Yuan Wu

Yupeng Chang, Xu Wang, Jindong Wang et al. · cmu, pku

Large language models (LLMs) are gaining increasing popularity in both academia and industry, owing to their unprecedented performance in various applications. As LLMs continue to play a vital role in both research and daily use, their evaluation becomes increasingly critical, not only at the task level, but also at the society level for better understanding of their potential risks. Over the past years, significant efforts have been made to examine LLMs from various perspectives. This paper presents a comprehensive review of these evaluation methods for LLMs, focusing on three key dimensions: what to evaluate, where to evaluate, and how to evaluate. Firstly, we provide an overview from the perspective of evaluation tasks, encompassing general natural language processing tasks, reasoning, medical usage, ethics, educations, natural and social sciences, agent applications, and other areas. Secondly, we answer the `where' and `how' questions by diving into the evaluation methods and benchmarks, which serve as crucial components in assessing performance of LLMs. Then, we summarize the success and failure cases of LLMs in different tasks. Finally, we shed light on several future challenges that lie ahead in LLMs evaluation. Our aim is to offer invaluable insights to researchers in the realm of LLMs evaluation, thereby aiding the development of more proficient LLMs. Our key point is that evaluation should be treated as an essential discipline to better assist the development of LLMs. We consistently maintain the related open-source materials at: https://github.com/MLGroupJLU/LLM-eval-survey.

Huangchen Xu, Yuan Wu, Yi Chang

Multimodal Large Language Models (MLLMs) are increasingly used for video understanding, yet their reliability under multi-video inputs remains poorly understood. We study positional bias in multi-video summarization, where the quality of a per-video summary can change with the video's input slot even when the underlying content is unchanged. We construct a benchmark from ActivityNet and News videos, covering Cooking, Domestic, Leisure, and News settings with two- and four-video inputs. We evaluate nine open-source and proprietary MLLMs and measure position effects with three complementary metrics: Coverage, Directional Positional Bias (DPB), and Middle-Edge Gap (MEG). Our results show that positional effects are domain- and model-dependent: signed directional bias can be small even when middle positions underperform, and increasing visual or generation budget does not uniformly remove the imbalance. We further analyze prompt-level mitigation methods. Together, the results show that multi-video summarization remains sensitive to input protocol and position, motivating more robust order-invariant multimodal systems.

Xumin Huang, Peichun Li, Hongyang Du et al.

Artificial intelligence generated content (AIGC) has emerged as a promising technology to improve the efficiency, quality, diversity and flexibility of the content creation process by adopting a variety of generative AI models. Deploying AIGC services in wireless networks has been expected to enhance the user experience. However, the existing AIGC service provision suffers from several limitations, e.g., the centralized training in the pre-training, fine-tuning and inference processes, especially their implementations in wireless networks with privacy preservation. Federated learning (FL), as a collaborative learning framework where the model training is distributed to cooperative data owners without the need for data sharing, can be leveraged to simultaneously improve learning efficiency and achieve privacy protection for AIGC. To this end, we present FL-based techniques for empowering AIGC, and aim to enable users to generate diverse, personalized, and high-quality content. Furthermore, we conduct a case study of FL-aided AIGC fine-tuning by using the state-of-the-art AIGC model, i.e., stable diffusion model. Numerical results show that our scheme achieves advantages in effectively reducing the communication cost and training latency and privacy protection. Finally, we highlight several major research directions and open issues for the convergence of FL and AIGC.

Huangchen Xu, Yuan Wu, Yi Chang

Multimodal large language models have made rapid progress in video understanding, yet existing benchmarks largely rely on simple prompts and provide limited evidence about whether models can satisfy explicit output constraints. We introduce VCIFBench, a benchmark for evaluating complex instruction following in video understanding. VCIFBench constructs constraint-rich instructions from both benchmark-adapted and directly video-grounded prompts, covering content, format, style, and structure requirements, and evaluates model outputs with a hybrid verification pipeline. The benchmark contains 306 satisfiable test instructions, a 540-pair DPO preference dataset, and a 30-item conflict diagnostic subset. Experiments on 10 MLLMs show that joint constraint satisfaction remains challenging. We further show that DPO training on VCIFBench data can improve instruction-following performance.

Peichun Li, Hanwen Zhang, Yuan Wu et al.

Distributed Artificial Intelligence (AI) model training over mobile edge networks encounters significant challenges due to the data and resource heterogeneity of edge devices. The former hampers the convergence rate of the global model, while the latter diminishes the devices' resource utilization efficiency. In this paper, we propose a generative AI-empowered federated learning to address these challenges by leveraging the idea of FIlling the MIssing (FIMI) portion of local data. Specifically, FIMI can be considered as a resource-aware data augmentation method that effectively mitigates the data heterogeneity while ensuring efficient FL training. We first quantify the relationship between the training data amount and the learning performance. We then study the FIMI optimization problem with the objective of minimizing the device-side overall energy consumption subject to required learning performance constraints. The decomposition-based analysis and the cross-entropy searching method are leveraged to derive the solution, where each device is assigned suitable AI-synthesized data and resource utilization policy. Experiment results demonstrate that FIMI can save up to 50% of the device-side energy to achieve the target global test accuracy in comparison with the existing methods. Meanwhile, FIMI can significantly enhance the converged global accuracy under the non-independently-and-identically distribution (non-IID) data.

Yuan Wu, Zhiqiang Yan, Zhengxue Wang et al.

The task of vision-based 3D occupancy prediction aims to reconstruct 3D geometry and estimate its semantic classes from 2D color images, where the 2D-to-3D view transformation is an indispensable step. Most previous methods conduct forward projection, such as BEVPooling and VoxelPooling, both of which map the 2D image features into 3D grids. However, the current grid representing features within a certain height range usually introduces many confusing features that belong to other height ranges. To address this challenge, we present Deep Height Decoupling (DHD), a novel framework that incorporates explicit height prior to filter out the confusing features. Specifically, DHD first predicts height maps via explicit supervision. Based on the height distribution statistics, DHD designs Mask Guided Height Sampling (MGHS) to adaptively decouple the height map into multiple binary masks. MGHS projects the 2D image features into multiple subspaces, where each grid contains features within reasonable height ranges. Finally, a Synergistic Feature Aggregation (SFA) module is deployed to enhance the feature representation through channel and spatial affinities, enabling further occupancy refinement. On the popular Occ3D-nuScenes benchmark, our method achieves state-of-the-art performance even with minimal input frames. Source code is released at https://github.com/yanzq95/DHD.

Peichun Li, Guoliang Cheng, Xumin Huang et al.

In this work, we investigate the challenging problem of on-demand federated learning (FL) over heterogeneous edge devices with diverse resource constraints. We propose a cost-adjustable FL framework, named AnycostFL, that enables diverse edge devices to efficiently perform local updates under a wide range of efficiency constraints. To this end, we design the model shrinking to support local model training with elastic computation cost, and the gradient compression to allow parameter transmission with dynamic communication overhead. An enhanced parameter aggregation is conducted in an element-wise manner to improve the model performance. Focusing on AnycostFL, we further propose an optimization design to minimize the global training loss with personalized latency and energy constraints. By revealing the theoretical insights of the convergence analysis, personalized training strategies are deduced for different devices to match their locally available resources. Experiment results indicate that, when compared to the state-of-the-art efficient FL algorithms, our learning framework can reduce up to 1.9 times of the training latency and energy consumption for realizing a reasonable global testing accuracy. Moreover, the results also demonstrate that, our approach significantly improves the converged global accuracy.

