arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 559 records · Page 31Linked to original sources

arXiv2025

DynamicMind: A Tri-Mode Thinking System for Large Language Models

Modern large language models (LLMs) often struggle to dynamically adapt their reasoning depth to varying task complexities, leading to suboptimal performance or inefficient resource utilization. To address this, we introduce DynamicMind, a novel tri-mode thinking system. DynamicMind empowers LLMs to autonomously select between Fast, Normal, and Slow thinking modes for zero-shot question answering (ZSQA) tasks through cognitive-inspired prompt engineering. Our framework's core innovations include: (1) expanding the established dual-process framework of fast and slow thinking into a tri-mode thinking system involving a normal thinking mode to preserve the intrinsic capabilities of LLM; (2) proposing the Thinking Density metric, which aligns computational resource allocation with problem complexity; and (3) developing the Thinking Mode Capacity (TMC) dataset and a lightweight Mind Router to predict the optimal thinking mode. Extensive experiments across diverse mathematical, commonsense, and scientific QA benchmarks demonstrate that DynamicMind achieves superior ZSQA capabilities while establishing an effective trade-off between performance and computational efficiency.

cs.CL↗

arXiv2025

ReasonGen-R1: CoT for Autoregressive Image generation models through SFT and RL

Although chain-of-thought reasoning and reinforcement learning (RL) have driven breakthroughs in NLP, their integration into generative vision models remains underexplored. We introduce ReasonGen-R1, a two-stage framework that first imbues an autoregressive image generator with explicit text-based "thinking" skills via supervised fine-tuning on a newly generated reasoning dataset of written rationales, and then refines its outputs using Group Relative Policy Optimization. To enable the model to reason through text before generating images, We automatically generate and release a corpus of model crafted rationales paired with visual prompts, enabling controlled planning of object layouts, styles, and scene compositions. Our GRPO algorithm uses reward signals from a pretrained vision language model to assess overall visual quality, optimizing the policy in each update. Evaluations on GenEval, DPG, and the T2I benchmark demonstrate that ReasonGen-R1 consistently outperforms strong baselines and prior state-of-the-art models. More: aka.ms/reasongen.

cs.CV↗

arXiv2025

Unpacking Graduate Students' Learning Experience with Generative AI Teaching Assistant in A Quantitative Methodology Course

The study was conducted in an Advanced Quantitative Research Methods course involving 20 graduate students. During the course, student inquiries made to the AI were recorded and coded using Bloom's taxonomy and the CLEAR framework. A series of independent sample t-tests and poisson regression analyses were employed to analyse the characteristics of different questions asked by students with different backgrounds. Post course interviews were conducted with 10 students to gain deeper insights into their perceptions. The findings revealed a U-shaped pattern in students' use of the AI assistant, with higher usage at the beginning and towards the end of the course, and a decrease in usage during the middle weeks. Most questions posed to the AI focused on knowledge and comprehension levels, with fewer questions involving deeper cognitive thinking. Students with a weaker mathematical foundation used the AI assistant more frequently, though their inquiries tended to lack explicit and logical structure compared to those with a strong mathematical foundation, who engaged less with the tool. These patterns suggest the need for targeted guidance to optimise the effectiveness of AI tools for students with varying levels of academic proficiency.

cs.HC↗

arXiv2025

Mapping Student-AI Interaction Dynamics in Multi-Agent Learning Environments: Supporting Personalised Learning and Reducing Performance Gaps

Multi-agent AI systems, which simulate diverse instructional roles such as teachers and peers, offer new possibilities for personalized and interactive learning. Yet, student-AI interaction patterns and their pedagogical implications remain unclear. This study explores how university students engaged with multiple AI agents, and how these interactions influenced cognitive outcomes (learning gains) and non-cognitive factors (motivation, technology acceptance). Based on MAIC, an online learning platform with multi-agent, the research involved 305 university students and 19,365 lines of dialogue data. Pre- and post-test scores, self-reported motivation and technology acceptance were also collected. The study identified two engagement patterns: co-construction of knowledge and co-regulation. Lag sequential analysis revealed that students with lower prior knowledge relied more on co-construction of knowledge sequences, showing higher learning gains and post-course motivation. In contrast, students with higher prior knowledge engaged more in co-regulation behaviors but exhibited limited learning improvement. Technology acceptance increased across all groups. These findings suggest that multi-agent AI systems can adapt to students' varying needs, support differentiated engagement, and reduce performance gaps. Implications for personalized system design and future research directions are discussed.

