arXiv subjects

Shuai Wang

Publications and source records attributed to Shuai Wang.

At least 379 records · Page 21Linked to original sources

arXiv2025

A Systematic Survey of Automatic Prompt Optimization Techniques

Since the advent of large language models (LLMs), prompt engineering has been a crucial step for eliciting desired responses for various Natural Language Processing (NLP) tasks. However, prompt engineering remains an impediment for end users due to rapid advances in models, tasks, and associated best practices. To mitigate this, Automatic Prompt Optimization (APO) techniques have recently emerged that use various automated techniques to help improve the performance of LLMs on various tasks. In this paper, we present a comprehensive survey summarizing the current progress and remaining challenges in this field. We provide a formal definition of APO, a 5-part unifying framework, and then proceed to rigorously categorize all relevant works based on their salient features therein. We hope to spur further research guided by our framework.

cs.CL↗

arXiv2025

$C^2$AV-TSE: Context and Confidence-aware Audio Visual Target Speaker Extraction

Audio-Visual Target Speaker Extraction (AV-TSE) aims to mimic the human ability to enhance auditory perception using visual cues. Although numerous models have been proposed recently, most of them estimate target signals by primarily relying on local dependencies within acoustic features, underutilizing the human-like capacity to infer unclear parts of speech through contextual information. This limitation results in not only suboptimal performance but also inconsistent extraction quality across the utterance, with some segments exhibiting poor quality or inadequate suppression of interfering speakers. To close this gap, we propose a model-agnostic strategy called the Mask-And-Recover (MAR). It integrates both inter- and intra-modality contextual correlations to enable global inference within extraction modules. Additionally, to better target challenging parts within each sample, we introduce a Fine-grained Confidence Score (FCS) model to assess extraction quality and guide extraction modules to emphasize improvement on low-quality segments. To validate the effectiveness of our proposed model-agnostic training paradigm, six popular AV-TSE backbones were adopted for evaluation on the VoxCeleb2 dataset, demonstrating consistent performance improvements across various metrics.

cs.SD↗

arXiv2025

SPA-Bench: A Comprehensive Benchmark for SmartPhone Agent Evaluation

Smartphone agents are increasingly important for helping users control devices efficiently, with (Multimodal) Large Language Model (MLLM)-based approaches emerging as key contenders. Fairly comparing these agents is essential but challenging, requiring a varied task scope, the integration of agents with different implementations, and a generalisable evaluation pipeline to assess their strengths and weaknesses. In this paper, we present SPA-Bench, a comprehensive SmartPhone Agent Benchmark designed to evaluate (M)LLM-based agents in an interactive environment that simulates real-world conditions. SPA-Bench offers three key contributions: (1) A diverse set of tasks covering system and third-party apps in both English and Chinese, focusing on features commonly used in daily routines; (2) A plug-and-play framework enabling real-time agent interaction with Android devices, integrating over ten agents with the flexibility to add more; (3) A novel evaluation pipeline that automatically assesses agent performance across multiple dimensions, encompassing seven metrics related to task completion and resource consumption. Our extensive experiments across tasks and agents reveal challenges like interpreting mobile user interfaces, action grounding, memory retention, and execution costs. We propose future research directions to ease these difficulties, moving closer to real-world smartphone agent applications. SPA-Bench is available at https://ai-agents-2030.github.io/SPA-Bench/.

cs.AI↗

arXiv2025

MoMuSE: Momentum Multi-modal Target Speaker Extraction for Real-time Scenarios with Impaired Visual Cues

Audio-visual Target Speaker Extraction (AV-TSE) aims to isolate the speech of a specific target speaker from an audio mixture using time-synchronized visual cues. In real-world scenarios, visual cues are not always available due to various impairments, which undermines the stability of AV-TSE. Despite this challenge, humans can maintain attentional momentum over time, even when the target speaker is not visible. In this paper, we introduce the Momentum Multi-modal target Speaker Extraction (MoMuSE), which retains a speaker identity momentum in memory, enabling the model to continuously track the target speaker. Designed for real-time inference, MoMuSE extracts the current speech window with guidance from both visual cues and dynamically updated speaker momentum. Experimental results demonstrate that MoMuSE exhibits significant improvement, particularly in scenarios with severe impairment of visual cues.