Xumin Huang, Yuan Wu, Jiawen Kang et al.

Metaverse enables users to communicate, collaborate and socialize with each other through their digital avatars. Due to the spatio-temporal characteristics, co-located users are served well by performing their software components in a collaborative manner such that a Metaverse service provider (MSP) eliminates redundant data transmission and processing, ultimately reducing the total energy consumption. The energyefficient service provision is crucial for enabling the green and sustainable Metaverse. In this article, we take an augmented reality (AR) application as an example to achieve this goal. Moreover, we study an economic issue on how the users reserve offloading services from the MSP and how the MSP determines an optimal charging price since each user is rational to decide whether to accept the offloading service by taking into account the monetary cost. A single-leader multi-follower Stackelberg game is formulated between the MSP and users while each user optimizes an offloading probability to minimize the weighted sum of time, energy consumption and monetary cost. Numerical results show that our scheme achieves energy savings and satisfies individual rationality simultaneously compared with the conventional schemes. Finally, we identify and discuss open directions on how several emerging technologies are combined with the sustainable green Metaverse.

Yuan Wu, Zongxian Yang, Jiayu Qian et al.

Large vision-language models (VLMs) often benefit from chain-of-thought (CoT) prompting in general domains, yet its efficacy in medical vision-language tasks remains underexplored. We report a counter-intuitive trend: on medical visual question answering, CoT frequently underperforms direct answering (DirA) across general-purpose and medical-specific models. We attribute this to a \emph{medical perception bottleneck}: subtle, domain-specific cues can weaken visual grounding, and CoT may compound early perceptual uncertainty rather than correct it. To probe this hypothesis, we introduce two training-free, inference-time grounding interventions: (i) \emph{perception anchoring} via region-of-interest cues and (ii) \emph{description grounding} via high-quality textual guidance. Across multiple benchmarks and model families, these interventions improve accuracy, mitigate CoT degradation, and in several settings reverse the CoT--DirA inversion. Our findings suggest that reliable clinical VLMs require robust visual grounding and cross-modal alignment, beyond extending text-driven reasoning chains. Code is available \href{https://github.com/TianYin123/Better_Eyes_Better_Thoughts}{here}.

Chenlu Guo, Nuo Xu, Yi Chang et al.

With the rapid development of large language models (LLMs), assessing their performance on health-related inquiries has become increasingly essential. The use of these models in real-world contexts-where misinformation can lead to serious consequences for individuals seeking medical advice and support-necessitates a rigorous focus on safety and trustworthiness. In this work, we introduce CHBench, the first comprehensive safety-oriented Chinese health-related benchmark designed to evaluate LLMs' capabilities in understanding and addressing physical and mental health issues with a safety perspective across diverse scenarios. CHBench comprises 6,493 entries on mental health and 2,999 entries on physical health, spanning a wide range of topics. Our extensive evaluations of four popular Chinese LLMs highlight significant gaps in their capacity to deliver safe and accurate health information, underscoring the urgent need for further advancements in this critical domain. The code is available at https://github.com/TracyGuo2001/CHBench.

Yahan Li, Yi Wang, Yi Chang et al.

Large language models (LLMs) have demonstrated remarkable capabilities across a range of natural language processing (NLP) tasks, capturing the attention of both practitioners and the broader public. A key question that now preoccupies the AI community concerns the capabilities and limitations of these models, with trustworthiness emerging as a central issue, particularly as LLMs are increasingly applied in sensitive fields like healthcare and finance, where errors can have serious consequences. However, most previous studies on the trustworthiness of LLMs have been limited to a single language, typically the predominant one in the dataset, such as English. In response to the growing global deployment of LLMs, we introduce XTRUST, the first comprehensive multilingual trustworthiness benchmark. XTRUST encompasses a diverse range of topics, including illegal activities, hallucination, out-of-distribution (OOD) robustness, physical and mental health, toxicity, fairness, misinformation, privacy, and machine ethics, across 10 different languages. Using XTRUST, we conduct an empirical evaluation of the multilingual trustworthiness of five widely used LLMs, offering an in-depth analysis of their performance across languages and tasks. Our results indicate that many LLMs struggle with certain low-resource languages, such as Arabic and Russian, highlighting the considerable room for improvement in the multilingual trustworthiness of current language models. The code is available at https://github.com/LluckyYH/XTRUST.

Yupeng Chang, Yi Chang, Yuan Wu

Parameter-efficient fine-tuning (PEFT) has become a de facto standard for adapting Large Language Models (LLMs). However, we identify a critical vulnerability within popular low-rank adaptation methods like LoRA: their tendency to exacerbate "Catastrophic Inheritance" - the unchecked propagation of biases, noise, and data imbalances from pre-training. This phenomenon can degrade model robustness and fairness, undermining the benefits of efficient adaptation. To address this, we introduce Bias-Alleviating Low-Rank Adaptation (BA-LoRA). Our approach is founded on a principled decomposition of Catastrophic Inheritance into three core challenges: Knowledge Drift, Representation Collapse, and Overfitting to Noise. BA-LoRA systematically mitigates these issues by incorporating a trio of targeted regularizers - consistency, diversity, and SVD - designed to preserve core knowledge, enforce representational richness, and promote robust, low-rank output representations. We conduct comprehensive evaluations on a suite of natural language understanding (NLU) and generation (NLG) tasks using diverse, prominent open-source language models (e.g., LLaMA-2-7B and DeBERTa-v3-base). Our results show that BA-LoRA not only outperforms state-of-the-art LoRA variants in terms of performance and stability, but also demonstrates quantitatively superior robustness and bias mitigation on targeted evaluations. This confirms its ability to counteract the adverse effects of Catastrophic Inheritance.

Li Ping Qian, Yi Zhang, Sikai Lyu et al.

With the rapid development of Artificial Intelligent Internet of Things (AIoT), the image data from AIoT devices has been witnessing the explosive increasing. In this paper, a novel deep image semantic communication model is proposed for the efficient image communication in AIoT. Particularly, at the transmitter side, a high-precision image semantic segmentation algorithm is proposed to extract the semantic information of the image to achieve significant compression of the image data. At the receiver side, a semantic image restoration algorithm based on Generative Adversarial Network (GAN) is proposed to convert the semantic image to a real scene image with detailed information. Simulation results demonstrate that the proposed image semantic communication model can improve the image compression ratio and recovery accuracy by 71.93% and 25.07% on average in comparison with WebP and CycleGAN, respectively. More importantly, our demo experiment shows that the proposed model reduces the total delay by 95.26% in the image communication, when comparing with the original image transmission.

Yuan Wu, Zhiqiang Yan, Jiawei Lian et al.

