reveals that enabling an agent to control how it sees is key to unlocking more accurate, robust, and interpretable video reasoning.

cs.CV↗

arXiv2026

.tmu: A Low-Entropy Tree-Structured Representation for LLM-Assisted Scientific Writing

As large language models (LLMs) increasingly assist scientific writing, the limitations and token costs of generating TeX become increasingly visible. This paper analyzes TeX's architectural mismatch with LLM workflows, stemming from its lack of an explicit structural representation, to illustrate its limitations on generated semantics and error localization. As an alternative, we introduce .tmu, a low-entropy tree-structured representation. With its efficient data structure and clear contextual boundaries, .tmu outperforms .tex in the above aspects. Experiments across four LLMs provide evidence for this claim in most evaluated settings. Furthermore, we show that due to its lower information entropy, fine-tuning LLMs on .tmu achieves approximately 43% lower final training loss than on .tex. Our work provides a more scalable and LLM-friendly data representation for LLM-assisted scientific writing.

cs.CL↗

arXiv2026

From Deferral to Learning: Online In-Context Knowledge Distillation for LLM Cascades

Standard LLM cascades improve efficiency by deferring difficult queries from weak to strong models. However, these systems are typically static: when faced with repeated or semantically similar queries, they redundantly consult the expensive model, failing to adapt during inference. To address this, we propose Inter-Cascade, an online, interactive framework that transforms the strong model from a temporary helper into a long-term teacher. In our approach, when the strong model resolves a deferred query, it generates a generalized, reusable problem-solving strategy. These strategies are stored in a dynamic repository and retrieved via similarity matching to augment the weak model's context for future queries. This enables the weak model to learn on the job without expensive parameter fine-tuning. We theoretically show that this mechanism improves the weak model's confidence calibration. Empirically, Inter-Cascade outperforms standard cascades on multiple benchmarks, improving weak model and overall system accuracy by up to 33.06 percent and 6.35 percent, while reducing strong model calls by up to 48.05 percent and saving fee by up to 49.63 percent. Inter-Cascade demonstrates effective in-context knowledge transfer between LLMs and provides a general, scalable framework applicable to both open-source and API-based LLMs.

cs.AI↗

arXiv2025

StarWhisper Telescope: An AI framework for automating end-to-end astronomical observations

The exponential growth of large-scale telescope arrays has boosted time-domain astronomy development but introduced operational bottlenecks, including labor-intensive observation planning, data processing, and real-time decision-making. Here we present the StarWhisper Telescope system, an AI agent framework automating end-to-end astronomical observations for surveys like the Nearby Galaxy Supernovae Survey. By integrating large language models with specialized function calls and modular workflows, StarWhisper Telescope autonomously generates site-specific observation lists, executes real-time image analysis via pipelines, and dynamically triggers follow-up proposals upon transient detection. The system reduces human intervention through automated observation planning, telescope controlling and data processing, while enabling seamless collaboration between amateur and professional astronomers. Deployed across Nearby Galaxy Supernovae Survey's network of 10 amateur telescopes, the StarWhisper Telescope has detected transients with promising response times relative to existing surveys. Furthermore, StarWhisper Telescope's scalable agent architecture provides a blueprint for future facilities like the Global Open Transient Telescope Array, where AI-driven autonomy will be critical for managing 60 telescopes.

astro-ph.IM↗

arXiv2025

Privacy-Preserving Parameter-Efficient Fine-Tuning for Large Language Model Services

Parameter-Efficient Fine-Tuning (PEFT) provides a practical way for users to customize Large Language Models (LLMs) with their private data in LLM service scenarios. However, the inherently sensitive nature of private data demands robust privacy preservation measures during the customization of LLM services to ensure data security, maintain user trust, and comply with stringent regulatory standards. Based on PEFT, we propose Privacy-Preserving Parameter-Efficient Fine-Tuning (RAPT), a framework that offers privacy protection for LLM services. RAPT adopts a local privacy approach, enabling users to privatize their data locally using a text-to-text local differential privacy mechanism. Since PEFT performs poorly when directly trained on privatized data, we introduce a novel privatized token reconstruction task that is trained jointly with the downstream task, allowing LLMs to learn better task-dependent representations. Despite the simplicity of our framework, experiments show that RAPT achieves competitive performance across tasks while providing privacy guarantees against adversaries.

cs.CL↗

arXiv2025

SV-TrustEval-C: Evaluating Structure and Semantic Reasoning in Large Language Models for Source Code Vulnerability Analysis

