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
- Dacheng TaoSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works102 from public data
- Why ResNet Works? Residuals Generalize410
According to the obtained generalization bound, regularization terms should be introduced to control the magnitude of the norms of weight matrices not to increase too much, in practice, to ensure a good generalization ability, which justifies the technique of weight decay.
- Control Batch Size and Learning Rate to Generalize Well: Theoretical and Empirical Evidence247
A PAC-Bayes generalization bound for neural networks trained by SGD is proved, which has a positive correlation with the ratio of batch size to learning rate, which builds the theoretical foundation of the training strategy.
- 188
- DisPFL: Towards Communication-Efficient Personalized Federated Learning via Decentralized Sparse Training186
This work proposes a novel personalized federated learning framework in a decentralized (peer-to-peer) communication protocol named Dis-PFL, which employs personalized sparse masks to customize sparse local models on the edge to save the communication and computation cost.
- Collect and Select: Semantic Alignment Metric Learning for Few-Shot Learning164
A Semantic Alignment Metric Learning (SAML) method for few-shot learning that aligns the semantically relevant dominant objects through a ``collect-and-select'' strategy and demonstrates the generalization ability of SAML and gives a theoretical guarantee.
- Fast Spatio-Temporal Residual Network for Video Super-Resolution156
A novel fast spatio-temporal residual network (FSTRN) to adopt 3D convolutions for the video SR task in order to enhance the performance while maintaining a low computational load and significantly outperforms the current state-of-the-art methods.
- 144
- Artificial Neural Variability for Deep Learning: On Overfitting, Noise Memorization, and Catastrophic Forgetting75
A neural variable risk minimization framework and neural variable optimizers are devised to achieve ANV for conventional network architectures in practice, and the empirical studies demonstrate that NVRM can effectively relieve overfitting, label noise memorization, and catastrophic forgetting at negligible costs.
- Visual Semantics Allow for Textual Reasoning Better in Scene Text Recognition69
The first attempt to perform textual reasoning based on visual semantics in this paper is made by paralleling GTR to the language model in a segmentation-based STR baseline, which can effectively exploit the visual-linguistic complementarity via mutual learning.
- 66
- Topology-aware Generalization of Decentralized SGD60
The theory indicates that the generalizability of D-SGD is positively correlated with the spectral gap, and can explain why consensus control in initial training phase can ensure better generalization.
- Bridged Transformer for Vision and Point Cloud 3D Object Detection58
Bridged Transformer is proposed, an end-to-end architecture for 3D object detection which learns to identify 3D and 2D object bounding boxes from both points and image patches and surpasses state-of-the-art methods on SUN RGB-D and ScanNetV2 datasets.
- Channel Exchanging Networks for Multimodal and Multitask Dense Image Prediction53
Channel-Exchanging-Network (CEN) is proposed which is self-adaptive, parameter-free, and more importantly, applicable for multimodal and multitask dense image prediction and adaptively exchanges channels between subnetworks of different modalities.
- 53
- Robust Unlearnable Examples: Protecting Data Against Adversarial Learning52
Experiments show that the unlearnability brought by robust error-minimizing noise can effectively protect data from adversarial training in various scenarios.
- InFi44
This work proposes the first end-to-end learnable input filtering framework that covers most state-of-the-art methods and surpasses them in feature embedding with robust discriminability and designs and implements an input filtering system InFi, the first to support text and sensor signal inputs and model partitioning deployments widely adopted by under-resourced mobile systems.
- Siamese Network with Interactive Transformer for Video Object Segmentation43
A novel Siamese network with a specifically designed interactive transformer, called SITVOS, to enable effective context propagation from historical to current frames and to extract backbone features of both past and current frames, which enables feature reuse and is more efficient than existing methods.
- Knowledge Removal in Sampling-based Bayesian Inference39
This paper proposes the first machine unlearning algorithm for Markov chain Monte Carlo (MCMC), designed to provably characterize the learned knowledge from data, and shows that MCMC unlearning would not compromise the generalizability of the MCMC models.
- Understanding Generalization in Recurrent Neural Networks35
This work proposes to add random noise to the input data and proves a generalization bound for training with random noise, which is an extension of the former one, and discovers that Fisher-Rao norm for RNNs can be interpreted as a measure of gradient, and incorporating this gradient measure can tighten the bound.
- $E(2)$-Equivariant Vision Transformer33
This paper designs a Group Equivariant Vision Transformer via a novel, effective positional encoding operator, and proves that GE-ViT meets all the theoretical requirements of an equivariant neural network.
- Understanding Deep Learning via Decision Boundary29
It is discovered that the neural network (NN) with lower decision boundary (DB) variability has better generalizability and an upper bound of order is proved which is convenient to estimate without the requirement of labels and does not explicitly depend on the network size which is usually prohibitively large in deep learning.
