arXiv subjects

Lei Huang

Publications and source records attributed to Lei Huang.

At least 109 records · Page 6Linked to original sources

arXiv2025

CC-Tuning: A Cross-Lingual Connection Mechanism for Improving Joint Multilingual Supervised Fine-Tuning

Current large language models (LLMs) often exhibit imbalanced multilingual capabilities due to their English-centric training corpora. To address this, existing fine-tuning approaches operating at the data-level (e.g., through data augmentation or distillation) typically introduce implicit cross-lingual alignment, overlooking the potential for more profound, latent-level cross-lingual interactions. In this work, we propose CC-Tuning, a novel multilingual fine-tuning paradigm that explicitly establishes a cross-lingual connection mechanism at the latent level. During training, CC-Tuning fuses the feed forward activations from both English and non-English inputs, enabling the model to benefit from both linguistic resources. This process is facilitated with a trainable Decision Maker that identifies beneficial activations. Furthermore, during inference, a Transform Matrix is utilized to simulate the cross-lingual connection under monolingual setting through representation transformation. Our experiments on six benchmarks covering 22 languages show that CC-Tuning outperforms vanilla SFT and offers a strong latent-level alternative to data-level augmentation methods. Further analysis also highlights the practicality of CC-Tuning and the potential of latent-level cross-lingual interactions in advancing the multilingual performance of LLMs.

cs.CL↗

arXiv2025

Unveiling Entity-Level Unlearning for Large Language Models: A Comprehensive Analysis

Large language model unlearning has garnered increasing attention due to its potential to address security and privacy concerns, leading to extensive research in the field. However, much of this research has concentrated on instance-level unlearning, specifically targeting the removal of predefined instances containing sensitive content. This focus has left a significant gap in the exploration of full entity-level unlearning, which is critical in real-world scenarios such as copyright protection. To this end, we propose a novel task of Entity-level unlearning, which aims to erase entity-related knowledge from the target model completely. To thoroughly investigate this task, we systematically evaluate trending unlearning algorithms, revealing that current methods struggle to achieve effective entity-level unlearning. Then, we further explore the factors that influence the performance of the unlearning algorithms, identifying that knowledge coverage and the size of the forget set play pivotal roles. Notably, our analysis also uncovers that entities introduced through fine-tuning are more vulnerable to unlearning than pre-trained entities. These findings collectively offer valuable insights for advancing entity-level unlearning for LLMs.

cs.CL↗

arXiv2025

Clustering Properties of Self-Supervised Learning

Self-supervised learning (SSL) methods via joint embedding architectures have proven remarkably effective at capturing semantically rich representations with strong clustering properties, magically in the absence of label supervision. Despite this, few of them have explored leveraging these untapped properties to improve themselves. In this paper, we provide an evidence through various metrics that the encoder's output $encoding$ exhibits superior and more stable clustering properties compared to other components. Building on this insight, we propose a novel positive-feedback SSL method, termed Representation Self-Assignment (ReSA), which leverages the model's clustering properties to promote learning in a self-guided manner. Extensive experiments on standard SSL benchmarks reveal that models pretrained with ReSA outperform other state-of-the-art SSL methods by a significant margin. Finally, we analyze how ReSA facilitates better clustering properties, demonstrating that it effectively enhances clustering performance at both fine-grained and coarse-grained levels, shaping representations that are inherently more structured and semantically meaningful.

cs.LG↗

arXiv2025

Metasurfaces in Adaptive Optics: A New Opportunity in Optical Wavefront Sensing

Over the past fifty years, wavefront sensing technology has continuously evolved from basic techniques to high-precision systems, serving as a core methodology in adaptive optics (AO). Beyond traditional wavefront retrieval methods based on spot displacement, direct phase retrieval techniques with greater accuracy have emerged, jointly driving advancements in wavefront sensing precision. This evolution is fueled by increasing demands for accuracy, which have prompted iterative upgrades in system architectures and algorithms. Recently, breakthroughs in metasurface technology have opened new possibilities for wavefront sensing. By utilizing subwavelength microstructures, metasurfaces enable multi-dimensional control over the phase, amplitude, and polarization of light fields. Their high degree of design flexibility presents transformative opportunities for advancing wavefront sensing capabilities. This review examines the fundamental principles of wavefront sensing and the development of key enabling devices, highlighting how metasurface technology is reshaping traditional paradigms. We discuss recent research progress and emerging innovations, aiming to establish a theoretical framework for next-generation wavefront sensing technologies. Ultimately, we hope this review provides technical insights for applications in astronomical observation, biological microscopy, laser engineering, and beyond.

physics.optics↗

arXiv2025

Search-Based Interaction For Conversation Recommendation via Generative Reward Model Based Simulated User