3D occupancy prediction aims to infer dense, voxel-wise scene semantics from sensor observations, where the 2D-to-3D view transformation serves as a crucial step in bridging image features and volumetric representations. Most previous methods rely on a fixed projection space, where 3D reference points are uniformly sampled along pillars. However, such sampling struggles to capture the sparsity and height variations of real-world scenes, leading to ambiguous correspondences and unreliable feature aggregation. To address these challenges, we propose HiPR, a camera-LiDAR occupancy framework with Height-Guided Projection Reparameterization. HiPR first encodes LiDAR into a BEV height map to capture the maximum height of the point cloud. HiPR then adjusts the sampling range of each pillar using the height prior, enabling adaptive reparameterization of the projection space. As a result, the projected points are redistributed into geometrically meaningful regions rather than fixed ranges. Meanwhile, we mask out the invalid parts of the height map to avoid misleading the feature aggregation. In addition, to alleviate the training instability caused by noisy LiDAR-derived heights, we introduce a training-time Progressive Height Conditioning strategy, which gradually transitions the conditioning signal from ground-truth heights to LiDAR heights. Extensive experiments demonstrate that HiPR consistently outperforms existing state-of-the-art methods while maintaining real-time inference. The code and pretrained models can be found at https://github.com/Rayn-Wu/HiPR.

Yue Zhou, Chenlu Guo, Xu Wang et al.

Large models, encompassing large language and diffusion models, have shown exceptional promise in approximating human-level intelligence, garnering significant interest from both academic and industrial spheres. However, the training of these large models necessitates vast quantities of high-quality data, and with continuous updates to these models, the existing reservoir of high-quality data may soon be depleted. This challenge has catalyzed a surge in research focused on data augmentation methods. Leveraging large models, these data augmentation techniques have outperformed traditional approaches. This paper offers an exhaustive review of large model-driven data augmentation methods, adopting a comprehensive perspective. We begin by establishing a classification of relevant studies into three main categories: image augmentation, text augmentation, and paired data augmentation. Following this, we delve into various data post-processing techniques pertinent to large model-based data augmentation. Our discussion then expands to encompass the array of applications for these data augmentation methods within natural language processing, computer vision, and audio signal processing. We proceed to evaluate the successes and limitations of large model-based data augmentation across different scenarios. Concluding our review, we highlight prospective challenges and avenues for future exploration in the field of data augmentation. Our objective is to furnish researchers with critical insights, ultimately contributing to the advancement of more sophisticated large models. We consistently maintain the related open-source materials at: https://github.com/MLGroup-JLU/LLM-data-aug-survey.

Xu Wang, Cheng Li, Yi Chang et al.

Large Language Models (LLMs) have become integral to a wide spectrum of applications, ranging from traditional computing tasks to advanced artificial intelligence (AI) applications. This widespread adoption has spurred extensive research into LLMs across various disciplines, including the social sciences. Notably, studies have revealed that LLMs possess emotional intelligence, which can be further developed through positive emotional stimuli. This discovery raises an intriguing question: can negative emotions similarly influence LLMs, potentially enhancing their performance? In response to this question, we introduce NegativePrompt, a novel approach underpinned by psychological principles, involving ten specifically designed negative emotional stimuli. We embark on rigorous experimental evaluations of five LLMs including Flan-T5-Large, Vicuna, Llama 2, ChatGPT, and GPT-4, across a set of 45 tasks. The results are revealing: NegativePrompt markedly enhances the performance of LLMs, evidenced by relative improvements of 12.89% in Instruction Induction tasks and 46.25% in BIG-Bench tasks. Moreover, we conduct attention visualization experiments to decipher the underlying mechanisms of NegativePrompt's influence. Our research contributes significantly to the understanding of LLMs and emotion interaction, demonstrating the practical efficacy of NegativePrompt as an emotion-driven method and offering novel insights for the enhancement of LLMs in real-world applications. The code is available at https://github.com/wangxu0820/NegativePrompt.

Jinnan Li, Jinzhe Li, Yue Wang et al.

Multi-turn instruction following capability constitutes a core competency of large language models (LLMs) in real-world applications. Existing evaluation benchmarks predominantly focus on fine-grained constraint satisfaction and domain-specific capability assessment, yet overlook the crucial structural dependencies between dialogue turns that distinguish multi-turn from single-turn interactions. These structural dependencies not only reflect user intent but also establish an essential second dimension for the instruction following evaluation beyond constraint satisfaction. To address this gap, we propose StructFlowBench, a multi-turn instruction following benchmark with structural flow modeling. The benchmark defines an innovative structural flow framework with six fundamental inter-turn relationships. These relationships introduce novel structural constraints for model evaluation and also serve as generation parameters for creating customized dialogue flows tailored to specific scenarios. Adopting established LLM-based automatic evaluation methodologies, we conduct systematic evaluations of 13 leading open-source and closed-source LLMs. Experimental results reveal significant deficiencies in current models' comprehension of multi-turn dialogue structures. The code is available at https://github.com/MLGroupJLU/StructFlowBench.

Gengxu Li, Tingyu Xia, Yi Chang et al.

Direct Preference Optimization (DPO) is a widely adopted offline algorithm for preference-based reinforcement learning from human feedback (RLHF), designed to improve training simplicity and stability by redefining reward functions. However, DPO is hindered by several limitations, including length bias, memory inefficiency, and probability degradation. To address these challenges, we propose Length-Controlled Margin-Based Preference Optimization (LMPO), a more efficient and robust alternative. LMPO introduces a uniform reference model as an upper bound for the DPO loss, enabling a more accurate approximation of the original optimization objective. Additionally, an average log-probability optimization strategy is employed to minimize discrepancies between training and inference phases. A key innovation of LMPO lies in its Length-Controlled Margin-Based loss function, integrated within the Bradley-Terry framework. This loss function regulates response length while simultaneously widening the margin between preferred and rejected outputs. By doing so, it mitigates probability degradation for both accepted and discarded responses, addressing a significant limitation of existing methods. We evaluate LMPO against state-of-the-art preference optimization techniques on two open-ended large language models, Mistral and LLaMA3, across six conditional benchmarks. Our experimental results demonstrate that LMPO effectively controls response length, reduces probability degradation, and outperforms existing approaches. The code is available at https://github.com/gengxuli/LMPO.

Zhiyuan Li, Tingyu Xia, Yi Chang et al.

The Receptance Weighted Key Value (RWKV) model offers a novel alternative to the Transformer architecture, merging the benefits of recurrent and attention-based systems. Unlike conventional Transformers, which depend heavily on self-attention, RWKV adeptly captures long-range dependencies with minimal computational demands. By utilizing a recurrent framework, RWKV addresses some computational inefficiencies found in Transformers, particularly in tasks with long sequences. RWKV has recently drawn considerable attention for its robust performance across multiple domains. Despite its growing popularity, no systematic review of the RWKV model exists. This paper seeks to fill this gap as the first comprehensive review of the RWKV architecture, its core principles, and its varied applications, such as natural language generation, natural language understanding, and computer vision. We assess how RWKV compares to traditional Transformer models, highlighting its capability to manage long sequences efficiently and lower computational costs. Furthermore, we explore the challenges RWKV encounters and propose potential directions for future research and advancement. We consistently maintain the related open-source materials at: https://github.com/MLGroupJLU/RWKV-Survey.

Haiqi Yang, Jinzhe Li, Gengxu Li et al.

