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
- Marimuthu Swami PalaniswamiSuggested from co-authorship
- Jiong JinSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works208 from public data
- Communication-efficient federated learning via knowledge distillation634
This work presents a communication-efficient federated learning method that saves a major fraction of communication cost, and reveals the advantage of reciprocal learning in machine knowledge transfer and the evolutional low-rank properties of deep model updates.
- 593
- Privacy and Robustness in Federated Learning: Attacks and Defenses582
A comprehensive survey on privacy and robustness in FL over the past five years is conducted and a concise introduction to the concept of FL is provided and a unique taxonomy covering privacy attacks and defenses is provided.
- Defending ChatGPT against jailbreak attack via self-reminders515
This work systematically documents the threats posed by jailbreak attacks, introduces and analyses a dataset for evaluating defensive interventions and proposes the psychologically inspired self-reminder technique that can efficiently and effectively mitigate against jailbreaks without further training.
- Local Differential Privacy-Based Federated Learning for Internet of Things449
This article proposes to integrate federated learning and local differential privacy (LDP) to facilitate the crowdsourcing applications to achieve the machine learning model, and proposes four LDP mechanisms to perturb gradients generated by vehicles.
- 434
- A federated graph neural network framework for privacy-preserving personalization347
A federated GNN framework named FedPerGNN, which can collaboratively train GNN models based on decentralized graphs inferred from local data and introduces a privacy-preserving graph expansion protocol to incorporate high-order information under privacy protection.
- Data Poisoning Attacks on Federated Machine Learning310
This work attempts to attacking a popular federated multitask learning framework, which uses a general multitask learning framework to handle statistical challenges in the federated learning setting, and proposes a novel systems-aware optimization method, called as attack on federated learning (AT2FL), to efficiently derive the implicit gradients for poisoned data, and further attain optimal attack strategies in the federated machine learning.
- Narcissus: A Practical Clean-Label Backdoor Attack with Limited Information287
An algorithm to launch clean-label backdoor attacks using only samples from the target class and public out-of-distribution data is designed and the synthesized Narcissus trigger contains durable features as persistent as the original target class features.
- 255
- 241
- Collaborative Fairness in Federated Learning229
This work investigates the collaborative fairness in FL, and proposes a novel Collaborative Fair Federated Learning (CFFL) framework which utilizes reputation to enforce participants to converge to different models, thus achieving fairness without compromising the predictive performance.
- 220
- FLEAM: A Federated Learning Empowered Architecture to Mitigate DDoS in Industrial IoT209
This article combines federated learning (FL) and fog/edge computing to combat malicious codes and trains a global optimized model based on distributed datasets of collaborators while removing the data and communication constraints.
- FedBERT : When Federated Learning Meets Pre-training182
This work proposes a new learning approach, FedBERT, that takes advantage of the federated learning and split learning approaches, resorting to pre-training BERT in a federated way and can prevent sharing the raw data information and obtain excellent performance.
- Privacy for Free: How does Dataset Condensation Help Privacy?173
This work identifies that dataset condensation (DC) which is originally designed for improving training efficiency is also a better solution to replace the traditional data generators for private data generation, thus providing privacy for free.
- A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment139
This paper introduces the concept of "full-stack" safety to systematically consider safety issues throughout the entire process of LLM training, deployment, and eventual commercialization, and represents the first safety survey to encompass the entire lifecycle of LLMs.
- 139
- TARGET: Federated Class-Continual Learning via Exemplar-Free Distillation136
A novel method called TARGET (federatTed clAss-continual leaRninG via Exemplar-free disTillation via Exemplar-free disTillation), which alleviates catastrophic forgetting in FCCL while preserving client data privacy.
- Federated Model Distillation with Noise-Free Differential Privacy135
This paper proposes a novel framework called FEDMD-NFDP, which applies a Noise-FreeDifferential Privacy (NFDP) mechanism into a federated model distillation framework, and demonstrates the feasibility of this framework by considering both IID and Non-IID settings, heterogeneous model architectures, and unlabelled public datasets from a different distribution.
