Searcharxiv⌕ Search

arXiv subjects

Lei Fang

Publications and source records attributed to Lei Fang.

At least 37 records · Page 2Linked to original sources
arXiv2025

R1-Searcher++: Incentivizing the Dynamic Knowledge Acquisition of LLMs via Reinforcement Learning

Large Language Models (LLMs) are powerful but prone to hallucinations due to static knowledge. Retrieval-Augmented Generation (RAG) helps by injecting external information, but current methods often are costly, generalize poorly, or ignore the internal knowledge of the model. In this paper, we introduce R1-Searcher++, a novel framework designed to train LLMs to adaptively leverage both internal and external knowledge sources. R1-Searcher++ employs a two-stage training strategy: an initial SFT Cold-start phase for preliminary format learning, followed by RL for Dynamic Knowledge Acquisition. The RL stage uses outcome-supervision to encourage exploration, incorporates a reward mechanism for internal knowledge utilization, and integrates a memorization mechanism to continuously assimilate retrieved information, thereby enriching the model's internal knowledge. By leveraging internal knowledge and external search engine, the model continuously improves its capabilities, enabling efficient retrieval-augmented reasoning. Our experiments demonstrate that R1-Searcher++ outperforms previous RAG and reasoning methods and achieves efficient retrieval. The code is available at https://github.com/RUCAIBox/R1-Searcher-plus.

cs.CL↗
arXiv2025

CAFE: Retrieval Head-based Coarse-to-Fine Information Seeking to Enhance Multi-Document QA Capability

Advancements in Large Language Models (LLMs) have extended their input context length, yet they still struggle with retrieval and reasoning in long-context inputs. Existing methods propose to utilize the prompt strategy and retrieval head to alleviate this limitation. However, they still face challenges in balancing retrieval precision and recall, impacting their efficacy in answering questions. To address this, we introduce $\textbf{CAFE}$, a two-stage coarse-to-fine method to enhance multi-document question-answering capacities. By gradually eliminating the negative impacts of background and distracting documents, CAFE makes the responses more reliant on the evidence documents. Initially, a coarse-grained filtering method leverages retrieval heads to identify and rank relevant documents. Then, a fine-grained steering method guides attention to the most relevant content. Experiments across benchmarks show CAFE outperforms baselines, achieving up to 22.1% and 13.7% SubEM improvement over SFT and RAG methods on the Mistral model, respectively.

cs.CL↗
arXiv2025

Rethinking Invariance in In-context Learning

In-Context Learning (ICL) has emerged as a pivotal capability of auto-regressive large language models, yet it is hindered by a notable sensitivity to the ordering of context examples regardless of their mutual independence. To address this issue, recent studies have introduced several variant algorithms of ICL that achieve permutation invariance. However, many of these do not exhibit comparable performance with the standard auto-regressive ICL algorithm. In this work, we identify two crucial elements in the design of an invariant ICL algorithm: information non-leakage and context interdependence, which are not simultaneously achieved by any of the existing methods. These investigations lead us to the proposed Invariant ICL (InvICL), a methodology designed to achieve invariance in ICL while ensuring the two properties. Empirically, our findings reveal that InvICL surpasses previous models, both invariant and non-invariant, in most benchmark datasets, showcasing superior generalization capabilities across varying input lengths. Code is available at https://github.com/PKU-ML/InvICL.

cs.CL↗
arXiv2025

R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning

Existing Large Reasoning Models (LRMs) have shown the potential of reinforcement learning (RL) to enhance the complex reasoning capabilities of Large Language Models~(LLMs). While they achieve remarkable performance on challenging tasks such as mathematics and coding, they often rely on their internal knowledge to solve problems, which can be inadequate for time-sensitive or knowledge-intensive questions, leading to inaccuracies and hallucinations. To address this, we propose \textbf{R1-Searcher}, a novel two-stage outcome-based RL approach designed to enhance the search capabilities of LLMs. This method allows LLMs to autonomously invoke external search systems to access additional knowledge during the reasoning process. Our framework relies exclusively on RL, without requiring process rewards or distillation for a cold start. % effectively generalizing to out-of-domain datasets and supporting both Base and Instruct models. Our experiments demonstrate that our method significantly outperforms previous strong RAG methods, even when compared to the closed-source GPT-4o-mini.