Large Multimodal Models (LMMs) have witnessed remarkable growth, showcasing formidable capabilities in handling intricate multimodal tasks with exceptional performance. Recent research has underscored the inclination of large language models to passively accept defective inputs, often resulting in futile reasoning on invalid prompts. However, the same critical question of whether LMMs can actively detect and scrutinize erroneous inputs still remains unexplored. To address this gap, we introduce the Input Scrutiny Ability Evaluation Framework (ISEval), which encompasses seven categories of flawed premises and three evaluation metrics. Our extensive evaluation of ten advanced LMMs has identified key findings. Most models struggle to actively detect flawed textual premises without guidance, which reflects a strong reliance on explicit prompts for premise error identification. Error type affects performance: models excel at identifying logical fallacies but struggle with surface-level linguistic errors and certain conditional flaws. Modality trust varies-Gemini 2.5 pro and Claude Sonnet 4 balance visual and textual info, while aya-vision-8b over-rely on text in conflicts. These insights underscore the urgent need to enhance LMMs' proactive verification of input validity and shed novel insights into mitigating the problem. The code is available at https://github.com/MLGroupJLU/LMM_ISEval.

Jinzhe Li, Gengxu Li, Yi Chang et al.

Large language models (LLMs) have witnessed rapid advancements, demonstrating remarkable capabilities. However, a notable vulnerability persists: LLMs often uncritically accept flawed or contradictory premises, leading to inefficient reasoning and unreliable outputs. This emphasizes the significance of possessing the \textbf{Premise Critique Ability} for LLMs, defined as the capacity to proactively identify and articulate errors in input premises. Most existing studies assess LLMs' reasoning ability in ideal settings, largely ignoring their vulnerabilities when faced with flawed premises. Thus, we introduce the \textbf{Premise Critique Bench (PCBench)}, designed by incorporating four error types across three difficulty levels, paired with multi-faceted evaluation metrics. We conducted systematic evaluations of 15 representative LLMs. Our findings reveal: (1) Most models rely heavily on explicit prompts to detect errors, with limited autonomous critique; (2) Premise critique ability depends on question difficulty and error type, with direct contradictions being easier to detect than complex or procedural errors; (3) Reasoning ability does not consistently correlate with the premise critique ability; (4) Flawed premises trigger overthinking in reasoning models, markedly lengthening responses due to repeated attempts at resolving conflicts. These insights underscore the urgent need to enhance LLMs' proactive evaluation of input validity, positioning premise critique as a foundational capability for developing reliable, human-centric systems. The code is available at https://github.com/MLGroupJLU/Premise_Critique.

Yuan Wu, Zhiqiang Yan, Yigong Zhang et al.

Occupancy prediction aims to estimate the 3D spatial distribution of occupied regions along with their corresponding semantic labels. Existing vision-based methods perform well on daytime benchmarks but struggle in nighttime scenarios due to limited visibility and challenging lighting conditions. To address these challenges, we propose LIAR, a novel framework that learns illumination-affined representations. LIAR first introduces Selective Low-light Image Enhancement (SLLIE), which leverages the illumination priors from daytime scenes to adaptively determine whether a nighttime image is genuinely dark or sufficiently well-lit, enabling more targeted global enhancement. Building on the illumination maps generated by SLLIE, LIAR further incorporates two illumination-aware components: 2D Illumination-guided Sampling (2D-IGS) and 3D Illumination-driven Projection (3D-IDP), to respectively tackle local underexposure and overexposure. Specifically, 2D-IGS modulates feature sampling positions according to illumination maps, assigning larger offsets to darker regions and smaller ones to brighter regions, thereby alleviating feature degradation in underexposed areas. Subsequently,3D-IDP enhances semantic understanding in overexposed regions by constructing illumination intensity fields and supplying refined residual queries to the BEV context refinement process. Extensive experiments on both real and synthetic datasets demonstrate the superior performance of LIAR under challenging nighttime scenarios. The source code and pretrained models are available [here](https://github.com/yanzq95/LIAR).

Yupeng Chang, Yi Chang, Yuan Wu

Large language models (LLMs) face significant challenges when balancing multiple high-level objectives, such as generating coherent, relevant, and high-quality responses while maintaining efficient task adaptation across diverse tasks. To address these challenges, we introduce Transfer-Prompting, a novel two-stage framework designed to enhance cross-task adaptation in prompt generation. The framework comprises two key components: (1) source prompt construction, which refines the original prompts on source task datasets to generate source prompts with enhanced generalization ability, and (2) target prompt generation, which enhances cross-task adaptation of target prompts by fine-tuning a set of high-scored source prompts on task-specific datasets. In each optimization cycle, a reference LLM generates candidate prompts based on historical prompt-score pairs and task descriptions in our designed reference prompt. These candidate prompts are refined iteratively, while a scorer LLM evaluates their effectiveness using the multi-dimensional metrics designed in the objective prompts evaluator-a novel contribution in this work that provides a holistic evaluation of prompt quality and task performance. This feedback loop facilitates continuous refinement, optimizing both prompt quality and task-specific outcomes. We validate Transfer-Prompting through extensive experiments across 25 LLMs, including 7 foundational models and 18 specialized models, evaluated on 9 diverse datasets. The results demonstrate that Transfer-Prompting significantly improves task-specific performance, highlighting its potential for enhancing cross-task adaptation in LLMs. The code is available at https://github.com/llm172/Transfer-Prompting.

Yahan Li, Tingyu Xia, Yi Chang et al.

As large language models (LLMs) continue to evolve, efficient evaluation metrics are vital for assessing their ability to compress information and reduce redundancy. While traditional metrics like Matrix Entropy offer valuable insights, they are computationally intensive for large-scale models due to their \( O(n^3) \) time complexity with Singular Value Decomposition (SVD). To mitigate this issue, we introduce the Matrix Nuclear-Norm, which not only serves as a metric to quantify the data compression proficiency of LLM but also provides a convex approximation of matrix rank to capture both predictive discriminability and diversity. By employing the \( L_{1,2}\text{-norm} \) to further approximate the nuclear norm, we can effectively assess the model's information compression capabilities. This approach reduces the time complexity to \( O(n^2) \) and eliminates the need for SVD computation. Consequently, the Matrix Nuclear-Norm achieves speeds 8 to 24 times faster than Matrix Entropy for the CEREBRAS-GPT model as sizes increase from 111M to 6.7B. This performance gap becomes more pronounced with larger models, as validated in tests with other models like Pythia. Additionally, evaluations on benchmarks and model responses confirm that our proposed Matrix Nuclear-Norm is a reliable, scalable, and efficient tool for assessing LLMs' performance, striking a balance between accuracy and computational efficiency. The code is available at https://github.com/MLGroupJLU/MatrixNuclearNorm.