- Protecting Intellectual Property of Language Generation APIs with Lexical Watermark134
This work presents a novel watermarking method for text generation APIs by conducting lexical modification to the original outputs that achieves better identifiable performance in terms of p-value, with fewer semantic losses.
- Detecting, Explaining, and Mitigating Memorization in Diffusion Models133
This work introduces a straightforward yet effective method for detecting memorized prompts by inspecting the magnitude of text-conditional predictions, and proposes two strategies to mitigate memorization by leveraging the magnitude of text-conditional predictions, either through minimization during inference or filtering during training.
- Vertically Federated Graph Neural Network for Privacy-Preserving Node Classification132
This paper proposes VFGNN, a federated GNN learning paradigm for privacy-preserving node classification task under data vertically partitioned setting, which can be generalized to existing GNN models.
- Model Extraction and Adversarial Transferability, Your BERT is Vulnerable!125
This work presents how an adversary can steal a BERT-based API service (the victim/target model) on multiple benchmark datasets with limited prior knowledge and queries, and shows that the extracted model can lead to highly transferable adversarial attacks against the victim model.
- Fog-Empowered Anomaly Detection in IoT Using Hyperellipsoidal Clustering118
This work proposes a novel anomaly detection method, called Fog-Empowered anomaly detection, by harnessing the processing power of the Fog computing platform and using an efficient hyperellipsoidal clustering algorithm to achieve anomaly detection in a timely manner.
- No One Left Behind: Inclusive Federated Learning over Heterogeneous Devices110
The core idea of InclusiveFL is to assign models of different sizes to clients with different computing capabilities, bigger models for powerful clients and smaller ones for weak clients, and an effective method to share the knowledge among local models with different sizes is proposed.
- Differentially Private Representation for NLP: Formal Guarantee and An Empirical Study on Privacy and Fairness108
Experimental results on benchmark datasets under various parameter settings demonstrate that DPNR largely reduces privacy leakage without significantly sacrificing the main task performance.
- Reconstructive Neuron Pruning for Backdoor Defense107
This paper proposes a novel defense called Reconstructive Neuron Pruning (RNP) to expose and prune backdoor neurons via an unlearning and then recovering process and shows that such an asymmetric process on only a few clean samples can effectively expose andPrune the backdoor neurons implanted by a wide range of attacks.
- CATER: Intellectual Property Protection on Text Generation APIs via Conditional Watermarks106
A novel Conditional wATERmarking framework (CATER) is proposed for protecting the IP of text generation APIs and it is proved that it is infeasible for even the savviest attacker to reveal the used watermarks from a large pool of potential word pairs based on statistical inspection.
- Differential Private Knowledge Transfer for Privacy-Preserving Cross-Domain Recommendation99
This work proposes a novel two stage based privacy-preserving CDR framework, PriCDR, and a novel heterogeneous CDR model (HeteroCDR), which uses deep auto-encoder and deep neural network to model the published source rating matrix and target rating matrix respectively.
- ${\sf FederBoost}$: Private Federated Learning for GBDT96
Experimental results show that both vertical and horizontalinline-formula-based federated learning solutions are 4-5 orders of magnitude faster than the state-of-the-art solutions for federated decision tree training; hence offering practical solutions for industrial applications.
- UltraRE: Enhancing RecEraser for Recommendation Unlearning via Error Decomposition91
This paper rethink RecEraser from an ensemble-based perspective and focus on its three potential losses, i.e., redundancy, relevance, and combination, and proposes a new framework named UltraRE, which simplifies RecEraser and powers RecEraser for recommendation tasks.
- Integration of large language models and federated learning85
This review aims to deeply explore the integration of LLMs and FL by proposing a research framework dividing the fusion of LLMs and FL into three parts: the combination of LLM sub-technologies with FL, the integration of FL sub-technologies with LLMs, and the overall merger of LLMs and FL.