Conversational recommendation systems (CRSs) use multi-turn interaction to capture user preferences and provide personalized recommendations. A fundamental challenge in CRSs lies in effectively understanding user preferences from conversations. User preferences can be multifaceted and complex, posing significant challenges for accurate recommendations even with access to abundant external knowledge. While interaction with users can clarify their true preferences, frequent user involvement can lead to a degraded user experience. To address this problem, we propose a generative reward model based simulated user, named GRSU, for automatic interaction with CRSs. The simulated user provides feedback to the items recommended by CRSs, enabling them to better capture intricate user preferences through multi-turn interaction. Inspired by generative reward models, we design two types of feedback actions for the simulated user: i.e., generative item scoring, which offers coarse-grained feedback, and attribute-based item critique, which provides fine-grained feedback. To ensure seamless integration, these feedback actions are unified into an instruction-based format, allowing the development of a unified simulated user via instruction tuning on synthesized data. With this simulated user, automatic multi-turn interaction with CRSs can be effectively conducted. Furthermore, to strike a balance between effectiveness and efficiency, we draw inspiration from the paradigm of reward-guided search in complex reasoning tasks and employ beam search for the interaction process. On top of this, we propose an efficient candidate ranking method to improve the recommendation results derived from interaction. Extensive experiments on public datasets demonstrate the effectiveness, efficiency, and transferability of our approach.

cs.IR↗

arXiv2025

Normalization of Quaternionic Polynomials in Coordinate-Free Quaternionic Variables in Conjugate-Alternating Order

Quaternionic polynomials occur naturally in applications of quaternions in science and engineering, and normalization of quaternionic polynomials is a basic manipulation. Once a Groebner basis is certified for the defining ideal I of the quaternionic polynomial algebra, the normal form of a quaternionic polynomial can be computed by routine top reduction with respect to the Groebner basis. In the literature, a Groebner basis under the conjugate-alternating order of quaternionic variables was conjectured for I in 2013, but no readable and convincing proof was found. In this paper, we present the first readable certification of the conjectured Groebner basis. The certification is based on several novel techniques for reduction in free associative algebras, which enables to not only make reduction to S-polynomials more efficiently, but also reduce the number of S-polynomials needed for the certification.

cs.SC↗

arXiv2025

TinyLLaVA-Video-R1: Towards Smaller LMMs for Video Reasoning

Recently, improving the reasoning ability of large multimodal models (LMMs) through reinforcement learning has made great progress. However, most existing works are based on highly reasoning-intensive datasets such as mathematics and code, and researchers generally choose large-scale models as the foundation. We argue that exploring small-scale models' reasoning capabilities remains valuable for researchers with limited computational resources. Moreover, enabling models to explain their reasoning processes on general question-answering datasets is equally meaningful. Therefore, we present the small-scale video reasoning model TinyLLaVA-Video-R1. Based on TinyLLaVA-Video, a traceably trained video understanding model with no more than 4B parameters, it not only demonstrates significantly improved reasoning and thinking capabilities after using reinforcement learning on general Video-QA datasets, but also exhibits the emergent characteristic of "aha moments". Furthermore, we share a series of experimental findings, aiming to provide practical insights for future exploration of video reasoning (thinking) abilities in small-scale models. It is available at https://github.com/ZhangXJ199/TinyLLaVA-Video-R1.

cs.CV↗

arXiv2025

Optimization over the weakly Pareto set and multi-task learning

We study the optimization problem over the weakly Pareto set of a convex multiobjective optimization problem given by polynomial functions. Using Lagrange multiplier expressions and the weight vector, we give three types of representations for the weakly Pareto set. Using these representations, we reformulate the optimization problem over the weakly Pareto set as a polynomial optimization problem. We then apply the Moment--SOS hierarchy to solve it and analyze its convergence properties under certain conditions. Numerical experiments are provided to demonstrate the effectiveness of our methods. Applications in multi-task learning are also presented.

math.OC↗

arXiv2025

Verification of Bit-Flip Attacks against Quantized Neural Networks

In the rapidly evolving landscape of neural network security, the resilience of neural networks against bit-flip attacks (i.e., an attacker maliciously flips an extremely small amount of bits within its parameter storage memory system to induce harmful behavior), has emerged as a relevant area of research. Existing studies suggest that quantization may serve as a viable defense against such attacks. Recognizing the documented susceptibility of real-valued neural networks to such attacks and the comparative robustness of quantized neural networks (QNNs), in this work, we introduce BFAVerifier, the first verification framework designed to formally verify the absence of bit-flip attacks or to identify all vulnerable parameters in a sound and rigorous manner. BFAVerifier comprises two integral components: an abstraction-based method and an MILP-based method. Specifically, we first conduct a reachability analysis with respect to symbolic parameters that represent the potential bit-flip attacks, based on a novel abstract domain with a sound guarantee. If the reachability analysis fails to prove the resilience of such attacks, then we encode this verification problem into an equivalent MILP problem which can be solved by off-the-shelf solvers. Therefore, BFAVerifier is sound, complete, and reasonably efficient. We conduct extensive experiments, which demonstrate its effectiveness and efficiency across various network architectures, quantization bit-widths, and adversary capabilities.