- 29
- Self-Ensembling GAN for Cross-Domain Semantic Segmentation26
Despite its simplicity, this paper finds SE-GAN can significantly boost the performance of adversarial training and enhance the stability of the model, the latter of which is a common barrier shared by most adversarialTraining-based methods.
- 26
- Decentralized SGD and Average-direction SAM are Asymptotically Equivalent23
It is proved that D-SGD implicitly minimizes the loss function of an average-direction Sharpness-aware minimization (SAM) algorithm under general non-convex non-$\beta$-smooth settings, and this surprising asymptotic equivalence reveals an intrinsic regularization-optimization trade-off and three advantages of decentralization.
- Self-paced Supervision for Multi-source Domain Adaptation22
A multi-source domain adaptation method to gradually improve the adaptation ability of each source domain by producing more high-confident pseudo-labels with self-paced learning for conditional distribution alignment and outperforms the state-of-the-art methods.
- 21
- 18
- 17
- 12
- Learning to Generalize Provably in Learning to Optimize12
This paper theoretically establish an implicit connection between the local entropy and the Hessian and hence unify their roles in the handcrafted design of generalizable optimizers as equivalent metrics of the landscape flatness of loss functions and proposes to incorporate these two metrics as flatness-aware regularizers into the L2O framework in order to meta-train optimizers to learn to generalize.
- 12
- AI Driven Online Advertising: Market Design, Generative AI, and Ethics11
This paper proposes to create a forum for researchers, developers, users, ventures, policymakers, and other stakeholders to exchange ideas, research, innovations, etc with emphasis on AI driven mechanism design for distributing advertisements, generative AI for creating content in advertisements, and ethics issues.
- 11
- 11
- Omniforce: on human-centered, large model empowered and cloud-edge collaborative AutoML system9
OmniForce is introduced, a human-centered AutoML system that yields both human-assisted ML and ML-assisted human techniques, putting an AutoML system into practice and building AI in open-environment scenarios, such as using large models in industrial supply chains.
- 9
- InFi: End-to-End Learning to Filter Input for Resource-Efficiency in Mobile-Centric Inference8
This work designs and implements the first end-to-end learnable input filtering framework that covers most state-of-the-art methods and surpasses them in feature embedding with robust discriminability and shows that InFi outperforms strong baselines in applicability, accuracy, and efficiency.
- DeXposure: A Dataset and Benchmarks for Inter-protocol Credit Exposure in Decentralized Financial Networks6
A token-to-protocol model is developed using DefiLlama metadata to infer inter-protocol credit exposure from the token's stock dynamics, as reported by the protocols, as reported by the protocols.
- 6
- 6
- Global Nash Equilibrium in Non-convex Multi-player Game: Theory and Algorithms5
This paper takes conjugate transformation to the formulation of non-convex multi-player games, and casts the complementary problem into a variational inequality (VI) problem with a continuous pseudo-gradient mapping, and proves the existence condition of global NE: the solution to the VI problem satisfies a duality relation.
- Heterogeneous multi-task Gaussian Cox processes5
To circumvent the non-conjugate Bayesian inference in the MOGP modulated heterogeneous multi-task framework, the data augmentation technique is employed and derive a mean-field approximation to realize closed-form iterative updates for estimating model parameters.
- DeXposure-FM: A Time-series, Graph Foundation Model for Credit Exposures and Stability on Decentralized Financial Networks4
The DeXposure-FM is empirically validated on two machine learning benchmarks; it consistently outperforms the state-of-the-art approaches, including a graph foundation model and temporal graph neural networks, for measuring and forecasting inter-protocol credit exposure on DeFi networks.
- AlphaMemo: Structured Search-Process Memory for Self-Evolving Alpha Mining Agents4
Experiments on CSI 500 and S\&P 500 show improved out-of-sample performance and fixed-budget discovery efficiency, with ablations validating the roles of residual learning, confidence gating, AST-diff motifs, and veto memory.
- Inverse Learning of Black-Box Aggregator for Robust Nash Equilibrium4
This article considers a scenario when only the uncertain parameters and associated perturbed NE are available, but the players’ weights in the aggregator are unknown, and proposes data-based optimization to inversely learn players’ weights in the black-box aggregator.
- DICE: Data Influence Cascade in Decentralized Learning4
The first method to estimate DICE, the first method to estimate influence cascade over arbitrary neighbor hops in a decentralized environment is designed, suggesting the influence cascade is determined by an interplay of data, communication topology, and the curvature of loss landscape.
- 4
- 4
- 3
- 3
- 3
- Boosting Fair Classifier Generalization through Adaptive Priority Reweighing3
A novel adaptive reweighing method that prioritizes samples closer to the decision boundary and assigns a higher weight to improve the generalizability of fair classifiers to eliminate the impact of the distribution shifts between training and test data on model generalizability.