- Privacy-Preserving Collaborative Deep Learning with Application to Human Activity Recognition82
A two-stage privacy-preserving scheme designed to mitigate maximum a posteriori (MAP) estimation attacks by perturbing each participant's data through a nonlinear function called repeated Gompertz (RG) and a novel LSTM-CNN model combining the merits of Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN).
- 73
- Defending against Backdoor Attacks in Natural Language Generation71
By giving a formal definition of backdoor attack and defense, this work finds that testing the backward probability of generating sources given targets yields effective defense performance against all different types of attacks, and is able to handle the one-to-many issue in many NLG tasks such as dialog generation.
- 69
- ASSET: Robust Backdoor Data Detection Across a Multiplicity of Deep Learning Paradigms67
A new detection method called Active Separation via Offset (ASSET), which actively induces different model behaviors between the backdoor and clean samples to promote their separation and is superior to existing methods in terms of consistency of defensive performance across different attacks and robustness to changes in poison ratios.
- Accelerated Federated Learning with Decoupled Adaptive Optimization64
A momentum decoupling adaptive optimization method, FedDA, is developed to fully utilize the global momentum on each local iteration and accelerate the training convergence and overcome the possible inconsistency caused by adaptive optimization methods.
- 63
- 62
- CO-SPY: Combining Semantic and Pixel Features to Detect Synthetic Images by AI57
This work proposes a novel framework, CO-SPY, that first enhances existing semantic features and artifact features and adaptively integrates them to achieve more general and robust synthetic image detection, and creates Co-SpyBench, a comprehensive dataset comprising 5 real image datasets and 22 state-of-the-art generative models.
- FedCTR: Federated Native Ad CTR Prediction with Cross-platform User Behavior Data55
A federated native ad CTR prediction method named FedCTR is proposed, which can learn user-interest representations from cross-platform user behaviors in a privacy-preserving way and proposes a federated framework for collaborative model training with distributed models and user behaviors.
- How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning54
A fair and differentially private decentralised deep learning framework called FDPDDL is built, which enables parties to derive more accurate local models in a fair and private manner by using a developed two-stage scheme.
- Towards Differentially Private Text Representations51
A new deep learning framework under an untrusted server setting, which includes a novel local differentially private (LDP) protocol to reduce the impact of privacy parameter ε on accuracy, and provide enhanced flexibility in choosing randomization probabilities for LDP.
- Unlearning through Knowledge Overwriting: Reversible Federated Unlearning via Selective Sparse Adapter45
A method named FUSED is proposed, which first identifies critical layers by analyzing each layer’s sensitivity to knowledge and constructs sparse unlearning adapters for sensitive ones, which enables FUSED to mitigate the effects of indiscriminate unlearning.
- 44
- Is Heterogeneity Notorious? Taming Heterogeneity to Handle Test-Time Shift in Federated Learning44
This work proposes a contrastive learning-based FL framework, namely FedICON, to capture invariant knowledge among heterogeneous clients and consistently tune the model to adapt to test data and extensive experiments validate the effectiveness of the proposed FedICON in taming heterogeneity to handle test-time.
- 44
- Pushing the Limits of ChatGPT on NLP Tasks43
Using the proposed assemble of techniques, this work is able to significantly boost the performance of ChatGPT on the selected NLP tasks, achieving performances comparable to or better than supervised baselines, or even existing SOTA performances.
- Reducing Communication for Split Learning by Randomized Top-k Sparsification43
This paper investigates multiple communication reduction methods for split learning, including cut layer size reduction, top-k sparsification, quantization, and L1 regularization and proposes randomized top- k sparsifiers, which achieves a better model performance under the same compression level.
