arXiv subjects

Yang Xiao

Publications and source records attributed to Yang Xiao.

At least 145 records · Page 8Linked to original sources

arXiv2025

LPGNet: A Lightweight Network with Parallel Attention and Gated Fusion for Multimodal Emotion Recognition

Emotion recognition in conversations (ERC) aims to predict the emotional state of each utterance by using multiple input types, such as text and audio. While Transformer-based models have shown strong performance in this task, they often face two major issues: high computational cost and heavy dependence on speaker information. These problems reduce their ability to generalize in real-world conversations. To solve these challenges, we propose LPGNet, a Lightweight network with Parallel attention and Gated fusion for multimodal ERC. The main part of LPGNet is the Lightweight Parallel Interaction Attention (LPIA) module. This module replaces traditional stacked Transformer layers with parallel dot-product attention, which can model both within-modality and between-modality relationships more efficiently. To improve emotional feature learning, LPGNet also uses a dual-gated fusion method. This method filters and combines features from different input types in a flexible and dynamic way. In addition, LPGNet removes speaker embeddings completely, which allows the model to work independently of speaker identity. Experiments on the IEMOCAP dataset show that LPGNet reaches over 87% accuracy and F1-score in 4-class emotion classification. It outperforms strong baseline models while using fewer parameters and showing better generalization across speakers.

eess.AS↗

arXiv2025

Noise-Robust Sound Event Detection and Counting via Language-Queried Sound Separation

Most sound event detection (SED) systems perform well on clean datasets but degrade significantly in noisy environments. Language-queried audio source separation (LASS) models show promise for robust SED by separating target events; existing methods require elaborate multi-stage training and lack explicit guidance for target events. To address these challenges, we introduce event appearance detection (EAD), a counting-based approach that counts event occurrences at both the clip and frame levels. Based on EAD, we propose a co-training-based multi-task learning framework for EAD and SED to enhance SED's performance in noisy environments. First, SED struggles to learn the same patterns as EAD. Then, a task-based constraint is designed to improve prediction consistency between SED and EAD. This framework provides more reliable clip-level predictions for LASS models and strengthens timestamp detection capability. Experiments on DESED and WildDESED datasets demonstrate better performance compared to existing methods, with advantages becoming more pronounced at higher noise levels.

cs.SD↗

arXiv2025

Multilingual Source Tracing of Speech Deepfakes: A First Benchmark

Recent progress in generative AI has made it increasingly easy to create natural-sounding deepfake speech from just a few seconds of audio. While these tools support helpful applications, they also raise serious concerns by making it possible to generate convincing fake speech in many languages. Current research has largely focused on detecting fake speech, but little attention has been given to tracing the source models used to generate it. This paper introduces the first benchmark for multilingual speech deepfake source tracing, covering both mono- and cross-lingual scenarios. We comparatively investigate DSP- and SSL-based modeling; examine how SSL representations fine-tuned on different languages impact cross-lingual generalization performance; and evaluate generalization to unseen languages and speakers. Our findings offer the first comprehensive insights into the challenges of identifying speech generation models when training and inference languages differ. The dataset, protocol and code are available at https://github.com/xuanxixi/Multilingual-Source-Tracing.

eess.AS↗

arXiv2025

LIMO: Less is More for Reasoning

We challenge the prevailing assumption that complex reasoning in large language models (LLMs) necessitates massive training data. We demonstrate that sophisticated mathematical reasoning can emerge with only a few examples. Specifically, through simple supervised fine-tuning, our model, LIMO, achieves 63.3\% accuracy on AIME24 and 95.6\% on MATH500, surpassing previous fine-tuned models (6.5\% on AIME24, 59.2\% on MATH500) while using only 1\% of the training data required by prior approaches. Furthermore, LIMO exhibits strong out-of-distribution generalization, achieving a 45.8\% absolute improvement across diverse benchmarks, outperforming models trained on 100x more data. Synthesizing these findings, we propose the Less-Is-More Reasoning Hypothesis (LIMO Hypothesis): In foundation models where domain knowledge has been comprehensively encoded during pre-training, sophisticated reasoning can emerge through minimal but strategically designed demonstrations of cognitive processes. This hypothesis suggests that the threshold for eliciting complex reasoning is not dictated by task complexity but rather by two key factors: (1) the completeness of the model's pre-trained knowledge base and (2) the effectiveness of post-training examples in serving as "cognitive templates" that guide reasoning.