cs.HC↗

arXiv2025

Line shapes in time- and angle-resolved photoemission spectroscopy explored by machine learning

Time- and angle-resolved photoemission spectroscopy is a powerful technique for investigating the dynamics of excited carriers in quantum materials. Typically, data analysis proceeds via the inspection of time distribution curves (TDCs), which represent the time-dependent photoemission intensity in a region of interest -- often chosen somewhat arbitrarily -- in energy-momentum space. Here, we employ $k$-means, an unsupervised machine learning technique, to systematically investigate trends in TDC line shape for quasi-free-standing monolayer graphene and for a simple analytical model. Our analysis reveals how finite energy and time resolution can affect the TDC line shape. We discuss how this can be taken into account in a quantitative analysis, and under what conditions the time-dependent photoemission intensity after laser excitation can be approximated by a simple exponential decay.

cond-mat.str-el↗

arXiv2025

Mixture of insighTful Experts (MoTE): The Synergy of Thought Chains and Expert Mixtures in Self-Alignment

As the capabilities of large language models (LLMs) continue to expand, aligning these models with human values remains a significant challenge. Recent studies show that reasoning abilities contribute significantly to model safety, while integrating Mixture-of-Experts (MoE) architectures can further enhance alignment. In this work, we address a fundamental question: How to effectively incorporate reasoning abilities and MoE architectures into self-alignment process in LLMs? We propose Mixture of insighTful Experts (MoTE), a novel framework that synergistically combines reasoning chains and expert mixtures to improve self-alignments. From a data perspective, MoTE employs a structured reasoning chain comprising four key stages: Question Analysis, Answer Guidance, Safe Answer, and Safety Checking. This approach enhances safety through multi-step reasoning and proves effective even for smaller and less powerful LLMs (e.g., 7B models). From an architectural perspective, MoTE adopts a multi-LoRA framework with step-level routing, where each expert is dedicated to a specific reasoning step. This design eliminates the need for balance losses, ensures stable training, and supports adaptive inference lengths. Experimental results demonstrate that MoTE significantly improves model safety, jailbreak resistance, and over-refusal capabilities, achieving performance comparable to OpenAI's state-of-the-art o1 model.

cs.CL↗

arXiv2025

Hidden Representation Clustering with Multi-Task Representation Learning towards Robust Online Budget Allocation

Marketing optimization, commonly formulated as an online budget allocation problem, has emerged as a pivotal factor in driving user growth. Most existing research addresses this problem by following the principle of 'first predict then optimize' for each individual, which presents challenges related to large-scale counterfactual prediction and solving complexity trade-offs. Note that the practical data quality is uncontrollable, and the solving scale tends to be tens of millions. Therefore, the existing approaches make the robust budget allocation non-trivial, especially in industrial scenarios with considerable data noise. To this end, this paper proposes a novel approach that solves the problem from the cluster perspective. Specifically, we propose a multi-task representation network to learn the inherent attributes of individuals and project the original features into high-dimension hidden representations through the first two layers of the trained network. Then, we divide these hidden representations into $K$ groups through partitioning-based clustering, thus reformulating the problem as an integer stochastic programming problem under different total budgets. Finally, we distill the representation module and clustering model into a multi-category model to facilitate online deployment. Offline experiments validate the effectiveness and superiority of our approach compared to six state-of-the-art marketing optimization algorithms. Online A/B tests on the Meituan platform indicate that the approach outperforms the online algorithm by 0.53% and 0.65%, considering order volume (OV) and gross merchandise volume (GMV), respectively.

cs.LG↗

arXiv2025

Utilizing Precise and Complete Code Context to Guide LLM in Automatic False Positive Mitigation