cs.SD↗

arXiv2025

Advancing THz Radio Map Construction and Obstacle Sensing: An Integrated Generative Framework in ISAC

Integrated sensing and communication (ISAC) in the terahertz (THz) band enables obstacle detection, which in turn facilitates efficient beam management to mitigate THz signal blockage. Simultaneously, a THz radio map, which captures signal propagation characteristics through the distribution of received signal strength (RSS), is well-suited for sensing, as it inherently contains obstacle-related information and reflects the unique properties of the THz channel. This means that communication-assisted sensing in ISAC can be effectively achieved using a THz radio map. However, constructing a radio map presents significant challenges due to the sparse deployment of THz sensors and their limited ability to accurately measure the RSS distribution, which directly affects obstacle sensing. In this paper, we formulate an integrated problem for the first time, leveraging the mutual enhancement between sensed obstacles and the constructed THz radio maps. To address this challenge while improving generalization, we propose an integration framework based on a conditional generative adversarial network (CGAN), which uncovers the manifold structure of THz radio maps embedded with obstacle information. Furthermore, recognizing the shared environmental semantics across THz radio maps from different beam directions, we introduce a novel voting-based sensing scheme, where obstacles are detected by aggregating votes from THz radio maps generated by the CGAN. Simulation results demonstrate that the proposed framework outperforms non-integrated baselines in both radio map construction and obstacle sensing, achieving up to 44.3% and 90.6% reductions in mean squared error (MSE), respectively, in a real-world scenario. These results validate the effectiveness of the proposed voting-based scheme.

eess.SP↗

arXiv2025

STShield: Single-Token Sentinel for Real-Time Jailbreak Detection in Large Language Models

Large Language Models (LLMs) have become increasingly vulnerable to jailbreak attacks that circumvent their safety mechanisms. While existing defense methods either suffer from adaptive attacks or require computationally expensive auxiliary models, we present STShield, a lightweight framework for real-time jailbroken judgement. STShield introduces a novel single-token sentinel mechanism that appends a binary safety indicator to the model's response sequence, leveraging the LLM's own alignment capabilities for detection. Our framework combines supervised fine-tuning on normal prompts with adversarial training using embedding-space perturbations, achieving robust detection while preserving model utility. Extensive experiments demonstrate that STShield successfully defends against various jailbreak attacks, while maintaining the model's performance on legitimate queries. Compared to existing approaches, STShield achieves superior defense performance with minimal computational overhead, making it a practical solution for real-world LLM deployment.

cs.CL↗

arXiv2025

Chiral Vortical Instability

We revisit the collective modes of chiral matter described by the second-order chiral hydrodynamics, noticing that chiral shear waves (CSWs) may become unstable for momenta above a characteristic scale. In the absence of sufficient dissipation, this instability emerges within the hydrodynamic regime, depending on the interplay between shear viscosity and the anomalous vortical contribution to the stress-energy tensor at second order in hydrodynamic expansion. We show that this instability generates helical flows and name it the {\it chiral vortical instability} (CVI). Alongside the chiral plasma and magnetovortical instabilities, CVI tends to transfer initial microscopic chirality into macroscopic helicities, which combine into a generalized axial charge. We further find that an elementary static Gromeka-Arnold-Beltrami-Childress flow, corresponding to a CSW at a specific momentum, solves the full nonlinear equations of second-order chiral hydrodynamics, whereas global rotation of a chiral medium is not a solution. This observation supports the relevance of CVI beyond the hydrodynamic regime. Finally, we briefly note that CVI may have multiple phenomenological implications across various systems, including QCD matter produced in heavy-ion collisions and primordial plasma in the early Universe.

hep-th↗

arXiv2025

Context-Aware Two-Step Training Scheme for Domain Invariant Speech Separation

Speech separation seeks to isolate individual speech signals from a multi-talk speech mixture. Despite much progress, a system well-trained on synthetic data often experiences performance degradation on out-of-domain data, such as real-world speech mixtures. To address this, we introduce a novel context-aware, two-stage training scheme for speech separation models. In this training scheme, the conventional end-to-end architecture is replaced with a framework that contains a context extractor and a segregator. The two modules are trained step by step to simulate the speech separation process of an auditory system. We evaluate the proposed training scheme through cross-domain experiments on both synthetic and real-world speech mixtures, and demonstrate that our new scheme effectively boosts separation quality across different domains without adaptation, as measured by signal quality metrics and word error rate (WER). Additionally, an ablation study on the real test set highlights that the context information, including phoneme and word representations from pretrained SSL models, serves as effective domain invariant training targets for separation models.