cs.CL↗

arXiv2025

Information-Assisted Carnot Engine Surpasses Standard Thermodynamic Bounds

Information can improve heat engine performance, but the underlying principles are still not so clear. Here we introduce a Carnot information machine (CIE) and obtain a quantitative relationship between the engine performance and information. We demonstrate that the presence of information changes allows the CIE to operate as a heat engine in the regime where the standard Carnot cycle is prohibited, ensures that the efficiency of the CIE is greater than or equal to the standard Carnot efficiency, and significantly enables it to achieve 100\% efficiency with positive work extraction for arbitrary two-level systems. We explicitly demonstrate these features using a spin-1/2 system and propose an experimental implementation scheme based on a trapped $^{40}\mathrm{Ca}^+$ ion.

cond-mat.stat-mech↗

arXiv2025

Machine Learning Left-Right Breaking from Gravitational Waves

First-order phase transitions in the early universe can generate stochastic gravitational waves (GWs), offering a unique probe of high-scale particle physics. The Left-Right Symmetric Model (LRSM), which restores parity symmetry at high energies and naturally incorporates the seesaw mechanism, allows for such transitions -- particularly during the spontaneous breaking of $SU(2)_R \times SU(2)_L \times U(1)_{B-L} \to SU(2)_L \times U(1)_Y$. This initial step, though less studied, is both theoretically motivated and potentially observable via GWs. In this work, we investigate the GW signatures associated with this first-step phase transition in the minimal LRSM. Due to the complexity and dimensionality of its parameter space, traditional scanning approaches are computationally intensive and inefficient. To overcome this challenge, we employ a Machine Learning Scan (MLS) strategy, integrated with the high-precision three-dimensional effective field theory framework -- using PhaseTracer as an interface to DRalgo -- to efficiently identify phenomenologically viable regions of the parameter space. Through successive MLS iterations, we identify a parameter region that yields GW signals detectable at forthcoming gravitational wave observatories, such as BBO and DECIGO. Additionally, we analyse the evolution of the MLS-recommended parameter space across iterations and perform a sensitivity analysis to identify the most influential parameters in the model. Our findings underscore both the observational prospects of gravitational waves from LRSM phase transitions and the efficacy of machine learning techniques in probing complex beyond the Standard-Model landscapes.

hep-ph↗

arXiv2025

RawTFNet: A Lightweight CNN Architecture for Speech Anti-spoofing

Automatic speaker verification (ASV) systems are often affected by spoofing attacks. Recent transformer-based models have improved anti-spoofing performance by learning strong feature representations. However, these models usually need high computing power. To address this, we introduce RawTFNet, a lightweight CNN model designed for audio signals. The RawTFNet separates feature processing along time and frequency dimensions, which helps to capture the fine-grained details of synthetic speech. We tested RawTFNet on the ASVspoof 2021 LA and DF evaluation datasets. The results show that RawTFNet reaches comparable performance to that of the state-of-the-art models, while also using fewer computing resources. The code and models will be made publicly available.

eess.AS↗

arXiv2025

Enhancing Text-Based Hierarchical Multilabel Classification for Mobile Applications via Contrastive Learning

A hierarchical labeling system for mobile applications (apps) benefits a wide range of downstream businesses that integrate the labeling with their proprietary user data, to improve user modeling. Such a label hierarchy can define more granular labels that capture detailed app features beyond the limitations of traditional broad app categories. In this paper, we address the problem of hierarchical multilabel classification for apps by using their textual information such as names and descriptions. We present: 1) HMCN (Hierarchical Multilabel Classification Network) for handling the classification from two perspectives: the first focuses on a multilabel classification without hierarchical constraints, while the second predicts labels sequentially at each hierarchical level considering such constraints; 2) HMCL (Hierarchical Multilabel Contrastive Learning), a scheme that is capable of learning more distinguishable app representations to enhance the performance of HMCN. Empirical results on our Tencent App Store dataset and two public datasets demonstrate that our approach performs well compared with state-of-the-art methods. The approach has been deployed at Tencent and the multilabel classification outputs for apps have helped a downstream task--credit risk management of user--improve its performance by 10.70% with regard to the Kolmogorov-Smirnov metric, for over one year.