- Towards Building a Robust and Fair Federated Learning System.43
This work proposes a novel Robust and Fair Federated Learning (RFFL) framework which utilizes reputation scores to address both issues, thus ensuring the high-contributing participants are rewarded with high-performing models while the low- or non-contributioning participants can be detected and removed.
- 42
- EdgeWatch: Collaborative Investigation of Data Integrity at the Edge based on Blockchain41
This paper proposes a novel scheme named EdgeWatch to enable robust and collaborative EDI investigation in a decentralized manner based on blockchain and evaluates it against three representative schemes experimentally to demonstrate the high precision, efficiency, and robustness of EdgeWatch.
- 40
- 38
- Defending against Indirect Prompt Injection by Instruction Detection37
This paper proposes InstructDetector, a novel detection-based approach that leverages the behavioral states of LLMs to identify potential IPI attacks, and demonstrates the hidden states and gradients from intermediate layers provide highly discriminative features for instruction detection.
- 37
- Privacy-Preserving Anomaly Detection in Cloud Manufacturing Via Federated Transformer36
This article customizes a weakly supervised edge computing anomaly detection framework, i.e., federated learning-based transformer framework (FedAnomaly), to deal with the anomaly detection problem in cloud manufacturing and introduces federatedLearning (FL) framework that allows edge devices to train an anomaly detection model in collaboration with the cloud without compromising privacy.
- RAIN: RegulArization on Input and Network for Black-Box Domain Adaptation35
This paper proposes a novel approach named RAIN (RegulArization on Input and Network) for Black-Box domain adaptation from both input-level and network-level regularization, which alleviates overfitting on the source domain by learning diverse target representations.
- FedSkip: Combatting Statistical Heterogeneity with Federated Skip Aggregation35
This work introduces a data-driven approach called FedSkip to improve the client optima by periodically skipping federated averaging and scattering local models to the cross devices and provides theoretical analysis of the possible benefit from FedSkip.
- 35
- A fast and scalable authentication scheme in IOT for smart living35
This paper uses fog computing and microservice to push certificate authority (CA) functions to the proximity of data sources, which can minimize attack surfaces and authentication latency, and result in a fast and scalable scheme in authenticating a large volume of resource-limited devices.
- Stretching Each Dollar: Diffusion Training from Scratch on a Micro-Budget33
A deferred masking strategy is proposed that preprocesses all patches using a patch-mixer before masking, thus significantly reducing the performance degradation with masking, making it superior to model downscaling in reducing computational cost.
- How to Trace Latent Generative Model Generated Images without Artificial Watermark?33
A latent inversion based method called LatentTracer is designed to trace the generated images of the inspected model by checking if the examined images can be well-reconstructed with an inverted latent input and suggests the intriguing possibility that today's latent generative generated images are naturally watermarked by the decoder used in the source models.
- FT-Shield: A Watermark Against Unauthorized Fine-tuning in Text-to-Image Diffusion Models31
The proposed FT-Shield addresses copyright protection challenges by designing new watermark generation and detection strategies and introduces an innovative algorithm for watermark generation that integrates a Mixture of Experts (MoE) approach for watermark detection.
- FedMef: Towards Memory-Efficient Federated Dynamic Pruning31
The proposed FedMef introduces the budget-aware extrusion that maintains pruning efficiency while preserving post-pruning performance by salvaging crucial information from parameters marked for pruning within a given budget, which is particularly beneficial for deploying FL to memory-limited devices.
- 31
- MLLM-as-a-Judge for Image Safety without Human Labeling30
A MLLM-based method includes objectifying safety rules, assessing the relevance between rules and images, making quick judgments based on debiased token probabilities with logically complete yet simplified precondition chains for safety rules, and conducting more in-depth reasoning with cascaded chain-of-thought processes if necessary.
- DEJA VU: Continual Model Generalization For Unseen Domains30
RaTP is a framework that focuses on improving models' target domain generalization (TDG) capability, while also achieving effective target domain adaptation (TDA) capability right after training on certain domains and forgetting alleviation (FA) capability on past domains.