Juntao Hu, Yuan Wu

We propose a novel data augmentation method termed You Only Need hAlf (YONA), which simplifies the augmentation process. YONA bisects an image, substitutes one half with noise, and applies data augmentation techniques to the remaining half. This method reduces the redundant information in the original image, encourages neural networks to recognize objects from incomplete views, and significantly enhances neural networks' robustness. YONA is distinguished by its properties of parameter-free, straightforward application, enhancing various existing data augmentation strategies, and thereby bolstering neural networks' robustness without additional computational cost. To demonstrate YONA's efficacy, extensive experiments were carried out. These experiments confirm YONA's compatibility with diverse data augmentation methods and neural network architectures, yielding substantial improvements in CIFAR classification tasks, sometimes outperforming conventional image-level data augmentation methods. Furthermore, YONA markedly increases the resilience of neural networks to adversarial attacks. Additional experiments exploring YONA's variants conclusively show that masking half of an image optimizes performance. The code is available at https://github.com/HansMoe/YONA.

Zheng Xing, Weibing Zhao, Guanghui Zhang et al.

Traditional radio map construction methods mandate labor-intensive data collection and precise location labeling. To address these limitations, we propose a novel survey-free approach for radio map construction that relies solely on unlabeled Received Signal Strength (RSS) measurements, thereby obviating the need for manual site surveys or auxiliary Inertial Measurement Units (IMUs). The key idea involves embedding multiple unlabeled RSS sequences into a known indoor layout, specifically targeting corridor-guided environments with a dominant unidirectional pedestrian flow. However, aligning the embedded coordinates with the RSS collection locations remains challenging due to the random fluctuations inherent in RSS data. To tackle this, we introduce a Hidden Markov Model (HMM)- based Coarse-to-Fine Inference (HCFI) framework. At the coarse level, we employ an HMM-based region label inference algorithm to partition RSS sequences and align the RSS segments with specific physical regions using graph-based inference. At the fine level, we develop an HMM-based location label inference technique to estimate RSS collection coordinates by leveraging RSS propagation principles while incorporating sequential spatio-temporal mobility probability. Empirical results from an office environment demonstrate that the proposed method achieves a radio map construction Mean Absolute Error (MAE) of 8.96 dB. Furthermore, based on the estimated radio map, k-Nearest Neighbor (KNN) localization yields an average positioning error of approximately 3.33 meters, offering a highly viable, survey-free solution for radio map construction under sequential topological assumptions.

Jinda Liu, Bo Cheng, Yi Chang et al.

Parameter-Efficient Fine-Tuning (PEFT) is essential for adapting Large Language Models (LLMs). In practice, LLMs are often required to handle a diverse set of tasks from multiple domains, a scenario naturally addressed by multi-task learning (MTL). Within this MTL context, a prevailing trend involves LoRA variants with multiple adapters or heads, which advocate for structural diversity to capture task-specific knowledge. Our findings present a direct challenge to this paradigm. We first show that a simplified multi-head architecture with high inter-head similarity substantially outperforms complex multi-adapter and multi-head systems. This leads us to question the multi-component paradigm itself, and we further demonstrate that a standard single-adapter LoRA, with a sufficiently increased rank, also achieves highly competitive performance. These results lead us to a new hypothesis: effective MTL generalization hinges on learning robust shared representations, not isolating task-specific features. To validate this, we propose Align-LoRA, which incorporates an explicit loss to align task representations within the shared adapter space. Experiments confirm that Align-LoRA significantly surpasses all baselines, establishing a simpler yet more effective paradigm for adapting LLMs to multiple tasks. The code is available at https://github.com/jinda-liu/Align-LoRA.

Yue Zhou, Yi Chang, Yuan Wu

Reasoning is a critical capability of multimodal large language models (MLLMs) for solving complex multimodal tasks, and judging the correctness of reasoning steps is crucial for improving this capability. Recently, MLLM-based process judges (MPJs) have been widely used to assess the correctness of reasoning steps in multimodal tasks. Therefore, evaluating MPJs is important for identifying their limitations and guiding future improvements. However, existing benchmarks for MPJs mainly focus on tasks such as step correctness classification and reasoning process search, while overlooking a key aspect: whether the confidence scores produced by MPJs at the step level are reliable. To address this gap, we propose ConfProBench, the first comprehensive benchmark designed to systematically evaluate the reliability of step-level confidence scores generated by MPJs. Our benchmark constructs three types of adversarially perturbed reasoning steps: Synonym Substitution, Syntactic Transformation, and Image Perturbation, to test the robustness of MPJ confidence under perturbations. In addition, we introduce three novel evaluation metrics: Confidence Robustness Score (CRS), Confidence Sensitivity Score (CSS), and Confidence Calibration Score (CCS), which evaluate robustness, sensitivity, and calibration, respectively. We evaluate 14 state-of-the-art MLLMs, including both proprietary and open-source models. Experiments reveal limitations in current MPJs' confidence performance and offer competitive baselines to support future research.

Yahan Li, Yuan Wu

Unsupervised domain adaptation (UDA) aims to transfer knowledge from a labeled source domain to an unlabeled target domain. The most recent UDA methods always resort to adversarial training to yield state-of-the-art results and a dominant number of existing UDA methods employ convolutional neural networks (CNNs) as feature extractors to learn domain invariant features. Vision transformer (ViT) has attracted tremendous attention since its emergence and has been widely used in various computer vision tasks, such as image classification, object detection, and semantic segmentation, yet its potential in adversarial domain adaptation has never been investigated. In this paper, we fill this gap by employing the ViT as the feature extractor in adversarial domain adaptation. Moreover, we empirically demonstrate that ViT can be a plug-and-play component in adversarial domain adaptation, which means directly replacing the CNN-based feature extractor in existing UDA methods with the ViT-based feature extractor can easily obtain performance improvement. The code is available at https://github.com/LluckyYH/VT-ADA.

Jinda Liu, Yi Chang, Yuan Wu

Fine-tuning large language models (LLMs) is computationally expensive, and Low-Rank Adaptation (LoRA) provides a cost-effective solution by approximating weight updates through low-rank matrices. In real-world scenarios, LLMs are fine-tuned on data from multiple domains to perform tasks across various fields, embodying multi-task learning (MTL). LoRA often underperforms in such complex scenarios. To enhance LoRA's capability in multi-task learning, we propose R-LoRA, which incorporates Multi-Head Randomization. Multi-Head Randomization diversifies the head matrices through Multi-Head Dropout and Multi-Head Random Initialization, enabling more efficient learning of task-specific features while maintaining shared knowledge representation. Our approach not only improves performance in MTL but also reduces GPU memory usage and training time. Experiments show that R-LoRA's gains stem from increased diversity in the head matrices, demonstrating its effectiveness for multi-task learning. The code is available at https://github.com/jinda-liu/R-LoRA

Gengxu Li, Yuan Wu

Source-free unsupervised domain adaptation (SFUDA) has gained significant attention as an alternative to traditional unsupervised domain adaptation (UDA), which relies on the constant availability of labeled source data. However, SFUDA approaches come with inherent limitations that are frequently overlooked. These challenges include performance degradation when the unlabeled target data fails to meet critical assumptions, such as having a closed-set label distribution identical to that of the source domain, or when sufficient unlabeled target data is unavailable-a common situation in real-world applications. To address these issues, we propose an asymmetric co-training (ACT) method specifically designed for the SFFSDA scenario. SFFSDA presents a more practical alternative to SFUDA, as gathering a few labeled target instances is more feasible than acquiring large volumes of unlabeled target data in many real-world contexts. Our ACT method begins by employing a weak-strong augmentation to enhance data diversity. Then we use a two-step optimization process to train the target model. In the first step, we optimize the label smoothing cross-entropy loss, the entropy of the class-conditional distribution, and the reverse-entropy loss to bolster the model's discriminative ability while mitigating overfitting. The second step focuses on reducing redundancy in the output space by minimizing classifier determinacy disparity. Extensive experiments across four benchmarks demonstrate the superiority of our ACT approach, which outperforms state-of-the-art SFUDA methods and transfer learning techniques. Our findings suggest that adapting a source pre-trained model using only a small amount of labeled target data offers a practical and dependable solution. The code is available at https://github.com/gengxuli/ACT.

