Author name cluster

Xuehao Wang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

9 papers

2 author rows

ICML Conference 2025 Conference Paper

Come Together, But Not Right Now: A Progressive Strategy to Boost Low-Rank Adaptation

Zhan Zhuang
Xiequn Wang
Wei Li
Yulong Zhang 0005
Qiushi Huang
Shuhao Chen
Xuehao Wang
Yanbin Wei

Low-rank adaptation (LoRA) has emerged as a leading parameter-efficient fine-tuning technique for adapting large foundation models, yet it often locks adapters into suboptimal minima near their initialization. This hampers model generalization and limits downstream operators such as adapter merging and pruning. Here, we propose CoTo, a progressive training strategy that gradually increases adapters’ activation probability over the course of fine-tuning. By stochastically deactivating adapters, CoTo encourages more balanced optimization and broader exploration of the loss landscape. We provide a theoretical analysis showing that CoTo promotes layer-wise dropout stability and linear mode connectivity, and we adopt a cooperative-game approach to quantify each adapter’s marginal contribution. Extensive experiments demonstrate that CoTo consistently boosts single-task performance, enhances multi-task merging accuracy, improves pruning robustness, and reduces training overhead, all while remaining compatible with diverse LoRA variants. Code is available at https: //github. com/zwebzone/coto.

Details

ICLR Conference 2025 Conference Paper

HeadMap: Locating and Enhancing Knowledge Circuits in LLMs

Xuehao Wang
Liyuan Wang
Binghuai Lin
Yu Zhang

Large language models (LLMs), through pretraining on extensive corpora, encompass rich semantic knowledge and exhibit the potential for efficient adaptation to diverse downstream tasks. However, the intrinsic mechanisms underlying LLMs remain unexplored, limiting the efficacy of applying these models to downstream tasks. In this paper, we explore the intrinsic mechanisms of LLMs from the perspective of knowledge circuits. Specifically, considering layer dependencies, we propose a layer-conditioned locating algorithm to identify a series of attention heads, which is a knowledge circuit of some tasks. Experiments demonstrate that simply masking a small portion of attention heads in the knowledge circuit can significantly reduce the model's ability to make correct predictions. This suggests that the knowledge flow within the knowledge circuit plays a critical role when the model makes a correct prediction. Inspired by this observation, we propose a novel parameter-efficient fine-tuning method called HeadMap, which maps the activations of these critical heads in the located knowledge circuit to the residual stream by two linear layers, thus enhancing knowledge flow from the knowledge circuit in the residual stream. Extensive experiments conducted on diverse datasets demonstrate the efficiency and efficacy of the proposed method. Our code is available at https://github.com/XuehaoWangFi/HeadMap.

Details

NeurIPS Conference 2025 Conference Paper

MoPFormer: Motion-Primitive Transformer for Wearable-Sensor Activity Recognition

Hao Zhang
Zhan Zhuang
Xuehao Wang
Xiaodong Yang
Yu Zhang

Human Activity Recognition (HAR) with wearable sensors is challenged by limited interpretability, which significantly impacts cross-dataset generalization. To address this challenge, we propose Motion-Primitive Transformer (MoPFormer), a novel self-supervised framework that enhances interpretability by tokenizing inertial measurement unit signals into semantically meaningful motion primitives and leverages a Transformer architecture to learn rich temporal representations. MoPFormer comprises two stages. The first stage is to partition multi-channel sensor streams into short segments and quantize them into discrete ``motion primitive'' codewords, while the second stage enriches those tokenized sequences through a context-aware embedding module and then processes them with a Transformer encoder. The proposed MoPFormer can be pre-trained using a masked motion-modeling objective that reconstructs missing primitives, enabling it to develop robust representations across diverse sensor configurations. Experiments on six HAR benchmarks demonstrate that MoPFormer not only outperforms state-of-the-art methods but also successfully generalizes across multiple datasets. More importantly, the learned motion primitives significantly enhance both interpretability and cross-dataset performance by capturing fundamental movement patterns that remain consistent across similar activities, regardless of dataset origin.

PDF Details

ICLR Conference 2025 Conference Paper

MTSAM: Multi-Task Fine-Tuning for Segment Anything Model

Xuehao Wang
Zhan Zhuang
Feiyang Ye 0001
Yu Zhang 0006

The Segment Anything Model (SAM), with its remarkable zero-shot capability, has the potential to be a foundation model for multi-task learning. However, adopting SAM to multi-task learning faces two challenges: (a) SAM has difficulty generating task-specific outputs with different channel numbers, and (b) how to fine-tune SAM to adapt multiple downstream tasks simultaneously remains unexplored. To address these two challenges, in this paper, we propose the Multi-Task SAM (MTSAM) framework, which enables SAM to work as a foundation model for multi-task learning. MTSAM modifies SAM's architecture by removing the prompt encoder and implementing task-specific no-mask embeddings and mask decoders, enabling the generation of task-specific outputs. Furthermore, we introduce Tensorized low-Rank Adaptation (ToRA) to perform multi-task fine-tuning on SAM. Specifically, ToRA injects an update parameter tensor into each layer of the encoder in SAM and leverages a low-rank tensor decomposition method to incorporate both task-shared and task-specific information. Extensive experiments conducted on benchmark datasets substantiate the efficacy of MTSAM in enhancing the performance of multi-task learning. Our code is available at https://github.com/XuehaoWangFi/MTSAM.