- PrivateRec: Differentially Private Model Training and Online Serving for Federated News Recommendation28
A federated news recommendation method for achieving better utility in model training and online serving under a DP guarantee and a privacy-preserving online serving mechanism based on the idea of decomposing user embeddings with public basic vectors and perturbing the lower-dimensional combination coefficients is proposed.
- Differentially private locality sensitive hashing based federated recommender system28
This paper conducts privacy analysis on the existing locality sensitive hashing (LSH) approach based privacy‐preserving recommender system and shows how an attacker can retrieve user's information under such aRecommender system.
- Bridging Model Heterogeneity in Federated Learning via Uncertainty-based Asymmetrical Reciprocity Learning26
FedType introduces small identical proxy models for clients, serving as agents for information exchange, ensuring model security, and achieving efficient communication simultaneously, redefines federated learning paradigms by bridging model heterogeneity.
- MAS: Towards Resource-Efficient Federated Multiple-Task Learning26
This work proposes the first FL system to effectively coordinate and train multiple simultaneous FL tasks, and presents a new approach, MAS (Merge and Split), to optimize the performance of training multiple simultaneous FL tasks.
- 26
- 26
- Privacy-preserving collaborative fuzzy clustering26
Results using seven cluster validity indices, root mean squared error (RMSE) and accuracy ratio show that clustering results based on two-stage-perturbed data are comparable to the clusteringResults based on raw data confirm the utility of the privacy-preserving scheme when used with either FCM or HCM.
- Notice of Violation of IEEE Publication Principles: DP-SIGNSGD: When Efficiency Meets Privacy and Robustness25
This paper presents an improved method called DP-SIGNSGD, which can meet all the aforementioned properties of SIGNSGD and proposes an error-feedback variant of DP-sigNSGD to improve accuracy.
- Towards Fair and Decentralized Privacy-Preserving Deep Learning with Blockchain25
The proposed DPPDL framework provides a viable solution to detect and isolate the cheating party in the system, and simultaneously provides fairness and privacy by proposing two novel algorithms: initial benchmarking and privacy-preserving collaborative deep learning.
- 24
- How to Sift Out a Clean Data Subset in the Presence of Data Poisoning?24
Examining the impact of poisoned samples on defenses when they are mistakenly mixed into the base set suggests that sifting out a base set with high precision is key to these defenses' performance, and proposes a practical countermeasure, Meta-Sift.
- Differentially Private Knowledge Distillation for Mobile Analytics24
A Differentially Private Knowledge Distillation framework to enable on-device deep learning as well as preserve training data privacy and shows that DPKD can not only benefit from the distilled knowledge but also provide a strong differential privacy guarantee with only marginal decreases in accuracy.
- V2X-Boosted Federated Learning for Cooperative Intelligent Transportation Systems with Contextual Client Selection23
A contextual client selection pipeline that uses Vehicle-to-Everything (V2X) messages to select clients based on the predicted communication latency and outperforms baselines on various datasets, particularly in non-iid settings is proposed.
- SacFL: Self-Adaptive Federated Continual Learning for Resource-Constrained End Devices22
A novel FCL framework named self-adaptive federated CL (SacFL) is proposed, which leverages contrastive learning to introduce an autonomous data shift detection mechanism, enabling it to discern whether a new task has emerged and whether it is a benign task.
- ResFed: Communication-Efficient Federated Learning With Deep Compressed Residuals22
A residual-based federated learning framework (ResFed) that transmits residuals instead of gradients or weights in networks, which outperforms current methods (weight- or gradient-based federated learning) on federated data sets and can be applied to reduce communication costs for server-to-client communication.
- When Alignment Fails: Multimodal Adversarial Attacks on Vision-Language-Action Models21
This paper introduces VLA-Fool, a comprehensive study of multimodal adversarial robustness in embodied VLA models under both white-box and black-box settings, and incorporates a VLA-aware semantic space into linguistic prompts, developing the first automatically crafted and semantically guided prompting framework.