Tingyu Xia, Bowen Yu, Kai Dang et al.

Supervised fine-tuning (SFT) is crucial for aligning Large Language Models (LLMs) with human instructions. The primary goal during SFT is to select a small yet representative subset of training data from the larger pool, such that fine-tuning with this subset achieves results comparable to or even exceeding those obtained using the entire dataset. However, most existing data selection techniques are designed for small-scale data pools, which fail to meet the demands of real-world SFT scenarios. In this paper, we replicated several self-scoring methods those that do not rely on external model assistance on two million scale datasets, and found that nearly all methods struggled to significantly outperform random selection when dealing with such large-scale data pools. Moreover, our comparisons suggest that, during SFT, diversity in data selection is more critical than simply focusing on high quality data. We also analyzed the limitations of several current approaches, explaining why they perform poorly on large-scale datasets and why they are unsuitable for such contexts. Finally, we found that filtering data by token length offers a stable and efficient method for improving results. This approach, particularly when training on long text data, proves highly beneficial for relatively weaker base models, such as Llama3.

Tingyu Xia, Bowen Yu, Yuan Wu et al.

In this paper, we initiate our discussion by demonstrating how Large Language Models (LLMs), when tasked with responding to queries, display a more even probability distribution in their answers if they are more adept, as opposed to their less skilled counterparts. Expanding on this foundational insight, we propose a new self-evaluation method ProbDiff for assessing the efficacy of various LLMs. This approach obviates the necessity for an additional evaluation model or the dependence on external, proprietary models like GPT-4 for judgment. It uniquely utilizes the LLMs being tested to compute the probability discrepancy between the initial response and its revised versions. A higher discrepancy for a given query between two LLMs indicates a relatively weaker capability. Our findings reveal that ProbDiff achieves results on par with those obtained from evaluations based on GPT-4, spanning a range of scenarios that include natural language generation (NLG) tasks such as translation, summarization, and our proposed Xiaohongshu blog writing task, and benchmarks for LLM evaluation like AlignBench, MT-Bench, and AlpacaEval, across LLMs of varying magnitudes.

Yuxing Cheng, Yi Chang, Yuan Wu

Recent advancements in Large Language Models (LLMs) have demonstrated significant progress in various areas, such as text generation and code synthesis. However, the reliability of performance evaluation has come under scrutiny due to data contamination-the unintended overlap between training and test datasets. This overlap has the potential to artificially inflate model performance, as LLMs are typically trained on extensive datasets scraped from publicly available sources. These datasets often inadvertently overlap with the benchmarks used for evaluation, leading to an overestimation of the models' true generalization capabilities. In this paper, we first examine the definition and impacts of data contamination. Secondly, we review methods for contamination-free evaluation, focusing on three strategies: data updating-based methods, data rewriting-based methods, and prevention-based methods. Specifically, we highlight dynamic benchmarks and LLM-driven evaluation methods. Finally, we categorize contamination detecting methods based on model information dependency: white-Box, gray-Box, and black-Box detection approaches. Our survey highlights the requirements for more rigorous evaluation protocols and proposes future directions for addressing data contamination challenges.

Zhiyuan Li, Yi Chang, Yuan Wu

Large reasoning models (LRMs) have achieved impressive performance in complex tasks, often outperforming conventional large language models (LLMs). However, the prevalent issue of overthinking severely limits their computational efficiency. Overthinking occurs when models generate excessive and redundant tokens that contribute little to accurate outcomes, especially in simple tasks, resulting in a significant waste of computational resources. To systematically investigate this issue, we introduce Think-Bench, a benchmark designed to evaluate the reasoning efficiency of LRMs. We also propose novel efficiency metrics and conduct a comprehensive evaluation of various LRMs across multiple dimensions, including the reasoning process, outcome quality, and chain-of-thought (CoT) characteristics. Our analysis reveals that most LRMs exhibit overthinking in handling easy questions, generating unnecessarily lengthy reasoning chains. While many LRMs demonstrate high CoT quality, several suffer from low efficiency. We hope that Think-Bench can serve as a robust foundation for advancing research into LRMs.

Yue Zhou, Yi Chang, Yuan Wu

Model merging aims to integrate multiple task-specific models into a unified model that inherits the capabilities of the task-specific models, without additional training. Existing model merging methods often lack consideration of the varying contribution ratios of different task-specific models to the final merged model. In this paper, we propose Mixup Model Merge (M3), a simple yet effective method inspired by the randomized linear interpolation strategy from the Mixup data augmentation technique. M3 performs randomized linear interpolation in parameter space between two task-specific LLMs, where interpolation coefficients are sampled from a Beta distribution to explore diverse contribution ratios. This controllable randomness allows M3 to outperform standard equal-ratio merging by discovering better contribution ratio combinations. Extensive experiments show that M3 significantly (1) improves merged LLM performance across tasks, (2) enhances out-of-distribution and adversarial robustness, (3) outperforms the positive effects of the sparsification method DARE on model merging and can be further combined with DARE to achieve superior results, and (4) balances exploration efficiency and diversity in contribution ratios by tuning the Beta distribution's shape parameters. The code is provided in the supplementary materials.

Zhengxue Wang, Yuan Wu, Xiang Li et al.

Depth super-resolution has achieved impressive performance, and the incorporation of multi-frame information further enhances reconstruction quality. Nevertheless, statistical analyses reveal that video depth super-resolution remains affected by pronounced long-tailed distributions, with the long-tailed effects primarily manifesting in spatial non-smooth regions and temporal variation zones. To address these challenges, we propose a novel SpatioTemporal Difference Network (STDNet) comprising two core branches: a spatial difference branch and a temporal difference branch. In the spatial difference branch, we introduce a spatial difference mechanism to mitigate the long-tailed issues in spatial non-smooth regions. This mechanism dynamically aligns RGB features with learned spatial difference representations, enabling intra-frame RGB-D aggregation for depth calibration. In the temporal difference branch, we further design a temporal difference strategy that preferentially propagates temporal variation information from adjacent RGB and depth frames to the current depth frame, leveraging temporal difference representations to achieve precise motion compensation in temporal long-tailed areas. Extensive experimental results across multiple datasets demonstrate the effectiveness of our STDNet, outperforming existing approaches.

Yupeng Chang, Chenlu Guo, Yi Chang et al.