cs.CR↗

arXiv2025

SessionRec: Next Session Prediction Paradigm For Generative Sequential Recommendation

We introduce SessionRec, a novel next-session prediction paradigm (NSPP) for generative sequential recommendation, addressing the fundamental misalignment between conventional next-item prediction paradigm (NIPP) and real-world recommendation scenarios. Unlike NIPP's item-level autoregressive generation that contradicts actual session-based user interactions, our framework introduces a session-aware representation learning through hierarchical sequence aggregation (intra/inter-session), reducing attention computation complexity while enabling implicit modeling of massive negative interactions, and a session-based prediction objective that better captures users' diverse interests through multi-item recommendation in next sessions. Moreover, we found that incorporating a rank loss for items within the session under the next session prediction paradigm can significantly improve the ranking effectiveness of generative sequence recommendation models. We also verified that SessionRec exhibits clear power-law scaling laws similar to those observed in LLMs. Extensive experiments conducted on public datasets and online A/B test in Meituan App demonstrate the effectiveness of SessionRec. The proposed paradigm establishes new foundations for developing industrial-scale generative recommendation systems through its model-agnostic architecture and computational efficiency.

cs.IR↗

arXiv2025

Improving Contextual Faithfulness of Large Language Models via Retrieval Heads-Induced Optimization

Ensuring contextual faithfulness in retrieval-augmented large language models (LLMs) is crucial for building trustworthy information-seeking systems, particularly in long-form question-answering (LFQA) scenarios. In this work, we identify a salient correlation between LFQA faithfulness and retrieval heads, a set of attention heads responsible for retrieving contextual information. Leveraging this insight, we propose RHIO, a framework designed to teach LLMs to explicitly discriminate between faithful and unfaithful generations. RHIO first augments unfaithful samples that simulate realistic model-intrinsic errors by selectively masking retrieval heads. Then, these samples are incorporated into joint training, enabling the model to distinguish unfaithful outputs from faithful ones conditioned on control tokens. Furthermore, these control tokens are leveraged to self-induce contrastive outputs, amplifying their difference through contrastive decoding. Additionally, to facilitate the evaluation of contextual faithfulness, we also introduce GroundBench, a comprehensive benchmark compiled from five existing LFQA datasets. Extensive experimental results on GroundBench demonstrate that RHIO significantly improves faithfulness, even outperforming GPT-4o.

cs.CL↗

arXiv2025

Finite convergence of the Moment-SOS hierarchy for polynomial matrix optimization

This paper studies the matrix Moment-SOS hierarchy for solving polynomial matrix optimization. Our first result is to show the finite convergence of this hierarchy, if the nondegeneracy condition, strict complementarity condition and second order sufficient condition hold at every minimizer, under the Archimedean property. A useful criterion for detecting the finite convergence is the flat truncation. Our second result is to show that every minimizer of the moment relaxation must have a flat truncation when the relaxation order is big enough, under the above mentioned optimality conditions. These results give connections between nonlinear semidefinite optimization theory and Moment-SOS methods for solving polynomial matrix optimization.

math.OC↗

arXiv2025

Joint Antenna Selection and Beamforming Design for Active RIS-aided ISAC Systems

Active reconfigurable intelligent surface (A-RIS) aided integrated sensing and communications (ISAC) system has been considered as a promising paradigm to improve spectrum efficiency. However, massive energy-hungry radio frequency (RF) chains hinder its large-scale deployment. To address this issue, an A-RIS-aided ISAC system with antenna selection (AS) is proposed in this work, where a target is sensed while multiple communication users are served with specifically selected antennas. Specifically, a cuckoo search-based scheme is first utilized to select the antennas associated with high-gain channels. Subsequently, with the properly selected antennas, the weighted sum-rate (WSR) of the system is optimized under the condition of radar probing power level, power budget for the A-RIS and transmitter. To solve the highly non-convex optimization problem, we develop an efficient algorithm based on weighted minimum mean square error (WMMSE) and fractional programming (FP). Simulation results show that the proposed AS scheme and the algorithm are effective, which reduce the number of RF chains without significant performance degradation.