cs.AI↗
arXiv2025

An Empirical Study on Eliciting and Improving R1-like Reasoning Models

In this report, we present the third technical report on the development of slow-thinking models as part of the STILL project. As the technical pathway becomes clearer, scaling RL training has become a central technique for implementing such reasoning models. We systematically experiment with and document the effects of various factors influencing RL training, conducting experiments on both base models and fine-tuned models. Specifically, we demonstrate that our RL training approach consistently improves the Qwen2.5-32B base models, enhancing both response length and test accuracy. Furthermore, we show that even when a model like DeepSeek-R1-Distill-Qwen-1.5B has already achieved a high performance level, it can be further refined through RL training, reaching an accuracy of 39.33% on AIME 2024. Beyond RL training, we also explore the use of tool manipulation, finding that it significantly boosts the reasoning performance of large reasoning models. This approach achieves a remarkable accuracy of 86.67% with greedy search on AIME 2024, underscoring its effectiveness in enhancing model capabilities. We release our resources at the STILL project website: https://github.com/RUCAIBox/Slow_Thinking_with_LLMs.

cs.CL↗
arXiv2025

Accurate Forgetting for Heterogeneous Federated Continual Learning

Recent years have witnessed a burgeoning interest in federated learning (FL). However, the contexts in which clients engage in sequential learning remain under-explored. Bridging FL and continual learning (CL) gives rise to a challenging practical problem: federated continual learning (FCL). Existing research in FCL primarily focuses on mitigating the catastrophic forgetting issue of continual learning while collaborating with other clients. We argue that the forgetting phenomena are not invariably detrimental. In this paper, we consider a more practical and challenging FCL setting characterized by potentially unrelated or even antagonistic data/tasks across different clients. In the FL scenario, statistical heterogeneity and data noise among clients may exhibit spurious correlations which result in biased feature learning. While existing CL strategies focus on a complete utilization of previous knowledge, we found that forgetting biased information is beneficial in our study. Therefore, we propose a new concept accurate forgetting (AF) and develop a novel generative-replay method~\method~which selectively utilizes previous knowledge in federated networks. We employ a probabilistic framework based on a normalizing flow model to quantify the credibility of previous knowledge. Comprehensive experiments affirm the superiority of our method over baselines.

cs.LG↗
arXiv2024

USTC-KXDIGIT System Description for ASVspoof5 Challenge

This paper describes the USTC-KXDIGIT system submitted to the ASVspoof5 Challenge for Track 1 (speech deepfake detection) and Track 2 (spoofing-robust automatic speaker verification, SASV). Track 1 showcases a diverse range of technical qualities from potential processing algorithms and includes both open and closed conditions. For these conditions, our system consists of a cascade of a frontend feature extractor and a back-end classifier. We focus on extensive embedding engineering and enhancing the generalization of the back-end classifier model. Specifically, the embedding engineering is based on hand-crafted features and speech representations from a self-supervised model, used for closed and open conditions, respectively. To detect spoof attacks under various adversarial conditions, we trained multiple systems on an augmented training set. Additionally, we used voice conversion technology to synthesize fake audio from genuine audio in the training set to enrich the synthesis algorithms. To leverage the complementary information learned by different model architectures, we employed activation ensemble and fused scores from different systems to obtain the final decision score for spoof detection. During the evaluation phase, the proposed methods achieved 0.3948 minDCF and 14.33% EER in the close condition, and 0.0750 minDCF and 2.59% EER in the open condition, demonstrating the robustness of our submitted systems under adversarial conditions. In Track 2, we continued using the CM system from Track 1 and fused it with a CNN-based ASV system. This approach achieved 0.2814 min-aDCF in the closed condition and 0.0756 min-aDCF in the open condition, showcasing superior performance in the SASV system.

cs.SD↗
arXiv2024

Artwork Protection Against Neural Style Transfer Using Locally Adaptive Adversarial Color Attack

Neural style transfer (NST) generates new images by combining the style of one image with the content of another. However, unauthorized NST can exploit artwork, raising concerns about artists' rights and motivating the development of proactive protection methods. We propose Locally Adaptive Adversarial Color Attack (LAACA), empowering artists to protect their artwork from unauthorized style transfer by processing before public release. By delving into the intricacies of human visual perception and the role of different frequency components, our method strategically introduces frequency-adaptive perturbations in the image. These perturbations significantly degrade the generation quality of NST while maintaining an acceptable level of visual change in the original image, ensuring that potential infringers are discouraged from using the protected artworks, because of its bad NST generation quality. Additionally, existing metrics often overlook the importance of color fidelity in evaluating color-mattered tasks, such as the quality of NST-generated images, which is crucial in the context of artistic works. To comprehensively assess the color-mattered tasks, we propose the Adversarial Color Distance Metric (ACDM), designed to quantify the color difference of images pre- and post-manipulations. Experimental results confirm that attacking NST using LAACA results in visually inferior style transfer, and the ACDM can efficiently measure color-mattered tasks. By providing artists with a tool to safeguard their intellectual property, our work relieves the socio-technical challenges posed by the misuse of NST in the art community.

