Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Shu‐Tao XiaSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works137 from public data
- Symmetric Cross Entropy for Robust Learning With Noisy Labels1,212
The proposed Symmetric cross entropy Learning (SL) approach simultaneously addresses both the under learning and overfitting problem of CE in the presence of noisy labels, and empirically shows that SL outperforms state-of-the-art methods.
- Adversarial Weight Perturbation Helps Robust Generalization923
This paper proposes a simple yet effective Adversarial Weight Perturbation (AWP) to explicitly regularize the flatness of weight loss landscape, forming a double-perturbation mechanism in the adversarial training framework that adversarially perturbs both inputs and weights.
- Improving Adversarial Robustness Requires Revisiting Misclassified Examples876
This paper proposes a new defense algorithm called MART, which explicitly differentiates the misclassified and correctly classified examples during the training, and shows that MART and its variant could significantly improve the state-of-the-art adversarial robustness.
- Understanding adversarial attacks on deep learning based medical image analysis systems567
It is found that medical DNN models can be more vulnerable to adversarial attacks compared to models for natural images, according to two different viewpoints, and surprisingly, medical adversarialattacks can be easily detected.
- 440
- Iterative Learning with Open-set Noisy Labels361
A novel iterative learning framework for training CNNs on datasets with open-set noisy labels that detects noisy labels and learns deep discriminative features in an iterative fashion and designs a Siamese network to encourage clean labels and noisy labels to be dissimilar.
- Adversarial Camouflage: Hiding Physical-World Attacks With Natural Styles285
Experimental evaluation shows that, in both digital and physical-world scenarios, adversarial examples crafted by Adversarial Camouflage are well camouflaged and highly stealthy, while remaining effective in fooling state-of-the-art DNN image classifiers.
- 217
- 166
- 159
- The EBS-A* algorithm: An improved A* algorithm for path planning146
An improved A*-based algorithm is proposed, called the EBS-A* algorithm, that introduces expansion distance, bidirectional search, and smoothing into path planning, and improves the path planning efficiency by 278% and reduces the number of critical nodes by 91.89%.
- Chaos is a Ladder: A New Theoretical Understanding of Contrastive Learning via Augmentation Overlap136
The theory suggests an alternative understanding of contrastive learning: the role of aligning positive samples is more like a surrogate task than an ultimate goal, and the overlapped augmented views create a ladder for Contrastive learning to gradually learn class-separated representations.
- 112
- 109
- Decoupled Networks92
A generic decoupled learning framework which models the intra-class variation and semantic difference independently and directly learn the operator from data is proposed.
- Fight Back Against Jailbreaking via Prompt Adversarial Tuning91
This paper proposes an approach named Prompt Adversarial Tuning (PAT) that trains a prompt control attached to the user prompt as a guard prefix that is effective against both grey-box and black-box attacks, while maintaining the model's utility on the benign task and incurring only negligible computational overhead.
- CFA: Class-Wise Calibrated Fair Adversarial Training91
This paper is the first to theoretically and empirically investigate the preference of different classes for adversarial configu-rations, including perturbation margin, regularization, and weight averaging, which customizes specific training con-figurations for each class automatically.
- 80
- What is Wrong with Perplexity for Long-context Language Modeling?77
Deeper insights are offered into the limitations of PPL and effective solutions for accurately evaluating and enhancing the long-context capabilities of LLMs are presented, leading to consistent improvements across diverse benchmarks.
- ContraNorm: A Contrastive Learning Perspective on Oversmoothing and Beyond67
A novel normalization layer called ContraNorm is proposed, inspired by the effectiveness of contrastive learning in preventing dimensional collapse, which implicitly shatters representations in the embedding space, leading to a more uniform distribution and a slighter dimensional collapse.
- 63
- Self-Ensemble Adversarial Training for Improved Robustness61
This work devise a simple but powerful Self-Ensemble Adversarial Training (SEAT) method for yielding a robust classifier by averaging weights of history models and discusses the relationship between the ensemble of predictions from different adversarially trained models and the prediction of weight-ensembled models.
- Improving Query Efficiency of Black-Box Adversarial Attack61
This paper explores the distribution of adversarial examples around benign inputs with the help of image structure information characterized by a Neural Process, and proposes a Neuralprocess based black-box adversarial attack (NP-Attack) in this paper.
- Analysis and Applications of Class-wise Robustness in Adversarial Training60
This work proposes to analyze the class-wise robustness in adversarial training and proposes a simple but effective attack method, named Temperature-PGD attack, which proposes to enlarge the robustness disparity among classes with a temperature factor on the confidence distribution of each image.
- 59
- 56
- Optimization-Induced Graph Implicit Nonlinear Diffusion48
It is shown that the learned representation can be formalized as the minimizer of an explicit convex optimization objective, and embedded prior properties to the equilibrium, as well as introducing skip connections to promote training stability.
- 43
- AttnGCG: Enhancing Jailbreaking Attacks on LLMs with Attention Manipulation42
This paper studies the vulnerabilities of transformer-based Large Language Models to jailbreaking attacks, focusing specifically on the optimization-based Greedy Coordinate Gradient (GCG) strategy, and introduces an enhanced method that manipulates models' attention scores to facilitate LLM jailbreaking, which is named AttnGCG.