Static Application Security Testing (SAST) tools are critical to software quality, identifying potential code issues early in development. However, they often produce false positive warnings that require manual review, slowing down development. Thus, automating false positive mitigation (FPM) is essential. The advent of Large Language Models (LLMs), with their strong abilities in natural language and code understanding, offers promising avenues for FPM. Yet current LLM-based FPM method faces two major limitations: 1. The warning-related code snippets extracted are overly broad and cluttered with irrelevant control/data flows, reducing precision; 2. Critical code contexts are missing, leading to incomplete representations that can mislead LLMs and cause inaccurate assessments. To overcome these limitations, we propose LLM4FPM , a lightweight and efficient false positive mitigation framework. It features eCPG-Slicer, which builds an extended code property graph (eCPG) to extract precise line-level code contexts for warnings. Furthermore, the integrated FARF algorithm builds a file reference graph to identify all files that are relevant to warnings in linear time. This enables eCPG-Slicer to obtain rich contextual information without resorting to expensive whole-program analysis. LLM4FPM outperforms the existing method on the Juliet dataset (F1 > 99% across various Common Weakness Enumerations) and improves label accuracy on the D2A dataset to 86%. By leveraging a lightweight open-source LLM, LLM4FPM can significantly save inspection costs up to \$2758 per run (\$0.384 per warning) on Juliet with an average inspection time of 4.7s per warning. Moreover, real-world tests on popular C/C++ projects demonstrate its practicality.

cs.SE↗

arXiv2025

StyleSinger: Style Transfer for Out-of-Domain Singing Voice Synthesis

Style transfer for out-of-domain (OOD) singing voice synthesis (SVS) focuses on generating high-quality singing voices with unseen styles (such as timbre, emotion, pronunciation, and articulation skills) derived from reference singing voice samples. However, the endeavor to model the intricate nuances of singing voice styles is an arduous task, as singing voices possess a remarkable degree of expressiveness. Moreover, existing SVS methods encounter a decline in the quality of synthesized singing voices in OOD scenarios, as they rest upon the assumption that the target vocal attributes are discernible during the training phase. To overcome these challenges, we propose StyleSinger, the first singing voice synthesis model for zero-shot style transfer of out-of-domain reference singing voice samples. StyleSinger incorporates two critical approaches for enhanced effectiveness: 1) the Residual Style Adaptor (RSA) which employs a residual quantization module to capture diverse style characteristics in singing voices, and 2) the Uncertainty Modeling Layer Normalization (UMLN) to perturb the style attributes within the content representation during the training phase and thus improve the model generalization. Our extensive evaluations in zero-shot style transfer undeniably establish that StyleSinger outperforms baseline models in both audio quality and similarity to the reference singing voice samples. Access to singing voice samples can be found at https://aaronz345.github.io/StyleSingerDemo/.

eess.AS↗

arXiv2025

TCSinger: Zero-Shot Singing Voice Synthesis with Style Transfer and Multi-Level Style Control

Zero-shot singing voice synthesis (SVS) with style transfer and style control aims to generate high-quality singing voices with unseen timbres and styles (including singing method, emotion, rhythm, technique, and pronunciation) from audio and text prompts. However, the multifaceted nature of singing styles poses a significant challenge for effective modeling, transfer, and control. Furthermore, current SVS models often fail to generate singing voices rich in stylistic nuances for unseen singers. To address these challenges, we introduce TCSinger, the first zero-shot SVS model for style transfer across cross-lingual speech and singing styles, along with multi-level style control. Specifically, TCSinger proposes three primary modules: 1) the clustering style encoder employs a clustering vector quantization model to stably condense style information into a compact latent space; 2) the Style and Duration Language Model (S\&D-LM) concurrently predicts style information and phoneme duration, which benefits both; 3) the style adaptive decoder uses a novel mel-style adaptive normalization method to generate singing voices with enhanced details. Experimental results show that TCSinger outperforms all baseline models in synthesis quality, singer similarity, and style controllability across various tasks, including zero-shot style transfer, multi-level style control, cross-lingual style transfer, and speech-to-singing style transfer. Singing voice samples can be accessed at https://aaronz345.github.io/TCSingerDemo/.

eess.AS↗

arXiv2025

TCSinger 2: Customizable Multilingual Zero-shot Singing Voice Synthesis

Customizable multilingual zero-shot singing voice synthesis (SVS) has various potential applications in music composition and short video dubbing. However, existing SVS models overly depend on phoneme and note boundary annotations, limiting their robustness in zero-shot scenarios and producing poor transitions between phonemes and notes. Moreover, they also lack effective multi-level style control via diverse prompts. To overcome these challenges, we introduce TCSinger 2, a multi-task multilingual zero-shot SVS model with style transfer and style control based on various prompts. TCSinger 2 mainly includes three key modules: 1) Blurred Boundary Content (BBC) Encoder, predicts duration, extends content embedding, and applies masking to the boundaries to enable smooth transitions. 2) Custom Audio Encoder, uses contrastive learning to extract aligned representations from singing, speech, and textual prompts. 3) Flow-based Custom Transformer, leverages Cus-MOE, with F0 supervision, enhancing both the synthesis quality and style modeling of the generated singing voice. Experimental results show that TCSinger 2 outperforms baseline models in both subjective and objective metrics across multiple related tasks. Singing voice samples are available at https://aaronz345.github.io/TCSinger2Demo/.