- Evaluating and Mitigating IP Infringement in Visual Generative AI20
A revised generation paradigm is developed that has the capability to recognize generated content that may be infringing on intellectual property rights, and mitigate such infringement by employing guidance methods throughout the diffusion process without retrain or fine-tune the pretrained models.
- 20
- FedP3: Federated Personalized and Privacy-friendly Network Pruning under Model Heterogeneity20
This paper presents an effective and adaptable federated framework FedP3, representing Federated Personalized and Privacy-friendly network Pruning, tailored for model heterogeneity scenarios, and offers a theoretical interpretation of FedP3 and its locally differential-private variant, DP-FedP3, and theoretically validate their efficiencies.
- 20
- Revisiting Data-Free Knowledge Distillation with Poisoned Teachers19
This work proposes Anti-Backdoor Data-Free KD (ABD), the first plug-in defensive method for data-free KD methods to mitigate the chance of potential backdoors being transferred and empirically evaluates the effectiveness of the proposed ABD in diminishing transferred backdoor knowledge while maintaining compatible downstream performances as the vanilla KD.
- InOR-Net: Incremental 3-D Object Recognition Network for Point Cloud Representation19
A novel Incremental 3-D Object Recognition Network (i.e., InOR-Net) is developed, which could recognize new classes of 2-D objects continuously by overcoming the catastrophic forgetting on old classes and a dual adaptive fairness compensations strategy is designed to overcome the forgetting brought by class imbalance.
- 19
- 18
- 18
- 18
- 17
- Cross-Network Social User Embedding with Hybrid Differential Privacy Guarantees17
A Cross-network Social User Embedding framework, namely DP-CroSUE, is proposed to learn the comprehensive representations of users in a privacy-preserving way and a novel cross-network GCN embedding model is designed to transfer knowledge across networks through those aligned users.
- The Perils of Learning From Unlabeled Data: Backdoor Attacks on Semi-supervised Learning16
This paper designs a backdoor poisoning attack on SSL that can be conducted by a weak adversary with no knowledge of the target SSL pipeline, and shows that by poisoning only 0.2% of the unlabeled training data, the adversary can successfully cause misclassification on more than 80% of test inputs.
- Towards Adversarially Robust Continual Learning16
This work is the first to study adversarial robustness in continual learning and proposes a novel method called Task-Aware Boundary Augmentation (TABA) to boost the robustness of continual learning models.
- 16
- 16
- 15
- 15
- An Unsupervised PM2.5 Estimation Method With Different Spatio-Temporal Resolutions Based on KIDW-TCGRU15
An effectively unsupervised PM2.5 estimation method based on time distributed convolutional gated recurrent unit (TCGRU) and an interpolation methodbased on $k$ -nearest neighbor inverse distance weighted (KIDW) to solve the problem of air quality monitoring in the areas without stations by different interpolation methods.
- DD-Ranking: Rethinking the Evaluation of Dataset Distillation14
DD-Ranking, a unified evaluation framework, along with new general evaluation metrics to uncover the true performance improvements achieved by different methods are proposed, which provide a more comprehensive and fair evaluation standard for future research advancements.
- FedSAC: Dynamic Submodel Allocation for Collaborative Fairness in Federated Learning14
FedSAC, a novel Federated learning framework with dynamic Submodel Allocation for Collaborative fairness, backed by a theoretical convergence guarantee, is proposed, which ensures fairness by tailoring rewards to individual clients based on their contributions.
- 14
- Minimum Topology Attacks for Graph Neural Networks14
An attack model, named MiBTack, based on a dynamic projected gradient descent algorithm, which can effectively solve the involving non-convex constraint optimization on discrete topology and can be used to measure node robustness, which is beyond what the current fixed-budget topology attacks can offer.