cs.CV↗
arXiv2024

Diffusion Transformer Model With Compact Prior for Low-dose PET Reconstruction

Positron emission tomography (PET) is an advanced medical imaging technique that plays a crucial role in non-invasive clinical diagnosis. However, while reducing radiation exposure through low-dose PET scans is beneficial for patient safety, it often results in insufficient statistical data. This scarcity of data poses significant challenges for accurately reconstructing high-quality images, which are essential for reliable diagnostic outcomes. In this research, we propose a diffusion transformer model (DTM) guided by joint compact prior (JCP) to enhance the reconstruction quality of low-dose PET imaging. In light of current research findings, we present a pioneering PET reconstruction model that integrates diffusion and transformer models for joint optimization. This model combines the powerful distribution mapping abilities of diffusion models with the capacity of transformers to capture long-range dependencies, offering significant advantages for low-dose PET reconstruction. Additionally, the incorporation of the lesion refining block and penalized weighted least squares (PWLS) enhance the recovery capability of lesion regions and preserves detail information, solving blurring problems in lesion areas and texture details of most deep learning frameworks. Experimental results demonstrate the effectiveness of DTM in enhancing image quality and preserving critical clinical information for low-dose PET scans. Our approach not only reduces radiation exposure risks but also provides a more reliable PET imaging tool for early disease detection and patient management.

cs.CV↗
arXiv2024

Integration of multiview microbiome data for deciphering microbiome-metabolome-disease pathways

The intricate interplay between host organisms and their gut microbiota has catalyzed research into the microbiome's role in disease, shedding light on novel aspects of disease pathogenesis. However, the mechanisms through which the microbiome exerts its influence on disease remain largely unclear. In this study, we first introduce a structural equation model to delineate the pathways connecting the microbiome, metabolome, and disease processes, utilizing a target multiview microbiome data. To mitigate the challenges posed by hidden confounders, we further propose an integrative approach that incorporates data from an external microbiome cohort. This method also supports the identification of disease-specific and microbiome-associated metabolites that are missing in the target cohort. We provide theoretical underpinnings for the estimations derived from our integrative approach, demonstrating estimation consistency and asymptotic normality. The effectiveness of our methodologies is validated through comprehensive simulation studies and an empirical application to inflammatory bowel disease, highlighting their potential to unravel the complex relationships between the microbiome, metabolome, and disease.

stat.ME↗
arXiv2024

A White-Box False Positive Adversarial Attack Method on Contrastive Loss Based Offline Handwritten Signature Verification Models

In this paper, we tackle the challenge of white-box false positive adversarial attacks on contrastive loss based offline handwritten signature verification models. We propose a novel attack method that treats the attack as a style transfer between closely related but distinct writing styles. To guide the generation of deceptive images, we introduce two new loss functions that enhance the attack success rate by perturbing the Euclidean distance between the embedding vectors of the original and synthesized samples, while ensuring minimal perturbations by reducing the difference between the generated image and the original image. Our method demonstrates state-of-the-art performance in white-box attacks on contrastive loss based offline handwritten signature verification models, as evidenced by our experiments. The key contributions of this paper include a novel false positive attack method, two new loss functions, effective style transfer in handwriting styles, and superior performance in white-box false positive attacks compared to other white-box attack methods.

cs.CV↗
arXiv2023

High Fidelity Human Trajectory Tracking Based on Surveillance Camera Data

Human crowds exhibit a wide range of interesting patterns, and measuring them is of great interest in areas ranging from psychology and social science to civil engineering. While \textit{in situ} measurements of human crowd patterns require large amounts of time and labor to obtain, human crowd experiments may result in statistics different from those that would emerge with a naturally emerging crowd. Here we present a simple, broadly applicable, highly accurate human crowd tracking technique to extract high-fidelity kinematic information from widely available surveillance camera videos. With the proposed technique, researchers can access scientific crowd data on a scale that is orders of magnitude larger than before. In addition to being able to measure an individual's time-resolved position and velocity, our technique also offers high validity time-resolved acceleration and step frequency, and step length. We demonstrate the applicability of our technique by applying it to surveillance camera videos in Tokyo Shinjuku streamed on YouTube and exploiting its high fidelity to expose the hidden contribution of walking speed variance at the crossroad. The high fidelity and simplicity of this powerful technique open up the way to utilize the large volume of existing surveillance camera data around the world for scientific studies.

nlin.AO↗
arXiv2023

Interaction between swarming active matter and flow: the impact on Lagrangian coherent structures