eess.AS↗

arXiv2025

Adversarial Semantic and Label Perturbation Attack for Pedestrian Attribute Recognition

Pedestrian Attribute Recognition (PAR) is an indispensable task in human-centered research and has made great progress in recent years with the development of deep neural networks. However, the potential vulnerability and anti-interference ability have still not been fully explored. To bridge this gap, this paper proposes the first adversarial attack and defense framework for pedestrian attribute recognition. Specifically, we exploit both global- and patch-level attacks on the pedestrian images, based on the pre-trained CLIP-based PAR framework. It first divides the input pedestrian image into non-overlapping patches and embeds them into feature embeddings using a projection layer. Meanwhile, the attribute set is expanded into sentences using prompts and embedded into attribute features using a pre-trained CLIP text encoder. A multi-modal Transformer is adopted to fuse the obtained vision and text tokens, and a feed-forward network is utilized for attribute recognition. Based on the aforementioned PAR framework, we adopt the adversarial semantic and label-perturbation to generate the adversarial noise, termed ASL-PAR. We also design a semantic offset defense strategy to suppress the influence of adversarial attacks. Extensive experiments conducted on both digital domains (i.e., PETA, PA100K, MSP60K, RAPv2) and physical domains fully validated the effectiveness of our proposed adversarial attack and defense strategies for the pedestrian attribute recognition. The source code of this paper will be released on https://github.com/Event-AHU/OpenPAR.

cs.CV↗

arXiv2025

AnchorAttention: Difference-Aware Sparse Attention with Stripe Granularity

Large Language Models (LLMs) with extended context lengths face significant computational challenges during the pre-filling phase, primarily due to the quadratic complexity of self-attention. Existing methods typically employ dynamic pattern matching and block-sparse low-level implementations. However, their reliance on local information for pattern identification fails to capture global contexts, and the coarse granularity of blocks leads to persistent internal sparsity, resulting in suboptimal accuracy and efficiency. To address these limitations, we propose \textbf{AnchorAttention}, a difference-aware, dynamic sparse attention mechanism that efficiently identifies critical attention regions at a finer stripe granularity while adapting to global contextual information, achieving superior speed and accuracy. AnchorAttention comprises three key components: (1) \textbf{Pattern-based Anchor Computation}, leveraging the commonalities present across all inputs to rapidly compute a set of near-maximum scores as the anchor; (2) \textbf{Difference-aware Stripe Sparsity Identification}, performing difference-aware comparisons with the anchor to quickly obtain discrete coordinates of significant regions in a stripe-like sparsity pattern; (3) \textbf{Fine-grained Sparse Computation}, replacing the traditional contiguous KV block loading approach with simultaneous discrete KV position loading to maximize sparsity rates while preserving full hardware computational potential. With its finer-grained sparsity strategy, \textbf{AnchorAttention} achieves higher sparsity rates at the same recall level, significantly reducing computation time. Compared to previous state-of-the-art methods, at a text length of 128k, it achieves a speedup of 1.44$\times$ while maintaining higher recall rates.

cs.LG↗

arXiv2025

DarkDiff: Advancing Low-Light Raw Enhancement by Retasking Diffusion Models for Camera ISP

High-quality photography in extreme low-light conditions is challenging but impactful for digital cameras. With advanced computing hardware, traditional camera image signal processor (ISP) algorithms are gradually being replaced by efficient deep networks that enhance noisy raw images more intelligently. However, existing regression-based models often minimize pixel errors and result in oversmoothing of low-light photos or deep shadows. Recent work has attempted to address this limitation by training a diffusion model from scratch, yet those models still struggle to recover sharp image details and accurate colors. We introduce a novel framework to enhance low-light raw images by retasking pre-trained generative diffusion models with the camera ISP. Extensive experiments demonstrate that our method outperforms the state-of-the-art in perceptual quality across three challenging low-light raw image benchmarks.