cs.LG↗

arXiv2025

DBA-DFL: Towards Distributed Backdoor Attacks with Network Detection in Decentralized Federated Learning

Distributed backdoor attacks (DBA) have shown a higher attack success rate than centralized attacks in centralized federated learning (FL). However, it has not been investigated in the decentralized FL. In this paper, we experimentally demonstrate that, while directly applying DBA to decentralized FL, the attack success rate depends on the distribution of attackers in the network architecture. Considering that the attackers can not decide their location, this paper aims to achieve a high attack success rate regardless of the attackers' location distribution. Specifically, we first design a method to detect the network by predicting the distance between any two attackers on the network. Then, based on the distance, we organize the attackers in different clusters. Lastly, we propose an algorithm to \textit{dynamically} embed local patterns decomposed from a global pattern into the different attackers in each cluster. We conduct a thorough empirical investigation and find that our method can, in benchmark datasets, outperform both centralized attacks and naive DBA in different decentralized frameworks.

cs.LG↗

arXiv2025

Sculpting Memory: Multi-Concept Forgetting in Diffusion Models via Dynamic Mask and Concept-Aware Optimization

Text-to-image (T2I) diffusion models have achieved remarkable success in generating high-quality images from textual prompts. However, their ability to store vast amounts of knowledge raises concerns in scenarios where selective forgetting is necessary, such as removing copyrighted content, reducing biases, or eliminating harmful concepts. While existing unlearning methods can remove certain concepts, they struggle with multi-concept forgetting due to instability, residual knowledge persistence, and generation quality degradation. To address these challenges, we propose \textbf{Dynamic Mask coupled with Concept-Aware Loss}, a novel unlearning framework designed for multi-concept forgetting in diffusion models. Our \textbf{Dynamic Mask} mechanism adaptively updates gradient masks based on current optimization states, allowing selective weight modifications that prevent interference with unrelated knowledge. Additionally, our \textbf{Concept-Aware Loss} explicitly guides the unlearning process by enforcing semantic consistency through superclass alignment, while a regularization loss based on knowledge distillation ensures that previously unlearned concepts remain forgotten during sequential unlearning. We conduct extensive experiments to evaluate our approach. Results demonstrate that our method outperforms existing unlearning techniques in forgetting effectiveness, output fidelity, and semantic coherence, particularly in multi-concept scenarios. Our work provides a principled and flexible framework for stable and high-fidelity unlearning in generative models. The code will be released publicly.

cs.CV↗

arXiv2025

Enhancing User Engagement in Socially-Driven Dialogue through Interactive LLM Alignments

Enhancing user engagement through interactions plays an essential role in socially-driven dialogues. While prior works have optimized models to reason over relevant knowledge or plan a dialogue act flow, the relationship between user engagement and knowledge or dialogue acts is subtle and does not guarantee user engagement in socially-driven dialogues. To this end, we enable interactive LLMs to learn user engagement by leveraging signals from the future development of conversations. Specifically, we adopt a more direct and relevant indicator of user engagement, i.e., the user's reaction related to dialogue intention after the interaction, as a reward to align interactive LLMs. To achieve this, we develop a user simulator to interact with target interactive LLMs and explore interactions between the user and the interactive LLM system via \textit{i$\times$MCTS} (\textit{M}onte \textit{C}arlo \textit{T}ree \textit{S}earch for \textit{i}nteraction). In this way, we collect a dataset containing pairs of higher and lower-quality experiences using \textit{i$\times$MCTS}, and align interactive LLMs for high-level user engagement by direct preference optimization (DPO) accordingly. Experiments conducted on two socially-driven dialogue scenarios (emotional support conversations and persuasion for good) demonstrate that our method effectively enhances user engagement in interactive LLMs.