As Large Language Models (LLMs) evolve in understanding and generating code, accurately evaluating their reliability in analyzing source code vulnerabilities becomes increasingly vital. While studies have examined LLM capabilities in tasks like vulnerability detection and repair, they often overlook the importance of both structure and semantic reasoning crucial for trustworthy vulnerability analysis. To address this gap, we introduce SV-TrustEval-C, a benchmark designed to evaluate LLMs' abilities for vulnerability analysis of code written in the C programming language through two key dimensions: structure reasoning - assessing how models identify relationships between code elements under varying data and control flow complexities; and semantic reasoning - examining their logical consistency in scenarios where code is structurally and semantically perturbed. Our results show that current LLMs are far from satisfactory in understanding complex code relationships and that their vulnerability analyses rely more on pattern matching than on robust logical reasoning. These findings underscore the effectiveness of the SV-TrustEval-C benchmark and highlight critical areas for enhancing the reasoning capabilities and trustworthiness of LLMs in real-world vulnerability analysis tasks. Our initial benchmark dataset is publicly available.

cs.SE↗

arXiv2025

HateSieve: A Contrastive Learning Framework for Detecting and Segmenting Hateful Content in Multimodal Memes

Amidst the rise of Large Multimodal Models (LMMs) and their widespread application in generating and interpreting complex content, the risk of propagating biased and harmful memes remains significant. Current safety measures often fail to detect subtly integrated hateful content within ``Confounder Memes''. To address this, we introduce \textsc{HateSieve}, a new framework designed to enhance the detection and segmentation of hateful elements in memes. \textsc{HateSieve} features a novel Contrastive Meme Generator that creates semantically paired memes, a customized triplet dataset for contrastive learning, and an Image-Text Alignment module that produces context-aware embeddings for accurate meme segmentation. Empirical experiments on the Hateful Meme Dataset show that \textsc{HateSieve} not only surpasses existing LMMs in performance with fewer trainable parameters but also offers a robust mechanism for precisely identifying and isolating hateful content. \textcolor{red}{Caution: Contains academic discussions of hate speech; viewer discretion advised.}

cs.AI↗

arXiv2025

Learning to Help in Multi-Class Settings

Deploying complex machine learning models on resource-constrained devices is challenging due to limited computational power, memory, and model retrainability. To address these limitations, a hybrid system can be established by augmenting the local model with a server-side model, where samples are selectively deferred by a rejector and then sent to the server for processing. The hybrid system enables efficient use of computational resources while minimizing the overhead associated with server usage. The recently proposed Learning to Help (L2H) model trains a server model given a fixed local (client) model, differing from the Learning to Defer (L2D) framework, which trains the client for a fixed (expert) server. In both L2D and L2H, the training includes learning a rejector at the client to determine when to query the server. In this work, we extend the L2H model from binary to multi-class classification problems and demonstrate its applicability in a number of different scenarios of practical interest in which access to the server may be limited by cost, availability, or policy. We derive a stage-switching surrogate loss function that is differentiable, convex, and consistent with the Bayes rule corresponding to the 0-1 loss for the L2H model. Experiments show that our proposed methods offer an efficient and practical solution for multi-class classification in resource-constrained environments.

cs.LG↗

arXiv2024

Renaissance of Literate Programming in the Era of LLMs: Enhancing LLM-Based Code Generation in Large-Scale Projects

Large Language Models (LLMs) have helped programmers increase efficiency through code generation, comprehension, and repair. However, their application to large-scale projects remains challenging due to complex interdependencies and the extensive size of modern codebases. Although Knuth's concept of Literate Programming (LP) combines code and natural language to convey logic and intent, its potential for enhancing relationships in large projects has not been fully explored. In this study, we introduce the idea of Interoperable LP (ILP), which leverages literate programming principles to enhance the development of both small-scale documents and large-scale projects with LLMs. We investigate how LLMs perform under ILP-style instructions for both document-oriented tasks and entire projects. Recognizing that many researchers rely on well-structured templates to guide LLMs, we propose a concise prompt engineering method to write LP documents so LLMs can better be involved in code generation. We also examine the capacity of various LLMs to generate Scheme and Python code on the RepoBench benchmark, illustrating the advantages of our approach. Our findings indicate that ILP with LLMs can enhance LLM-based code generation in large-scale project development.

cs.SE↗

arXiv2024

Sonicmesh: Enhancing 3D Human Mesh Reconstruction in Vision-Impaired Environments With Acoustic Signals

3D Human Mesh Reconstruction (HMR) from 2D RGB images faces challenges in environments with poor lighting, privacy concerns, or occlusions. These weaknesses of RGB imaging can be complemented by acoustic signals, which are widely available, easy to deploy, and capable of penetrating obstacles. However, no existing methods effectively combine acoustic signals with RGB data for robust 3D HMR. The primary challenges include the low-resolution images generated by acoustic signals and the lack of dedicated processing backbones. We introduce SonicMesh, a novel approach combining acoustic signals with RGB images to reconstruct 3D human mesh. To address the challenges of low resolution and the absence of dedicated processing backbones in images generated by acoustic signals, we modify an existing method, HRNet, for effective feature extraction. We also integrate a universal feature embedding technique to enhance the precision of cross-dimensional feature alignment, enabling SonicMesh to achieve high accuracy. Experimental results demonstrate that SonicMesh accurately reconstructs 3D human mesh in challenging environments such as occlusions, non-line-of-sight scenarios, and poor lighting.