- Reliable and Privacy-Preserving Task Matching in Blockchain-Based Crowdsourcing14
This paper proposes a Reliable and Privacy-preserving Task Matching scheme (RPTM) for crowdsourcing that supports multi-keyword fuzzy matching by exploiting locality sensitive hashing and Bloom filter, which can tolerate keyword spelling errors and different expression formats.
- 13
- Self-Comparison for Dataset-Level Membership Inference in Large (Vision-)Language Model12
It is found that a member prefix followed by a non-member suffix (paraphrased from a member suffix) can further trigger the model's memorization on training data, and a novel dataset-level membership inference method based on Self-Comparison is proposed.
- 12
- DIAGNOSIS: Detecting Unauthorized Data Usages in Text-to-image Diffusion Models12
This paper alters the protected image dataset by adding unique contents on the images such as stealthy image wrapping functions that are imperceptible to human vision but can be captured and memorized by diffusion models to detect unauthorized data usages.
- 12
- 11
- Multiple-Agent Deep Reinforcement Learning for Avatar Migration in Vehicular Metaverses11
A new avatar task migration framework for vehicular Metaverses is proposed, formulated as a Partially Observable Markov Decision Process (POMDP), and applied a Multi-Agent Deep Reinforcement Learning (MADRL) algorithm to dynamically make migration decisions for avatar tasks.
- EdgeDis: Enabling Fast, Economical, and Reliable Data Dissemination for Mobile Edge Computing11
A novel scheme named EdgeDis is presented that coordinates data dissemination by distributed consensus among those servers that reduces dissemination costs and performance loss caused by dissemination failures and is compared with baseline and state-of-the-art schemes.
- 10
- International Workshop on Federated Learning for Distributed Data Mining10
This workshop will encourage ideas exchange on the trustworthiness, scalability, robustness, and broad applications of Federated Learning in data mining.
- 10
- Clustering-based Efficient Privacy-preserving Face Recognition Scheme without Compromising Accuracy10
Recently, biometric identification has been extensively used for border control and some face recognition systems have been designed based on Internet of Things, but the rich personal information provided by these systems cannot be compared to that provided by face-based systems.
- Six-CD: Benchmarking Concept Removals for Text-to-image Diffusion Models9
In this benchmark, a thorough evaluation of concept removals is conducted by introducing a new dataset, Six-CD, along with a novel evaluation metric, and conducting a thorough evaluation of concept removal methods.
- 9
- 9
- 9
- 9
- Protecting Split Learning by Potential Energy Loss8
The potential energy loss is proposed to make the forward embeddings of split learning more 'complicated', by pushing embeddings of the same class towards the decision boundary, so that it is hard for the attacker to learn from the forward embeddings.
- Is Synthetic Image Useful for Transfer Learning? An Investigation into Data Generation, Volume, and Utilization8
A novel two-stage framework called bridged transfer is introduced, which initially employs synthetic images for fine-tuning a pre-trained model to improve its transferability and subsequently uses real data for rapid adaptation and a dataset style inversion strategy to improve the stylistic alignment between synthetic and real images.
- Privacy-preserving machine learning and data aggregation for Internet of Things8
This thesis systematically investigates state-of-the-art techniques for privacy-preserving machine learning and data aggregation in a range of IoT applications and tackles security challenges in collaborative anomaly detection with a two-stage scheme called RG+RT.
- 7
- 7
- 7
- MLAN: Language-Based Instruction Tuning Preserves and Transfers Knowledge in Multimodal Language Models6
The comprehensive evaluation shows that the text-heavy instruction tuning approach is able to perform on-par with traditional vision-heavy mixtures on both modalities across 12 general datasets while using as low as half the total training tokens.
- RLCP: A Reinforcement Learning-based Copyright Protection Method for Text-to-Image Diffusion Model6
This work introduces a novel copyright metric grounded in legal precedents and employs the Denoising Diffusion Policy Optimization (DDPO) framework to minimize copyright-infringing content while preserving image quality in text-to-image diffusion models.