Parameter-efficient fine-tuning (PEFT), particularly Low-Rank Adaptation (LoRA), adapts large language models (LLMs) by training only a small fraction of parameters. However, as the rank of the low-rank matrices used for adaptation increases, LoRA often exhibits an unstable "double descent" phenomenon, characterized by transient divergence in the training loss, which delays convergence and impairs generalization by causing instability due to the attraction to sharp local minima. To address this, we introduce LoRA-MGPO, a framework that incorporates Momentum-Guided Perturbation Optimization (MGPO). MGPO stabilizes training dynamics by mitigating the double descent phenomenon and guiding weight perturbations using momentum vectors from the optimizer's state, thus avoiding dual gradient computations. Additionally, an adaptive normalization scheme scales the magnitude of perturbations based on an exponential moving average (EMA) of gradient norms, further enhancing stability. While EMA controls the magnitude of the perturbations, MGPO guides their direction, ensuring a more stable optimization trajectory. Experiments on a suite of natural language understanding and generation benchmarks show that LoRA-MGPO consistently achieves superior performance over LoRA and other PEFT methods. The analysis indicates that LoRA-MGPO leads to smoother loss curves, faster convergence, and improved generalization by stabilizing the training process and mitigating the attraction to sharp minima.

Juntao Hu, Yuan Wu

The most successful multi-domain text classification (MDTC) approaches employ the shared-private paradigm to facilitate the enhancement of domain-invariant features through domain-specific attributes. Additionally, they employ adversarial training to align marginal feature distributions. Nevertheless, these methodologies encounter two primary challenges: (1) Neglecting class-aware information during adversarial alignment poses a risk of misalignment; (2) The limited availability of labeled data across multiple domains fails to ensure adequate discriminative capacity for the model. To tackle these issues, we propose a method called Regularized Conditional Alignment (RCA) to align the joint distributions of domains and classes, thus matching features within the same category and amplifying the discriminative qualities of acquired features. Moreover, we employ entropy minimization and virtual adversarial training to constrain the uncertainty of predictions pertaining to unlabeled data and enhance the model's robustness. Empirical results on two benchmark datasets demonstrate that our RCA approach outperforms state-of-the-art MDTC techniques.

Zhengxue Wang, Zhiqiang Yan, Yuan Wu et al.

Recent guided depth super-resolution methods are premised on the assumption of strictly spatial alignment between depth and RGB, achieving high-quality depth reconstruction. However, in real-world scenarios, the acquisition of strictly aligned RGB-D is hindered by inherent hardware limitations (e.g., physically separate RGB-D sensors) and unavoidable calibration drift induced by mechanical vibrations or temperature variations. Consequently, existing approaches often suffer inevitable performance degradation when applied to misaligned real-world scenes. In this paper, we propose the Multi-Order Matching Network (MOMNet), a novel alignment-free framework that adaptively retrieves and selects the most relevant information from misaligned RGB. Specifically, our method begins with a multi-order matching mechanism, which jointly performs zero-order, first-order, and second-order matching to comprehensively identify RGB information consistent with depth across multi-order feature spaces. To effectively integrate the retrieved RGB and depth, we further introduce a multi-order aggregation composed of multiple structure detectors. This strategy uses multi-order priors as prompts to facilitate the selective feature transfer from RGB to depth. Extensive experiments demonstrate that MOMNet achieves state-of-the-art performance and exhibits outstanding robustness.

Bo Cheng, Xu Wang, Jinda Liu et al.

Low-Rank Adaptation (LoRA) has emerged as one of the most widely used parameter-efficient fine-tuning (PEFT) methods for adapting large language models (LLMs) to downstream tasks. While highly effective in single-task settings, it struggles to efficiently leverage inter-task knowledge in complex multi-task learning scenarios, often requiring substantial task-specific data to achieve optimal performance. To address this limitation, we introduce MeTA-LoRA, a two-stage optimization framework that significantly improves data efficiency in multi-task adaptation. In the first stage, task-specific LoRA adapters are learned using only a few samples from each involved dataset, enabling rapid adaptation without large-scale supervision. In the second stage, the shared LoRA adapter is updated by aggregating gradients from multiple tasks to promote knowledge transfer across tasks, further reducing data usage by leveraging common patterns. In both multi-task learning and multilingual learning scenarios, our method matches or surpasses the performance of traditional full-data LoRA fine-tuning approaches, while using significantly less task-specific data.

Yanghe Pan, Yuntao Wang, Shaolong Guo et al.

Semantic communication (SemCom) has the potential to significantly reduce communication delay in vehicle-to-everything (V2X) communications within vehicular networks (VNs). However, the deployment of vehicular SemCom networks (VN-SemComNets) faces critical trust challenges in information transmission, semantic encoding, and communication entity reliability. This paper proposes an innovative three-layer trustworthy VN-SemComNet architecture. Specifically, we introduce a semantic camouflage transmission mechanism leveraging defensive adversarial noise for active eavesdropping defense, a robust federated encoder-decoder training framework to mitigate encoder-decoder poisoning attacks, and an audit game-based distributed vehicle trust management mechanism to deter untrustworthy vehicles. A case study validates the effectiveness of the proposed solutions. Lastly, essential future research directions are pointed out to advance this emerging field.

Jialin Li, Jinzhe Li, Gengxu Li et al.

With the advancement of code generation capabilities in large language models (LLMs), their reliance on input premises has intensified. When users provide inputs containing faulty premises, the probability of code generation hallucinations rises significantly, exposing deficiencies in their self-scrutiny capabilities. This paper proposes Faulty Premises Bench (FPBench), the first code generation evaluation framework targeting faulty premises. By systematically constructing three categories of faulty premises and integrating multi-dimensional evaluation metrics, it conducts in-depth assessments of 15 representative LLMs. The key findings are as follows: (1) Most models exhibit poor reasoning abilities and suboptimal code generation performance under faulty premises, heavily relying on explicit prompts for error detection, with limited self-scrutiny capabilities; (2) Faulty premises trigger a point of diminishing returns in resource investment, leading to blindly increasing length fails to enhance quality; (3) The three types of faulty premises respectively activate distinct defect patterns in models, revealing a triple dissociation in the cognitive mechanisms of code generation models. This study not only highlights the urgent need for LLMs to proactively verify premises in code generation but also, through the proposed FPBench framework and multi-dimensional evaluation system, provides a theoretical foundation and practical pathway for developing reliable, human-centric code generation models.

Xiaolin Yan, Yangxing Liu, Jiazhang Zheng et al.

Large language models (LLMs) have recently achieved significant advances in reasoning and demonstrated their advantages in solving challenging problems. Yet, their effectiveness in the semiconductor display industry remains limited due to a lack of domain-specific training and expertise. To bridge this gap, we present X-Intelligence 3.0, the first high-performance reasoning model specifically developed for the semiconductor display industry. This model is designed to deliver expert-level understanding and reasoning for the industry's complex challenges. Leveraging a carefully curated industry knowledge base, the model undergoes supervised fine-tuning and reinforcement learning to enhance its reasoning and comprehension capabilities. To further accelerate development, we implemented an automated evaluation framework that simulates expert-level assessments. We also integrated a domain-specific retrieval-augmented generation (RAG) mechanism, resulting in notable performance gains on benchmark datasets. Despite its relatively compact size of 32 billion parameters, X-Intelligence 3.0 outperforms SOTA DeepSeek-R1-671B across multiple evaluations. This demonstrates its exceptional efficiency and establishes it as a powerful solution to the longstanding reasoning challenges faced by the semiconductor display industry.