cs.CV↗

arXiv2024

Generalizing End-To-End Autonomous Driving In Real-World Environments Using Zero-Shot LLMs

Traditional autonomous driving methods adopt a modular design, decomposing tasks into sub-tasks. In contrast, end-to-end autonomous driving directly outputs actions from raw sensor data, avoiding error accumulation. However, training an end-to-end model requires a comprehensive dataset; otherwise, the model exhibits poor generalization capabilities. Recently, large language models (LLMs) have been applied to enhance the generalization capabilities of end-to-end driving models. Most studies explore LLMs in an open-loop manner, where the output actions are compared to those of experts without direct feedback from the real world, while others examine closed-loop results only in simulations. This paper proposes an efficient architecture that integrates multimodal LLMs into end-to-end driving models operating in closed-loop settings in real-world environments. In our architecture, the LLM periodically processes raw sensor data to generate high-level driving instructions, effectively guiding the end-to-end model, even at a slower rate than the raw sensor data. This architecture relaxes the trade-off between the latency and inference quality of the LLM. It also allows us to choose from a wide variety of LLMs to improve high-level driving instructions and minimize fine-tuning costs. Consequently, our architecture reduces data collection requirements because the LLMs do not directly output actions; we only need to train a simple imitation learning model to output actions. In our experiments, the training data for the end-to-end model in a real-world environment consists of only simple obstacle configurations with one traffic cone, while the test environment is more complex and contains multiple obstacles placed in various positions. Experiments show that the proposed architecture enhances the generalization capabilities of the end-to-end model even without fine-tuning the LLM.

cs.RO↗

arXiv2024

EAPCR: A Universal Feature Extractor for Scientific Data without Explicit Feature Relation Patterns

Conventional methods, including Decision Tree (DT)-based methods, have been effective in scientific tasks, such as non-image medical diagnostics, system anomaly detection, and inorganic catalysis efficiency prediction. However, most deep-learning techniques have struggled to surpass or even match this level of success as traditional machine-learning methods. The primary reason is that these applications involve multi-source, heterogeneous data where features lack explicit relationships. This contrasts with image data, where pixels exhibit spatial relationships; textual data, where words have sequential dependencies; and graph data, where nodes are connected through established associations. The absence of explicit Feature Relation Patterns (FRPs) presents a significant challenge for deep learning techniques in scientific applications that are not image, text, and graph-based. In this paper, we introduce EAPCR, a universal feature extractor designed for data without explicit FRPs. Tested across various scientific tasks, EAPCR consistently outperforms traditional methods and bridges the gap where deep learning models fall short. To further demonstrate its robustness, we synthesize a dataset without explicit FRPs. While Kolmogorov-Arnold Network (KAN) and feature extractors like Convolutional Neural Networks (CNNs), Graph Convolutional Networks (GCNs), and Transformers struggle, EAPCR excels, demonstrating its robustness and superior performance in scientific tasks without FRPs.

cs.LG↗

arXiv2024

Predicting and Accelerating Nanomaterials Synthesis Using Machine Learning Featurization

Materials synthesis optimization is constrained by serial feedback processes that rely on manual tools and intuition across multiple siloed modes of characterization. We automate and generalize feature extraction of reflection high-energy electron diffraction (RHEED) data with machine learning to establish quantitatively predictive relationships in small sets (\~10) of expert-labeled data, saving significant time on subsequently grown samples. These predictive relationships are evaluated in a representative material system (\ce{W_{1-x}V_xSe2} on c-plane sapphire (0001)) with two aims: 1) predicting grain alignment of the deposited film using pre-growth substrate data, and 2) estimating vanadium dopant concentration using in-situ RHEED as a proxy for ex-situ methods (e.g. x-ray photoelectron spectroscopy). Both tasks are accomplished using the same materials-agnostic features, avoiding specific system retraining and leading to a potential 80\% time saving over a 100-sample synthesis campaign. These predictions provide guidance to avoid doomed trials, reduce follow-on characterization, and improve control resolution for materials synthesis.

cond-mat.mtrl-sci↗

arXiv2024

When to Trust Your Data: Enhancing Dyna-Style Model-Based Reinforcement Learning With Data Filter