cs.SD↗

arXiv2025

VALL-T: Decoder-Only Generative Transducer for Robust and Decoding-Controllable Text-to-Speech

Recent TTS models with decoder-only Transformer architecture, such as SPEAR-TTS and VALL-E, achieve impressive naturalness and demonstrate the ability for zero-shot adaptation given a speech prompt. However, such decoder-only TTS models lack monotonic alignment constraints, sometimes leading to hallucination issues such as mispronunciation, word skipping and repeating. To address this limitation, we propose VALL-T, a generative Transducer model that introduces shifting relative position embeddings for input phoneme sequence, explicitly indicating the monotonic generation process while maintaining the architecture of decoder-only Transformer. Consequently, VALL-T retains the capability of prompt-based zero-shot adaptation and demonstrates better robustness against hallucinations with a relative reduction of 28.3% in the word error rate.

eess.AS↗

arXiv2025

LaRA: Benchmarking Retrieval-Augmented Generation and Long-Context LLMs -- No Silver Bullet for LC or RAG Routing

Effectively incorporating external knowledge into Large Language Models (LLMs) is crucial for enhancing their capabilities and addressing real-world needs. Retrieval-Augmented Generation (RAG) offers an effective method for achieving this by retrieving the most relevant fragments into LLMs. However, the advancements in context window size for LLMs offer an alternative approach, raising the question of whether RAG remains necessary for effectively handling external knowledge. Several existing studies provide inconclusive comparisons between RAG and long-context (LC) LLMs, largely due to limitations in the benchmark designs. In this paper, we present LaRA, a novel benchmark specifically designed to rigorously compare RAG and LC LLMs. LaRA encompasses 2326 test cases across four practical QA task categories and three types of naturally occurring long texts. Through systematic evaluation of seven open-source and four proprietary LLMs, we find that the optimal choice between RAG and LC depends on a complex interplay of factors, including the model's parameter size, long-text capabilities, context length, task type, and the characteristics of the retrieved chunks. Our findings provide actionable guidelines for practitioners to effectively leverage both RAG and LC approaches in developing and deploying LLM applications. Our code and dataset is provided at: \href{https://github.com/Alibaba-NLP/LaRA}{\textbf{https://github.com/Alibaba-NLP/LaRA}}.

cs.CL↗

arXiv2025

CHOP: Mobile Operating Assistant with Constrained High-frequency Optimized Subtask Planning

The advancement of visual language models (VLMs) has enhanced mobile device operations, allowing simulated human-like actions to address user requirements. Current VLM-based mobile operating assistants can be structured into three levels: task, subtask, and action. The subtask level, linking high-level goals with low-level executable actions, is crucial for task completion but faces two challenges: ineffective subtasks that lower-level agent cannot execute and inefficient subtasks that fail to contribute to the completion of the higher-level task. These challenges stem from VLM's lack of experience in decomposing subtasks within GUI scenarios in multi-agent architecture. To address these, we propose a new mobile assistant architecture with constrained high-frequency o}ptimized planning (CHOP). Our approach overcomes the VLM's deficiency in GUI scenarios planning by using human-planned subtasks as the basis vector. We evaluate our architecture in both English and Chinese contexts across 20 Apps, demonstrating significant improvements in both effectiveness and efficiency. Our dataset and code is available at https://github.com/Yuqi-Zhou/CHOP

cs.AI↗

arXiv2025

pFedLVM: A Large Vision Model (LVM)-Driven and Latent Feature-Based Personalized Federated Learning Framework in Autonomous Driving