- 3
- 2
- 2
- 2
- 2
- Reject Decoding via Language-Vision Models for Text-to-Image Synthesis2
A reject decoding algorithm with tiny multi-modal models is exploited to enlarge the searching space and exclude the useless paths as early as possible and improve the synthesizing quality by consuming less time.
- Semantic-Aware Feature Aggregation for Few-Shot Image Classification2
This work directly model the interdependencies between prototype features and image regions, resulting in a novel Semantic-Aware Feature Aggregation (SAFA) framework that can place more weights on category-relevant image regions.
- Efficient Counterfactual Reasoning in ProbLog via Single World Intervention Programs1
This work makes complex counterfactual reasoning more computationally tractable and reliable, providing a crucial step towards developing more robust and explainable AI systems.
- 1
- 1
- Super-model ecosystem: A domain-adaptation perspective1
The theory suggests that the generalization is determined by a new measure that characterizes the domain discrepancy between the source domain and target domain, based on the covariance matrices and the shift of the converged local minimum.
- 1
- 1
- 1
- –
- Near-Constant Strong Violation and Last-Iterate Convergence for Online CMDPs via Decaying Safety Margins–
The Flexible safety Domain Optimization via Margin-regularized Exploration (FlexDOME) algorithm is proposed, the first to provably achieve near-constant $\tilde{O}(1)$ strong constraint violation alongside sublinear strong regret and non-asymptotic last-iterate convergence.
- –
- –
- –
- Beyond Global Divergences: A Local-Mass Perspective on Bayesian Inference–
Two mathematical tools are introduced and used: Mass Index for recording the polynomial and logarithmic decay scales of local mass, and regularised extended KL (RE-KL), a set-localised divergence that can be formulated in the presence of singular components.
- –
- DeXposure-Claw: An Agentic System for DeFi Risk Supervision–
DeXposure-Claw is introduced, a forecast-grounded agentic supervision system that routes LLM decisions through structured evidence and develops DeXposure-Bench, a six-axis evaluation harness, whose decision axis scores tickets against a regulator-aligned absolute-loss ground truth and an explicit false-intervention rate.
- Generalisation of RLHF under Reward Shift and Clipped KL Regularisation–
Generalisation theory for RLHF is developed, suggesting that the generalisation error stems from a sampling error from prompts and rollouts, a reward shift error, and a KL clipping error.
- PRISM: Parallel Reward Integration with Symmetry for MORL–
A Parallel Reward Integration with Symmetry (PRISM) algorithm that enforces reflectional symmetry as an inductive bias in aligning reward channels, and introduces ReSymNet, a theory-motivated model that reconciles temporal-frequency mismatches across objectives.
- Rationality Measurement and Theory for Reinforcement Learning Agents–
This paper proposes a suite of rationality measures and associated theory for reinforcement learning agents and suggests hypotheses on the benefits from regularisers and domain randomisation, as well as the harm from environment shifts.
- DeepSWIP: Quotient-WMC Counterfactuals for Neural Probabilistic Logic Programs–
DeepSWIP is introduced, a single-world counterfactual semantics for DeepProbLog programs, using neural materialization to reduce fixed-context neural predicates to ordinary ProbLog choices, apply Single World Intervention Programs (SWIPs), and compute counterfactuals by weighted model counting (WMC) over a single transformed program.
- –
- In LLM Reasoning, there is Irrationality on top of Value Misalignment–
It is argued that, even when an LLM has been well aligned in (post-)training, it may still fail to maximise the aligned value in reasoning, and the utility discrepancy between a model's deployed reasoning strategy and its rational counterpart whose responses maximise utility in the steepest direction is formalised.
- LLM-OSDA: An Optimal-Stopping Dynamic Auction for Native Advertising in Multi-Turn LLM Conversations–
- When a Reinforcement Learning Agent Encounters Unknown Unknowns–
Though an unknown unknown is surprising, it will be asymptotically properly discovered with decent speed and an affordable cost, collectively suggesting that though an unknown unknown is surprising, it will be asymptotically properly discovered with decent speed and an affordable cost.
- –
- –
- Ensuring Safety in an Uncertain Environment: Constrained MDPs via Stochastic Thresholds–
Stochastic Pessimistic-Optimistic Thresholding (SPOT), a novel model-based primal-dual algorithm for multiple constraints against stochastic thresholds, is designed, which is the first reinforcement learning algorithm that realises theoretical guaranteed performance in an uncertain environment where even thresholds are unknown.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Exploring High-Order Structure for Robust Graph Structure Learning–
This paper analyzes the adversarial attack on graphs from the perspective of feature smoothness which further contributes to an efficient new adversarial defensive algorithm for GNNs and proposes a novel algorithm that incorporates the high-order structural information into the graph structure learning.
- –
- –
- –
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. 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.