- 40
- 32
- 31
- 27
- A Message Passing Perspective on Learning Dynamics of Contrastive Learning26
This paper shows that if the authors cast a contrastive objective equivalently into the feature space, then its learning dynamics admits an interpretable form, and theoretically characterize how contrastive learning gradually learns discriminative features with the alignment update and the uniformity update.
- Dirichlet Latent Variable Hierarchical Recurrent Encoder-Decoder in Dialogue Generation26
There is redundancy among the dimensions of latent variable, and the lengths and sentence patterns of the responses can be strongly correlated to each dimension of the latent variable.
- Generalist: Decoupling Natural and Robust Generalization24
This work decouple the natural generalization and the robust generalization from joint training and formulate different training strategies for each one to propose a hi-expert framework called Generalist, where base learners are simultaneously trained with task-aware strategies so that they can specialize in their own fields.
- Unbiased Multivariate Correlation Analysis24
UMC is a cumulative entropy based non-parametric multivariate correlation measure, which can capture both linear and non-linear correlations for groups of three or more variables, and employs a correction for chance using a statistical model of independence to address the issue of bias.
- Fw‐fuzz: A code coverage‐guided fuzzing framework for network protocols on firmware23
The Fw‐fuzz is proposed, a coverage‐guided and crossplatform framework for fuzzing network services running in the context of firmware on embedded architectures, which can generate more valuable test cases by introspecting program runtime information and using a genetic algorithm model.
- Generalization in Deep RL for TSP Problems via Equivariance and Local Search21
This work designs a novel deep RL approach with a focus on generalizability, consisting of a simple deep learning architecture, which learns with novel RL training techniques, and exploits equivariance to facilitate training.
- Student-t Process Regression with Student-t Likelihood21
Student-t Process Regression with Student-t Likelihood is theoretically show that TPRT is more robust to both input and target outliers than GPR and GPRT, and it is proved that both G PR and G PRT are special cases of TPRt.
- Staged Method of Code Similarity Analysis for Firmware Vulnerability Detection20
This work presents a method of staged firmware vulnerability detection based on code similarity, which can improve the accuracy of vulnerability detection and compared with state-of-the-art approaches.
- A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training19
This paper provides the first probabilistic characterization of AT through a unified understanding of robustness and generative ability and proposes a principled method to develop adversarial learning and sampling methods.
- Improving Generalization of Deep Reinforcement Learning-based TSP Solvers19
This work proposes a novel approach named MAGIC that includes a deep learning architecture and a DRL training method that defines a stochastic policy that sequentially generates a TSP solution.
- Proving Common Mechanisms Shared by Twelve Methods of Boosting Adversarial Transferability17
This paper summarizes the common mechanism shared by twelve previous transferability-boosting methods in a unified view, i.e., these methods all reduce game-theoretic interactions between regional adversarial perturbations.
- 16
- 15
- 15
- Wearable sensing for understanding and influencing human movement in ecological contexts14
The purpose, means, and impact of using wearable sensors to assess movement context, kinematics, and kinetics during locomotion, and how this information can be used to better understand and influence movement are discussed.
- 14
- 14
- 13
- Link sign prediction by Variational Bayesian Probabilistic Matrix Factorization with Student-t Prior12
A novel Variational Bayesian Probabilistic Matrix Factorization with Student-t prior model (TBMF) is proposed, which improves the prediction performance significantly compared with the state-of-the-art algorithms, especially when the observed links are few.
- 11
- A Value Set Analysis Refinement Approach Based on Conditional Merging and Lazy Constraint Solving11
A variable dependence analysis algorithm to divide program paths into subsets and only merge the states which satisfy the condition that the states are from the same subset, which reduces the imprecision from the merging operation.
- 9
- 9
- 9
- Bernoulli random forests: closing the gap between theoretical consistency and empirical soundness9
A Bernoulli random forests model (BRF), which intends to close the gap between the theoretical consistency and the empirical soundness of random forests classification and demonstrates the best performance among all theoretical random forests.
- 8
- 7
- Reparameterized Sampling for Generative Adversarial Networks7
This work proposes REP-GAN, a novel sampling method that allows general dependent proposals by REParameterizing the Markov chains into the latent space of the generator.
- 7
- 7
- SSL4RL: Revisiting Self-supervised Learning as Intrinsic Reward for Visual-Language Reasoning6
This work proposes SSL4RL, a novel framework that leverages self-supervised learning tasks as a source of verifiable rewards for RL-based fine-tuning, and establishes a versatile and effective paradigm for aligning multimodal models using verifiable, self-supervised objectives.
- 6
- 6
- 5
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- 4
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- Robust Survey Aggregation with Student-t Distribution and Sparse Representation3
A robust survey aggregation method based on Student-t distribution and sparse representation that achieves significant improvement over the state-of-the-art methods on both synthetic and real datasets is proposed.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- Generalist++: A Meta-learning Framework for Mitigating Trade-off in Adversarial Training1
Both theoretical analysis and extensive experiments demonstrate that Generalist achieves lower generalization error and significantly alleviates the trade-off problems compared with baseline methods, and suggest that Generalist provides a promising step toward developing fully robust classifiers in the future.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Robust Long-Tailed Learning via Label-Aware Bounded CVaR–
Two novel approaches based on CVaR (Conditional Value at Risk) are proposed to improve the performance of long-tailed learning with a solid theoretical ground and further design the optimal weight bounds for LAB-CVaR theoretically.
- –
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-16. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.