Guanting Liu, Yi Wang, Xi Chen et al.

Landslides are among the most common natural disasters globally, posing significant threats to human society. Deep learning (DL) has proven to be an effective method for rapidly generating landslide inventories in large-scale disaster areas. However, DL models rely heavily on high-quality labeled landslide data for strong feature extraction capabilities. And landslide detection using DL urgently needs a benchmark dataset to evaluate the generalization ability of the latest models. To solve the above problems, we construct a Large-scale Multi-source High-resolution Landslide Dataset (LMHLD) for Landslide Detection based on DL. LMHLD collects remote sensing images from five different satellite sensors across seven study areas worldwide: Wenchuan, China (2008); Rio de Janeiro, Brazil (2011); Gorkha, Nepal (2015); Jiuzhaigou, China (2015); Taiwan, China (2018); Hokkaido, Japan (2018); Emilia-Romagna, Italy (2023). The dataset includes a total of 25,365 patches, with different patch sizes to accommodate different landslide scales. Additionally, a training module, LMHLDpart, is designed to accommodate landslide detection tasks at varying scales and to alleviate the issue of catastrophic forgetting in multi-task learning. Furthermore, the models trained by LMHLD is applied in other datasets to highlight the robustness of LMHLD. Five dataset quality evaluation experiments designed by using seven DL models from the U-Net family demonstrate that LMHLD has the potential to become a benchmark dataset for landslide detection. LMHLD is open access and can be accessed through the link: https://doi.org/10.5281/zenodo.11424988. This dataset provides a strong foundation for DL models, accelerates the development of DL in landslide detection, and serves as a valuable resource for landslide prevention and mitigation efforts.

Chenlu Guo, Yuan Wu, Yi Chang

Parameter-efficient fine-tuning (PEFT) is essential for adapting large language models (LLMs), with low-rank adaptation (LoRA) being the most popular approach. However, LoRA suffers from slow convergence, and some recent LoRA variants, such as PiSSA, primarily rely on Singular Value Decomposition (SVD) for initialization, leading to expensive computation. To mitigate these problems, we use the Nyström method, which follows a three-matrix manipulation. We first introduce StructuredLoRA (SLoRA), which investigates adding a small intermediate matrix between the low-rank matrices A and B. Secondly, we propose NyströmLoRA (NLoRA), which leverages Nyström-based initialization for SLoRA to improve its effectiveness and efficiency. Finally, we propose IntermediateTune (IntTune), which explores fine-tuning exclusively on the intermediate matrix of NLoRA to further boost LLM efficiency. We evaluate our methods on five natural language generation (NLG) tasks and eight natural language understanding (NLU) tasks. On GSM8K, SLoRA and NLoRA achieve accuracies of 56.48% and 57.70%, surpassing LoRA by 33.52% and 36.41%, with only 3.67 million additional trainable parameters. IntTune improves average NLG performance over LoRA by 7.45% while using only 1.25% of its parameters. These results demonstrate the efficiency and effectiveness of our approach in enhancing model performance with minimal parameter overhead.

Xueluan Gong, Bowei Tian, Meng Xue et al.

Recent studies have revealed the vulnerability of Deep Neural Network (DNN) models to backdoor attacks. However, existing backdoor attacks arbitrarily set the trigger mask or use a randomly selected trigger, which restricts the effectiveness and robustness of the generated backdoor triggers. In this paper, we propose a novel attention-based mask generation methodology that searches for the optimal trigger shape and location. We also introduce a Quality-of-Experience (QoE) term into the loss function and carefully adjust the transparency value of the trigger in order to make the backdoored samples to be more natural. To further improve the prediction accuracy of the victim model, we propose an alternating retraining algorithm in the backdoor injection process. The victim model is retrained with mixed poisoned datasets in even iterations and with only benign samples in odd iterations. Besides, we launch the backdoor attack under a co-optimized attack framework that alternately optimizes the backdoor trigger and backdoored model to further improve the attack performance. Apart from DNN models, we also extend our proposed attack method against vision transformers. We evaluate our proposed method with extensive experiments on VGG-Flower, CIFAR-10, GTSRB, CIFAR-100, and ImageNette datasets. It is shown that we can increase the attack success rate by as much as 82\% over baselines when the poison ratio is low and achieve a high QoE of the backdoored samples. Our proposed backdoor attack framework also showcases robustness against state-of-the-art backdoor defenses.

Yuan Wu

Multi-domain text classification (MDTC) endeavors to harness available resources from correlated domains to enhance the classification accuracy of the target domain. Presently, most MDTC approaches that embrace adversarial training and the shared-private paradigm exhibit cutting-edge performance. Unfortunately, these methods face a non-negligible challenge: the absence of theoretical guarantees in the design of MDTC algorithms. The dearth of theoretical underpinning poses a substantial impediment to the advancement of MDTC algorithms. To tackle this problem, we first provide a theoretical analysis of MDTC by decomposing the MDTC task into multiple domain adaptation tasks. We incorporate the margin discrepancy as the measure of domain divergence and establish a new generalization bound based on Rademacher complexity. Subsequently, we propose a margin discrepancy-based adversarial training (MDAT) approach for MDTC, in accordance with our theoretical analysis. To validate the efficacy of the proposed MDAT method, we conduct empirical studies on two MDTC benchmarks. The experimental results demonstrate that our MDAT approach surpasses state-of-the-art baselines on both datasets.

Xi Chen, Yang Cai, Yuan Wu et al.

Recently, MBConv blocks, initially designed for efficiency in resource-limited settings and later adapted for cutting-edge image classification performances, have demonstrated significant potential in image classification tasks. Despite their success, their application in semantic segmentation has remained relatively unexplored. This paper introduces a novel adaptation of MBConv blocks specifically tailored for semantic segmentation. Our modification stems from the insight that semantic segmentation requires the extraction of more detailed spatial information than image classification. We argue that to effectively perform multi-scale semantic segmentation, each branch of a U-Net architecture, regardless of its resolution, should possess equivalent segmentation capabilities. By implementing these changes, our approach achieves impressive mean Intersection over Union (IoU) scores of 84.5% and 84.0% on the Cityscapes test and validation datasets, respectively, demonstrating the efficacy of our proposed modifications in enhancing semantic segmentation performance.

Kaiyi Zhang, Ximing Yang, Yuan Wu et al.

Given partial objects and some complete ones as references, point cloud completion aims to recover authentic shapes. However, existing methods pay little attention to general shapes, which leads to the poor authenticity of completion results. Besides, the missing patterns are diverse in reality, but existing methods can only handle fixed ones, which means a poor generalization ability. Considering that a partial point cloud is a subset of the corresponding complete one, we regard them as different samples of the same distribution and propose Structure Retrieval based Point Completion Network (SRPCN). It first uses k-means clustering to extract structure points and disperses them into distributions, and then KL Divergence is used as a metric to find the complete structure point cloud that best matches the input in a database. Finally, a PCN-like decoder network is adopted to generate the final results based on the retrieved structure point clouds. As structure plays an important role in describing the general shape of an object and the proposed structure retrieval method is robust to missing patterns, experiments show that our method can generate more authentic results and has a stronger generalization ability.