Details

ICML Conference 2025 Conference Paper

Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction

Yanbin Wei
Xuehao Wang
Zhan Zhuang
Yang Chen 0031
Shuhao Chen
Yulong Zhang 0005
James T. Kwok
Yu Zhang 0006

Message-passing graph neural networks (MPNNs) and structural features (SFs) are cornerstones for the link prediction task. However, as a common and intuitive mode of understanding, the potential of visual perception has been overlooked in the MPNN community. For the first time, we equip MPNNs with vision structural awareness by proposing an effective framework called Graph Vision Network (GVN), along with a more efficient variant (E-GVN). Extensive empirical results demonstrate that with the proposed frameworks, GVN consistently benefits from the vision enhancement across seven link prediction datasets, including challenging large-scale graphs. Such improvements are compatible with existing state-of-the-art (SOTA) methods and GVNs achieve new SOTA results, thereby underscoring a promising novel direction for link prediction.

Details

ICLR Conference 2025 Conference Paper

Sharpness-Aware Black-Box Optimization

Feiyang Ye 0001
Yueming Lyu
Xuehao Wang
Masashi Sugiyama
Yu Zhang 0006
Ivor W. Tsang

Black-box optimization algorithms have been widely used in various machine learning problems, including reinforcement learning and prompt fine-tuning. However, directly optimizing the training loss value, as commonly done in existing black-box optimization methods, could lead to suboptimal model quality and generalization performance. To address those problems in black-box optimization, we propose a novel Sharpness-Aware Black-box Optimization (SABO) algorithm, which applies a sharpness-aware minimization strategy to improve the model generalization. Specifically, the proposed SABO method first reparameterizes the objective function by its expectation over a Gaussian distribution. Then it iteratively updates the parameterized distribution by approximated stochastic gradients of the maximum objective value within a small neighborhood around the current solution in the Gaussian distribution space. Theoretically, we prove the convergence rate and generalization bound of the proposed SABO algorithm. Empirically, extensive experiments on the black-box prompt fine-tuning tasks demonstrate the effectiveness of the proposed SABO method in improving model generalization performance.

Details

ICLR Conference 2024 Conference Paper

Adaptive Stochastic Gradient Algorithm for Black-box Multi-Objective Learning

Feiyang Ye 0001
Yueming Lyu
Xuehao Wang
Yu Zhang 0006
Ivor W. Tsang

Multi-objective optimization (MOO) has become an influential framework for various machine learning problems, including reinforcement learning and multi-task learning. In this paper, we study the black-box multi-objective optimization problem, where we aim to optimize multiple potentially conflicting objectives with function queries only. To address this challenging problem and find a Pareto optimal solution or the Pareto stationary solution, we propose a novel adaptive stochastic gradient algorithm for black-box MOO, called ASMG. Specifically, we use the stochastic gradient approximation method to obtain the gradient for the distribution parameters of the Gaussian smoothed MOO with function queries only. Subsequently, an adaptive weight is employed to aggregate all stochastic gradients to optimize all objective functions effectively. Theoretically, we explicitly provide the connection between the original MOO problem and the corresponding Gaussian smoothed MOO problem and prove the convergence rate for the proposed ASMG algorithm in both convex and non-convex scenarios. Empirically, the proposed ASMG method achieves competitive performance on multiple numerical benchmark problems. Additionally, the state-of-the-art performance on the black-box multi-task learning problem demonstrates the effectiveness of the proposed ASMG method.

Details

NeurIPS Conference 2024 Conference Paper

Time-Varying LoRA: Towards Effective Cross-Domain Fine-Tuning of Diffusion Models

Zhan Zhuang
Yulong Zhang
Xuehao Wang
Jiangang Lu
Ying Wei
Yu Zhang

Large-scale diffusion models are adept at generating high-fidelity images and facilitating image editing and interpolation. However, they have limitations when tasked with generating images in dynamic, evolving domains. In this paper, we introduce Terra, a novel Time-varying low-rank adapter that offers a fine-tuning framework specifically tailored for domain flow generation. The key innovation of Terra lies in its construction of a continuous parameter manifold through a time variable, with its expressive power analyzed theoretically. This framework not only enables interpolation of image content and style but also offers a generation-based approach to address the domain shift problems in unsupervised domain adaptation and domain generalization. Specifically, Terra transforms images from the source domain to the target domain and generates interpolated domains with various styles to bridge the gap between domains and enhance the model generalization, respectively. We conduct extensive experiments on various benchmark datasets, empirically demonstrate the effectiveness of Terra. Our source code is publicly available on https: //github. com/zwebzone/terra.

PDF Details DOI

IJCAI Conference 2023 Conference Paper

Multi-Task Learning via Time-Aware Neural ODE

Feiyang Ye
Xuehao Wang
Yu Zhang
Ivor W. Tsang

Multi-Task Learning (MTL) is a well-established paradigm for learning shared models for a diverse set of tasks. Moreover, MTL improves data efficiency by jointly training all tasks simultaneously. However, directly optimizing the losses of all the tasks may lead to imbalanced performance on all the tasks due to the competition among tasks for the shared parameters in MTL models. Many MTL methods try to mitigate this problem by dynamically weighting task losses or manipulating task gradients. Different from existing studies, in this paper, we propose a Neural Ordinal diffeRential equation based Multi-tAsk Learning (NORMAL) method to alleviate this issue by modeling task-specific feature transformations from the perspective of dynamic flows built on the Neural Ordinary Differential Equation (NODE). Specifically, the proposed NORMAL model designs a time-aware neural ODE block to learn task-specific time information, which determines task positions of feature transformations in the dynamic flow, in NODE automatically via gradient descent methods. In this way, the proposed NORMAL model handles the problem of competing shared parameters by learning task positions. Moreover, the learned task positions can be used to measure the relevance among different tasks. Extensive experiments show that the proposed NORMAL model outperforms state-of-the-art MTL models.

PDF Details DOI