cs.IT↗

arXiv2025

SLAM: Towards Efficient Multilingual Reasoning via Selective Language Alignment

Despite the significant improvements achieved by large language models (LLMs) in English reasoning tasks, these models continue to struggle with multilingual reasoning. Recent studies leverage a full-parameter and two-stage training paradigm to teach models to first understand non-English questions and then reason. However, this method suffers from both substantial computational resource computing and catastrophic forgetting. The fundamental cause is that, with the primary goal of enhancing multilingual comprehension, an excessive number of irrelevant layers and parameters are tuned during the first stage. Given our findings that the representation learning of languages is merely conducted in lower-level layers, we propose an efficient multilingual reasoning alignment approach that precisely identifies and fine-tunes the layers responsible for handling multilingualism. Experimental results show that our method, SLAM, only tunes 6 layers' feed-forward sub-layers including 6.5-8% of all parameters within 7B and 13B LLMs, achieving superior average performance than all strong baselines across 10 languages. Meanwhile, SLAM only involves one training stage, reducing training time by 4.1-11.9 compared to the two-stage method.

cs.CL↗

arXiv2024

On the complexity of matrix Putinar's Positivstellensatz

This paper studies the complexity of matrix Putinar's Positivstellens{ä}tz on the semialgebraic set that is given by the polynomial matrix inequality. \rev{When the quadratic module generated by the constrained polynomial matrix is Archimedean}, we prove a polynomial bound on the degrees of terms appearing in the representation of matrix Putinar's Positivstellens{ä}tz. Estimates on the exponent and constant are given. As a byproduct, a polynomial bound on the convergence rate of matrix sum-of-squares relaxations is obtained, which resolves an open question raised by Dinh and Pham. When the constraining set is unbounded, we also prove a similar bound for the matrix version of Putinar--Vasilescu's Positivstellens{ä}tz by exploiting homogenization techniques.

math.OC↗

arXiv2024

Length Controlled Generation for Black-box LLMs

Large language models (LLMs) have demonstrated impressive instruction following capabilities, while still struggling to accurately manage the length of the generated text, which is a fundamental requirement in many real-world applications. Existing length control methods involve fine-tuning the parameters of LLMs, which is inefficient and suboptimal for practical use. In this paper, we propose a novel iterative sampling framework for text length control, integrating the Metropolis-Hastings algorithm with an importance sampling acceleration strategy. This framework efficiently and reliably regulates LLMs to generate length-constrained text without modifying the underlying parameters, thereby preserving the original capabilities of LLMs. Experimental results demonstrate that our framework achieves almost 100\% success rates of length control on Llama3.1 for tasks such as length-controlled abstractive summarization and length-constrained instruction following, with minimal additional computational overhead. This also highlights the significant potential of our method for precise length control across a broader range of applications, without compromising the versatility of LLMs.

cs.CL↗

arXiv2024

Bench-CoE: a Framework for Collaboration of Experts from Benchmark

Large Language Models (LLMs) are key technologies driving intelligent systems to handle multiple tasks. To meet the demands of various tasks, an increasing number of LLMs-driven experts with diverse capabilities have been developed, accompanied by corresponding benchmarks to evaluate their performance. This paper proposes the Bench-CoE framework, which enables Collaboration of Experts (CoE) by effectively leveraging benchmark evaluations to achieve optimal performance across various tasks. Bench-CoE includes a set of expert models, a router for assigning tasks to corresponding experts, and a benchmark dataset for training the router. Moreover, we formulate Query-Level and Subject-Level approaches based on our framework, and analyze the merits and drawbacks of these two approaches. Finally, we conduct a series of experiments with vary data distributions on both language and multimodal tasks to validate that our proposed Bench-CoE outperforms any single model in terms of overall performance. We hope this method serves as a baseline for further research in this area. The code is available at \url{https://github.com/ZhangXJ199/Bench-CoE}.

cs.AI↗

arXiv2024

A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions

The emergence of large language models (LLMs) has marked a significant breakthrough in natural language processing (NLP), fueling a paradigm shift in information acquisition. Nevertheless, LLMs are prone to hallucination, generating plausible yet nonfactual content. This phenomenon raises significant concerns over the reliability of LLMs in real-world information retrieval (IR) systems and has attracted intensive research to detect and mitigate such hallucinations. Given the open-ended general-purpose attributes inherent to LLMs, LLM hallucinations present distinct challenges that diverge from prior task-specific models. This divergence highlights the urgency for a nuanced understanding and comprehensive overview of recent advances in LLM hallucinations. In this survey, we begin with an innovative taxonomy of hallucination in the era of LLM and then delve into the factors contributing to hallucinations. Subsequently, we present a thorough overview of hallucination detection methods and benchmarks. Our discussion then transfers to representative methodologies for mitigating LLM hallucinations. Additionally, we delve into the current limitations faced by retrieval-augmented LLMs in combating hallucinations, offering insights for developing more robust IR systems. Finally, we highlight the promising research directions on LLM hallucinations, including hallucination in large vision-language models and understanding of knowledge boundaries in LLM hallucinations.

cs.CL↗