Deep learning-based Autonomous Driving (AD) models often exhibit poor generalization due to data heterogeneity in an ever domain-shifting environment. While Federated Learning (FL) could improve the generalization of an AD model (known as FedAD system), conventional models often struggle with under-fitting as the amount of accumulated training data progressively increases. To address this issue, instead of conventional small models, employing Large Vision Models (LVMs) in FedAD is a viable option for better learning of representations from a vast volume of data. However, implementing LVMs in FedAD introduces three challenges: (I) the extremely high communication overheads associated with transmitting LVMs between participating vehicles and a central server; (II) lack of computing resource to deploy LVMs on each vehicle; (III) the performance drop due to LVM focusing on shared features but overlooking local vehicle characteristics. To overcome these challenges, we propose pFedLVM, a LVM-Driven, Latent Feature-Based Personalized Federated Learning framework. In this approach, the LVM is deployed only on central server, which effectively alleviates the computational burden on individual vehicles. Furthermore, the exchange between central server and vehicles are the learned features rather than the LVM parameters, which significantly reduces communication overhead. In addition, we utilize both shared features from all participating vehicles and individual characteristics from each vehicle to establish a personalized learning mechanism. This enables each vehicle's model to learn features from others while preserving its personalized characteristics, thereby outperforming globally shared models trained in general FL. Extensive experiments demonstrate that pFedLVM outperforms the existing state-of-the-art approaches.

cs.RO↗

arXiv2025

AdaNDV: Adaptive Number of Distinct Value Estimation via Learning to Select and Fuse Estimators

Estimating the Number of Distinct Values (NDV) is fundamental for numerous data management tasks, especially within database applications. However, most existing works primarily focus on introducing new statistical or learned estimators, while identifying the most suitable estimator for a given scenario remains largely unexplored. Therefore, we propose AdaNDV, a learned method designed to adaptively select and fuse existing estimators to address this issue. Specifically, (1) we propose to use learned models to distinguish between overestimated and underestimated estimators and then select appropriate estimators from each category. This strategy provides a complementary perspective by integrating overestimations and underestimations for error correction, thereby improving the accuracy of NDV estimation. (2) To further integrate the estimation results, we introduce a novel fusion approach that employs a learned model to predict the weights of the selected estimators and then applies a weighted sum to merge them. By combining these strategies, the proposed AdaNDV fundamentally distinguishes itself from previous works that directly estimate NDV. Moreover, extensive experiments conducted on real-world datasets, with the number of individual columns being several orders of magnitude larger than in previous studies, demonstrate the superior performance of our method.

cs.DB↗

arXiv2025

DiffRhythm: Blazingly Fast and Embarrassingly Simple End-to-End Full-Length Song Generation with Latent Diffusion

Recent advancements in music generation have garnered significant attention, yet existing approaches face critical limitations. Some current generative models can only synthesize either the vocal track or the accompaniment track. While some models can generate combined vocal and accompaniment, they typically rely on meticulously designed multi-stage cascading architectures and intricate data pipelines, hindering scalability. Additionally, most systems are restricted to generating short musical segments rather than full-length songs. Furthermore, widely used language model-based methods suffer from slow inference speeds. To address these challenges, we propose DiffRhythm, the first latent diffusion-based song generation model capable of synthesizing complete songs with both vocal and accompaniment for durations of up to 4m45s in only ten seconds, maintaining high musicality and intelligibility. Despite its remarkable capabilities, DiffRhythm is designed to be simple and elegant: it eliminates the need for complex data preparation, employs a straightforward model structure, and requires only lyrics and a style prompt during inference. Additionally, its non-autoregressive structure ensures fast inference speeds. This simplicity guarantees the scalability of DiffRhythm. Moreover, we release the complete training code along with the pre-trained model on large-scale data to promote reproducibility and further research.

eess.AS↗

arXiv2025

CLIP-Optimized Multimodal Image Enhancement via ISP-CNN Fusion for Coal Mine IoVT under Uneven Illumination

Clear monitoring images are crucial for the safe operation of coal mine Internet of Video Things (IoVT) systems. However, low illumination and uneven brightness in underground environments significantly degrade image quality, posing challenges for enhancement methods that often rely on difficult-to-obtain paired reference images. Additionally, there is a trade-off between enhancement performance and computational efficiency on edge devices within IoVT systems.To address these issues, we propose a multimodal image enhancement method tailored for coal mine IoVT, utilizing an ISP-CNN fusion architecture optimized for uneven illumination. This two-stage strategy combines global enhancement with detail optimization, effectively improving image quality, especially in poorly lit areas. A CLIP-based multimodal iterative optimization allows for unsupervised training of the enhancement algorithm. By integrating traditional image signal processing (ISP) with convolutional neural networks (CNN), our approach reduces computational complexity while maintaining high performance, making it suitable for real-time deployment on edge devices.Experimental results demonstrate that our method effectively mitigates uneven brightness and enhances key image quality metrics, with PSNR improvements of 2.9%-4.9%, SSIM by 4.3%-11.4%, and VIF by 4.9%-17.8% compared to seven state-of-the-art algorithms. Simulated coal mine monitoring scenarios validate our method's ability to balance performance and computational demands, facilitating real-time enhancement and supporting safer mining operations.