cs.CV↗

arXiv2025

AI instructional agent improves student's perceived learner control and learning outcome: empirical evidence from a randomized controlled trial

This study examines the impact of an AI instructional agent on students' perceived learner control and academic performance in a medium demanding course with lecturing as the main teaching strategy. Based on a randomized controlled trial, three instructional conditions were compared: a traditional human teacher, a self-paced MOOC with chatbot support, and an AI instructional agent capable of delivering lectures and responding to questions in real time. Students in the AI instructional agent group reported significantly higher levels of perceived learner control compared to the other groups. They also completed the learning task more efficiently and engaged in more frequent interactions with the instructional system. Regression analyzes showed that perceived learner control positively predicted post-test performance, with behavioral indicators such as reduced learning time and higher interaction frequency supporting this relationship. These findings suggest that AI instructional agents, when designed to support personalized pace and responsive interaction, can enhance both students' learning experience and learning outcomes.

cs.CY↗

arXiv2025

Corrupted but Not Broken: Understanding and Mitigating the Negative Impacts of Corrupted Data in Visual Instruction Tuning

Visual Instruction Tuning (VIT) aims to enhance Multimodal Large Language Models (MLLMs), yet its effectiveness is often compromised by corrupted datasets with issues such as hallucinated content, incorrect responses, and poor OCR quality. Previous approaches to address these challenges have focused on refining datasets through high-quality data collection or rule-based filtering that can be costly or limited in scope. In this paper, we conduct a systematic investigation into the impact of corrupted data on MLLMs and discover that, although corrupted data degrade model performance, such adverse effects are largely reversible, and MLLMs are {\bf corrupted but not broken}. Specifically, we find that disabling a small subset of parameters can almost fully restore performance. Moreover, corrupted MLLMs inherently possess the capability to differentiate between clean and corrupted samples, facilitating dataset cleaning without external intervention. Building on these insights, we introduce a corruption-robust training paradigm that significantly surpasses existing strategies for mitigating the effects of corrupted data.

cs.CV↗

arXiv2025

Large Language Model-enhanced Reinforcement Learning for Low-Altitude Economy Networking

Low-Altitude Economic Networking (LAENet) aims to support diverse flying applications below 1,000 meters by deploying various aerial vehicles for flexible and cost-effective aerial networking. However, complex decision-making, resource constraints, and environmental uncertainty pose significant challenges to the development of the LAENet. Reinforcement learning (RL) offers a potential solution in response to these challenges but has limitations in generalization, reward design, and model stability. The emergence of large language models (LLMs) offers new opportunities for RL to mitigate these limitations. In this paper, we first present a tutorial about integrating LLMs into RL by using the capacities of generation, contextual understanding, and structured reasoning of LLMs. We then propose an LLM-enhanced RL framework for the LAENet in terms of serving the LLM as information processor, reward designer, decision-maker, and generator. Moreover, we conduct a case study by using LLMs to design a reward function to improve the learning performance of RL in the LAENet. Finally, we provide a conclusion and discuss future work.

cs.AI↗

arXiv2025

Latent label distribution grid representation for modeling uncertainty

Although \textbf{L}abel \textbf{D}istribution \textbf{L}earning (LDL) has promising representation capabilities for characterizing the polysemy of an instance, the complexity and high cost of the label distribution annotation lead to inexact in the construction of the label space. The existence of a large number of inexact labels generates a label space with uncertainty, which misleads the LDL algorithm to yield incorrect decisions. To alleviate this problem, we model the uncertainty of label distributions by constructing a \textbf{L}atent \textbf{L}abel \textbf{D}istribution \textbf{G}rid (LLDG) to form a low-noise representation space. Specifically, we first construct a label correlation matrix based on the differences between labels, and then expand each value of the matrix into a vector that obeys a Gaussian distribution, thus building a LLDG to model the uncertainty of the label space. Finally, the LLDG is reconstructed by the LLDG-Mixer to generate an accurate label distribution. Note that we enforce a customized low-rank scheme on this grid, which assumes that the label relations may be noisy and it needs to perform noise-reduction with the help of a Tucker reconstruction technique. Furthermore, we attempt to evaluate the effectiveness of the LLDG by considering its generation as an upstream task to achieve the classification of the objects. Extensive experimental results show that our approach performs competitively on several benchmarks.

cs.LG↗