In recent years, research topics concerning active matter have attracted interest from diverse communities. It has been suggested that active matter-as represented by zooplankton-has potential in ocean mixing due to its intrinsic mobility and the sheer amount of biomass. However, prior investigations have predominantly overlooked the influence of external background flow, despite the ubiquity of flow driven by various sources in nature. The interaction between active matter and external flow structures has long been neglected. Here, we conducted experiments using a typical centimeter swimmer, \textit{A. salina}, and an electromagnetically driven quasi-two-dimensional flow to study the interaction between active matter and flow. We focused on the impact of swarming active matter on hyperbolic Lagrangian coherent structures (LCSs) that mark the most straining regions in the flow. We illustrated that the impact of active matter on LCSs was much more significant compared to localized random noise with similar energy input. In addition, we revealed that the perturbation generated by active matter could couple with the background flow and further deform the LCSs. In addition, we also revealed that the rotational elliptical region of the flow was much more susceptible to active matter perturbation. We further described how the influence of active matter changed with their number densities and background flow intensities. We revealed that the LCSs could be decently altered even at a small number density of active matter. We aim to provide valuable insights and draw attention to the problem regarding the interaction between active matter and external flow structures.

physics.flu-dyn↗
arXiv2023

Retro-BLEU: Quantifying Chemical Plausibility of Retrosynthesis Routes through Reaction Template Sequence Analysis

Computer-assisted methods have emerged as valuable tools for retrosynthesis analysis. However, quantifying the plausibility of generated retrosynthesis routes remains a challenging task. We introduce Retro-BLEU, a statistical metric adapted from the well-established BLEU score in machine translation, to evaluate the plausibility of retrosynthesis routes based on reaction template sequences analysis. We demonstrate the effectiveness of Retro-BLEU by applying it to a diverse set of retrosynthesis routes generated by state-of-the-art algorithms and compare the performance with other evaluation metrics. The results show that Retro-BLEU is capable of differentiating between plausible and implausible routes. Furthermore, we provide insights into the strengths and weaknesses of Retro-BLEU, paving the way for future developments and improvements in this field.

cs.LG↗
arXiv2023

Biologically generated turbulent energy cascade in shear flow depends on tensor geometry

It has been proposed that biologically generated turbulence plays an important role in material transport and ocean mixing. Both experimental and numerical studies have reported evidence of the non-negligible mixing by moderate Reynolds number swimmers in quiescent water, such as zooplankton, especially at aggregation scales. However, the interaction between biologically generated agitation and the background flow as a key factor in biologically generated turbulence that could reshape our previous knowledge of biologically generated turbulence, has long been ignored. Here we show that the geometry between the biologically generated agitation and the background hydrodynamic shear can determine both the intensity and direction of biologically generated turbulent energy cascade. Measuring the migration of a centimeter-scale swimmer-as represented by the brine shrimp \textit{Artemia salina}-in a shear flow and verifying through an analogue experiment with an artificial jet revealed that different geometries between the biologically generated agitation and the background shear can result in spectral energy transferring toward larger or smaller scales, which consequently intensifies or attenuates the large scale hydrodynamic shear. Our results suggest that the long ignored geometry between the biologically generated agitation and the background flow field is an important factor that should be taken into consideration in future studies of biologically generated turbulence.

physics.flu-dyn↗
arXiv2023

High-dimensional Variable Screening via Conditional Martingale Difference Divergence

Variable screening has been a useful research area that deals with ultrahigh-dimensional data. When there exist both marginally and jointly dependent predictors to the response, existing methods such as conditional screening or iterative screening often suffer from instability against the selection of the conditional set or the computational burden, respectively. In this article, we propose a new independence measure, named conditional martingale difference divergence (CMDH), that can be treated as either a conditional or a marginal independence measure. Under regularity conditions, we show that the sure screening property of CMDH holds for both marginally and jointly active variables. Based on this measure, we propose a kernel-based model-free variable screening method, which is efficient, flexible, and stable against high correlation among predictors and heterogeneity of the response. In addition, we provide a data-driven method to select the conditional set. In simulations and real data applications, we demonstrate the superior performance of the proposed method.

stat.ME↗
arXiv2022

Mixing patterns in graphs with higher-order structure

In this paper we examine the percolation properties of higher-order networks that have non-trivial clustering and subgraph-based assortative mixing (the tendency of vertices to connect to other vertices based on subgraph joint degree). Our analytical method is based on generating functions. We also propose a Monte Carlo graph generation algorithm to draw random networks from the ensemble of graphs with fixed statistics. We use our model to understand the effect that network microstructure has, through the arrangement of clustering, on the global properties. Finally, we use an edge disjoint clique cover to represent empirical networks using our formulation, finding the resultant model offers a significant improvement over edge-based theory.

physics.soc-ph↗