Reinforcement learning (RL) algorithms can be divided into two classes: model-free algorithms, which are sample-inefficient, and model-based algorithms, which suffer from model bias. Dyna-style algorithms combine these two approaches by using simulated data from an estimated environmental model to accelerate model-free training. However, their efficiency is compromised when the estimated model is inaccurate. Previous works address this issue by using model ensembles or pretraining the estimated model with data collected from the real environment, increasing computational and sample complexity. To tackle this issue, we introduce an out-of-distribution (OOD) data filter that removes simulated data from the estimated model that significantly diverges from data collected in the real environment. We show theoretically that this technique enhances the quality of simulated data. With the help of the OOD data filter, the data simulated from the estimated model better mimics the data collected by interacting with the real model. This improvement is evident in the critic updates compared to using the simulated data without the OOD data filter. Our experiment integrates the data filter into the model-based policy optimization (MBPO) algorithm. The results demonstrate that our method requires fewer interactions with the real environment to achieve a higher level of optimality than MBPO, even without a model ensemble.

cs.LG↗

arXiv2023

Another virtue of wavelet forests?

A wavelet forest for a text $T [1..n]$ over an alphabet $σ$ takes $n H_0 (T) + o (n \log σ)$ bits of space and supports access and rank on $T$ in $O (\log σ)$ time. Kärkkäinen and Puglisi (2011) implicitly introduced wavelet forests and showed that when $T$ is the Burrows-Wheeler Transform (BWT) of a string $S$, then a wavelet forest for $T$ occupies space bounded in terms of higher-order empirical entropies of $S$ even when the forest is implemented with uncompressed bitvectors. In this paper we show experimentally that wavelet forests also have better access locality than wavelet trees and are thus interesting even when higher-order compression is not effective on $S$, or when $T$ is not a BWT at all.

cs.DS↗

arXiv2023

Solving Strongly Convex and Smooth Stackelberg Games Without Modeling the Follower

Stackelberg games have been widely used to model interactive decision-making problems in a variety of domains such as energy systems, transportation, cybersecurity, and human-robot interaction. However, existing algorithms for solving Stackelberg games often require knowledge of the follower's cost function or learning dynamics and may also require the follower to provide an exact best response, which can be difficult to obtain in practice. To circumvent this difficulty, we develop an algorithm that does not require knowledge of the follower's cost function or an exact best response, making it more applicable to real-world scenarios. Specifically, our algorithm only requires the follower to provide an approximately optimal action in response to the leader's action. The inexact best response is used in computing an approximate gradient of the leader's objective function, with which zeroth-order bilevel optimization can be applied to obtain an optimal action for the leader. Our algorithm is proved to converge at a linear rate to a neighborhood of the optimal point when the leader's cost function under the follower's best response is strongly convex and smooth.

math.OC↗

arXiv2022

Facilitating Self-monitored Physical Rehabilitation with Virtual Reality and Haptic feedback

Physical rehabilitation is essential to recovery from joint replacement operations. As a representation, total knee arthroplasty (TKA) requires patients to conduct intensive physical exercises to regain the knee's range of motion and muscle strength. However, current joint replacement physical rehabilitation methods rely highly on therapists for supervision, and existing computer-assisted systems lack consideration for enabling self-monitoring, making at-home physical rehabilitation difficult. In this paper, we investigated design recommendations that would enable self-monitored rehabilitation through clinical observations and focus group interviews with doctors and therapists. With this knowledge, we further explored Virtual Reality(VR)-based visual presentation and supplemental haptic motion guidance features in our implementation VReHab, a self-monitored and multimodal physical rehabilitation system with VR and vibrotactile and pneumatic feedback in a TKA rehabilitation context. We found that the third point of view real-time reconstructed motion on a virtual avatar overlaid with the target pose effectively provides motion awareness and guidance while haptic feedback helps enhance users' motion accuracy and stability. Finally, we implemented \systemname to facilitate self-monitored post-operative exercises and validated its effectiveness through a clinical study with 10 patients.

cs.HC↗

arXiv2022

Composite FORCE learning of chaotic echo state networks for time-series prediction

Echo state network (ESN), a kind of recurrent neural networks, consists of a fixed reservoir in which neurons are connected randomly and recursively and obtains the desired output only by training output connection weights. First-order reduced and controlled error (FORCE) learning is an online supervised training approach that can change the chaotic activity of ESNs into specified activity patterns. This paper proposes a composite FORCE learning method based on recursive least squares to train ESNs whose initial activity is spontaneously chaotic, where a composite learning technique featured by dynamic regressor extension and memory data exploitation is applied to enhance parameter convergence. The proposed method is applied to a benchmark problem about predicting chaotic time series generated by the Mackey-Glass system, and numerical results have shown that it significantly improves learning and prediction performances compared with existing methods.

cs.LG↗