cs.CL↗

arXiv2025

Stereo sound event localization and detection based on PSELDnet pretraining and BiMamba sequence modeling

Pre-training methods have achieved significant performance improvements in sound event localization and detection (SELD) tasks, but existing Transformer-based models suffer from high computational complexity. In this work, we propose a stereo sound event localization and detection system based on pre-trained PSELDnet and bidirectional Mamba sequence modeling. We replace the Conformer module with a BiMamba module and introduce asymmetric convolutions to more effectively model the spatiotemporal relationships between time and frequency dimensions. Experimental results demonstrate that the proposed method achieves significantly better performance than the baseline and the original PSELDnet with Conformer decoder architecture on the DCASE2025 Task 3 development dataset, while also reducing computational complexity. These findings highlight the effectiveness of the BiMamba architecture in addressing the challenges of the SELD task.

eess.AS↗

arXiv2025

Boosting Vulnerability Detection of LLMs via Curriculum Preference Optimization with Synthetic Reasoning Data

Large language models (LLMs) demonstrate considerable proficiency in numerous coding-related tasks; however, their capabilities in detecting software vulnerabilities remain limited. This limitation primarily stems from two factors: (1) the absence of reasoning data related to vulnerabilities, which hinders the models' ability to capture underlying vulnerability patterns; and (2) their focus on learning semantic representations rather than the reason behind them, thus failing to recognize semantically similar vulnerability samples. Furthermore, the development of LLMs specialized in vulnerability detection is challenging, particularly in environments characterized by the scarcity of high-quality datasets. In this paper, we propose a novel framework ReVD that excels at mining vulnerability patterns through reasoning data synthesizing and vulnerability-specific preference optimization. Specifically, we construct forward and backward reasoning processes for vulnerability and corresponding fixed code, ensuring the synthesis of high-quality reasoning data. Moreover, we design the triplet supervised fine-tuning followed by curriculum online preference optimization for enabling ReVD to better understand vulnerability patterns. The extensive experiments conducted on PrimeVul and SVEN datasets demonstrate that ReVD sets new state-of-the-art for LLM-based software vulnerability detection, e.g., 12.24\%-22.77\% improvement in the accuracy. The source code and data are available at https://github.com/Xin-Cheng-Wen/PO4Vul.

cs.AI↗

arXiv2025

Towards Dynamic Theory of Mind: Evaluating LLM Adaptation to Temporal Evolution of Human States

As Large Language Models (LLMs) increasingly participate in human-AI interactions, evaluating their Theory of Mind (ToM) capabilities - particularly their ability to track dynamic mental states - becomes crucial. While existing benchmarks assess basic ToM abilities, they predominantly focus on static snapshots of mental states, overlooking the temporal evolution that characterizes real-world social interactions. We present \textsc{DynToM}, a novel benchmark specifically designed to evaluate LLMs' ability to understand and track the temporal progression of mental states across interconnected scenarios. Through a systematic four-step framework, we generate 1,100 social contexts encompassing 5,500 scenarios and 78,100 questions, each validated for realism and quality. Our comprehensive evaluation of ten state-of-the-art LLMs reveals that their average performance underperforms humans by 44.7\%, with performance degrading significantly when tracking and reasoning about the shift of mental states. This performance gap highlights fundamental limitations in current LLMs' ability to model the dynamic nature of human mental states.

cs.CL↗

arXiv2025

JointSplat: Probabilistic Joint Flow-Depth Optimization for Sparse-View Gaussian Splatting