cs.CV↗

arXiv2025

Memory-Free and Parallel Computation for Quantized Spiking Neural Networks

Quantized Spiking Neural Networks (QSNNs) offer superior energy efficiency and are well-suited for deployment on resource-limited edge devices. However, limited bit-width weight and membrane potential result in a notable performance decline. In this study, we first identify a new underlying cause for this decline: the loss of historical information due to the quantized membrane potential. To tackle this issue, we introduce a memory-free quantization method that captures all historical information without directly storing membrane potentials, resulting in better performance with less memory requirements. To further improve the computational efficiency, we propose a parallel training and asynchronous inference framework that greatly increases training speed and energy efficiency. We combine the proposed memory-free quantization and parallel computation methods to develop a high-performance and efficient QSNN, named MFP-QSNN. Extensive experiments show that our MFP-QSNN achieves state-of-the-art performance on various static and neuromorphic image datasets, requiring less memory and faster training speeds. The efficiency and efficacy of the MFP-QSNN highlight its potential for energy-efficient neuromorphic computing.

cs.NE↗

arXiv2025

Towards Accurate Binary Spiking Neural Networks: Learning with Adaptive Gradient Modulation Mechanism

Binary Spiking Neural Networks (BSNNs) inherit the eventdriven paradigm of SNNs, while also adopting the reduced storage burden of binarization techniques. These distinct advantages grant BSNNs lightweight and energy-efficient characteristics, rendering them ideal for deployment on resource-constrained edge devices. However, due to the binary synaptic weights and non-differentiable spike function, effectively training BSNNs remains an open question. In this paper, we conduct an in-depth analysis of the challenge for BSNN learning, namely the frequent weight sign flipping problem. To mitigate this issue, we propose an Adaptive Gradient Modulation Mechanism (AGMM), which is designed to reduce the frequency of weight sign flipping by adaptively adjusting the gradients during the learning process. The proposed AGMM can enable BSNNs to achieve faster convergence speed and higher accuracy, effectively narrowing the gap between BSNNs and their full-precision equivalents. We validate AGMM on both static and neuromorphic datasets, and results indicate that it achieves state-of-the-art results among BSNNs. This work substantially reduces storage demands and enhances SNNs' inherent energy efficiency, making them highly feasible for resource-constrained environments.

cs.CV↗

arXiv2025

Spiking Vision Transformer with Saccadic Attention

The combination of Spiking Neural Networks (SNNs) and Vision Transformers (ViTs) holds potential for achieving both energy efficiency and high performance, particularly suitable for edge vision applications. However, a significant performance gap still exists between SNN-based ViTs and their ANN counterparts. Here, we first analyze why SNN-based ViTs suffer from limited performance and identify a mismatch between the vanilla self-attention mechanism and spatio-temporal spike trains. This mismatch results in degraded spatial relevance and limited temporal interactions. To address these issues, we draw inspiration from biological saccadic attention mechanisms and introduce an innovative Saccadic Spike Self-Attention (SSSA) method. Specifically, in the spatial domain, SSSA employs a novel spike distribution-based method to effectively assess the relevance between Query and Key pairs in SNN-based ViTs. Temporally, SSSA employs a saccadic interaction module that dynamically focuses on selected visual areas at each timestep and significantly enhances whole scene understanding through temporal interactions. Building on the SSSA mechanism, we develop a SNN-based Vision Transformer (SNN-ViT). Extensive experiments across various visual tasks demonstrate that SNN-ViT achieves state-of-the-art performance with linear computational complexity. The effectiveness and efficiency of the SNN-ViT highlight its potential for power-critical edge vision applications.

cs.CV↗