Reconstructing 3D scenes from sparse viewpoints is a long-standing challenge with wide applications. Recent advances in feed-forward 3D Gaussian sparse-view reconstruction methods provide an efficient solution for real-time novel view synthesis by leveraging geometric priors learned from large-scale multi-view datasets and computing 3D Gaussian centers via back-projection. Despite offering strong geometric cues, both feed-forward multi-view depth estimation and flow-depth joint estimation face key limitations: the former suffers from mislocation and artifact issues in low-texture or repetitive regions, while the latter is prone to local noise and global inconsistency due to unreliable matches when ground-truth flow supervision is unavailable. To overcome this, we propose JointSplat, a unified framework that leverages the complementarity between optical flow and depth via a novel probabilistic optimization mechanism. Specifically, this pixel-level mechanism scales the information fusion between depth and flow based on the matching probability of optical flow during training. Building upon the above mechanism, we further propose a novel multi-view depth-consistency loss to leverage the reliability of supervision while suppressing misleading gradients in uncertain areas. Evaluated on RealEstate10K and ACID, JointSplat consistently outperforms state-of-the-art (SOTA) methods, demonstrating the effectiveness and robustness of our proposed probabilistic joint flow-depth optimization approach for high-fidelity sparse-view 3D reconstruction.

cs.CV↗

arXiv2025

Optimal Pricing Strategies for Heterogeneous Customers in Dual-Channel Closed-Loop Supply Chains: A Modeling Approach

Dual-channel closed-loop supply chains (DCCLSCs) play a vital role in attaining both sustainability and profitability. This paper introduces a game-theoretic model to analyze optimal pricing strategies for primary and replacement customers within three distinct recycling frameworks: manufacturer-led, retailer-led, and collaborative recycling. The model identifies equilibrium pricing and subsidy decisions for each scenario, considering the primary customer's preference for the direct channel and the specific roles in recycling. The findings indicate that manufacturers tend to set lower prices in direct channels compared to retailers, aiming to stimulate demand and promote trade-ins. Manufacturer-led recycling initiatives result in stable pricing, whereas retailer-led recycling necessitates higher subsidies. Collaborative recycling strategies yield lower prices and an increase in trade-ins. Primary customers' preference for the direct channel significantly impacts pricing strategies, with a stronger preference leading to lower direct-channel prices and higher manufacturer subsidies. This paper contributes to the field by incorporating primary customer channel preferences and diverse recycling frameworks into DCCLSC pricing models. These insights assist manufacturers and retailers in adjusting pricing strategies and trade-in incentives according to primary customer preferences and associated costs, thereby enhancing profitability and recycling efficiency within DCCLSCs.

math.OC↗

arXiv2025

Pura: An Efficient Privacy-Preserving Solution for Face Recognition

Face recognition is an effective technology for identifying a target person by facial images. However, sensitive facial images raises privacy concerns. Although privacy-preserving face recognition is one of potential solutions, this solution neither fully addresses the privacy concerns nor is efficient enough. To this end, we propose an efficient privacy-preserving solution for face recognition, named Pura, which sufficiently protects facial privacy and supports face recognition over encrypted data efficiently. Specifically, we propose a privacy-preserving and non-interactive architecture for face recognition through the threshold Paillier cryptosystem. Additionally, we carefully design a suite of underlying secure computing protocols to enable efficient operations of face recognition over encrypted data directly. Furthermore, we introduce a parallel computing mechanism to enhance the performance of the proposed secure computing protocols. Privacy analysis demonstrates that Pura fully safeguards personal facial privacy. Experimental evaluations demonstrate that Pura achieves recognition speeds up to 16 times faster than the state-of-the-art.

cs.CV↗

arXiv2025

TF-Mamba: A Time-Frequency Network for Sound Source Localization

Sound source localization (SSL) determines the position of sound sources using multi-channel audio data. It is commonly used to improve speech enhancement and separation. Extracting spatial features is crucial for SSL, especially in challenging acoustic environments. Recently, a novel structure referred to as Mamba demonstrated notable performance across various sequence-based modalities. This study introduces the Mamba for SSL tasks. We consider the Mamba-based model to analyze spatial features from speech signals by fusing both time and frequency features, and we develop an SSL system called TF-Mamba. This system integrates time and frequency fusion, with Bidirectional Mamba managing both time-wise and frequency-wise processing. We conduct the experiments on the simulated and real datasets. Experiments show that TF-Mamba significantly outperforms other advanced methods. The code will be publicly released in due course.

eess.AS↗