- 6
- Federated Learning under Partially Class-Disjoint Data via Manifold Reshaping6
This work proposes a manifold reshaping approach called FedMR, which adds two interplaying losses to the vanilla federated learning: one is intra-class loss to decorrelate feature dimensions for anti-collapse; and the other one is inter-class loss to guarantee the proper margin among categories in the feature expansion.
- Alteration-free and Model-agnostic Origin Attribution of Generated Images6
An alteration-free and model-agnostic origin attribution method via input reverse-engineering on image generation models, which effectively distinguishes between generated images from a specific generative model and other images, including those generated by different models and real images.
- FedWon: Triumphing Multi-domain Federated Learning Without Normalization6
This research addresses a crucial yet frequently overlooked problem known as multi-domain FL, where clients’ data originate from diverse domains with distinct feature distributions, as opposed to label distributions, and eliminates all normalizations in FL and reparameterizes convolution layers with scaled weight standardization.
- Contour Accentuation for Transfer Learning-Based Ship Recognition Method6
In practical experiments, the results showed that the proposed transfer learning-based ship recognition method with contour accentuation can obtain higher accuracy, and the accuracy of the proposed method was 97.79%.
- On the Hardness of Robustness Transfer: A Perspective from Rademacher Complexity over Symmetric Difference Hypothesis Space5
The results initiate the study of the generalization theory of adversarially robust domain adaptation, and could shed lights on distributed adversarial robust learning from heterogeneous sources, e.g., federated learning scenario.
- Privacy-preserving design of graph neural networks with applications to vertical federated learning5
This paper presents an end-to-end GRL framework in the VFL setting called VESPER, which is built upon a general privatization scheme termed perturbed message passing (PMP) that allows the privatization of many popular graph neural architectures.
- SplitGNN: Splitting GNN for Node Classification with Heterogeneous Attention5
A split learning-based GNN (SplitGNN), where this model is divided into two sub-models: the local GNN model includes all the private data related computation to generate local node embeddings, whereas the global model calculates global embeddings by aggregating all the participants' local embeddings.
- 5
- IDEAL: Query-Efficient Data-Free Learning from Black-box Models5
This work proposes a novel method called Query-Efficient Knowledge Distillation (QEKD), which aims to query-efficiently learn from black-box model APIs to train a good student without any real data.
- ResFed: Communication Efficient Federated Learning by Transmitting Deep Compressed Residuals5
This work introduces a residual-based federated learning framework (ResFed), where residuals rather than model parameters are transmitted in communication networks for training, and integrates two pairs of shared predictors for the model prediction in both server-to-client and client- to-server communication.
- 5
- 4
- EnTruth: Enhancing the Traceability of Unauthorized Dataset Usage in Text-to-image Diffusion Models with Minimal and Robust Alterations4
This work is the first to investigate the positive application of memorization and use it for copyright protection, which turns a curse into a blessing and offers a pioneering perspective for unauthorized usage detection in generative models.
- 4
- 4
- 4
- Distributed Privacy-Preserving Prediction.4
This work demonstrates a generally applicable Distributed Privacy-Preserving Prediction (DPPP) framework, in which instead of sharing more sensitive data or model parameters, an untrusted aggregator combines only multiple models' predictions under provable privacy guarantee.
- Asymmetrical Reciprocity-based Federated Learning for Resolving Disparities in Medical Diagnosis3
This work proposes a novel cross-silo federated learning framework, named FedHelp, aimed at alleviating geographic health disparities and fortifying the diagnostic capabilities of underserved regions, and introduces a novel asymmetric dual knowledge distillation module to manage the issue of asymmetric reciprocity.
- 3
- 3
- DADFNet: Dual Attention and Dual Frequency-Guided Dehazing Network for Video-Empowered Intelligent Transportation3
A dual attention and dual frequency-guided dehazing network (termed DADFNet) for real-time visibility enhancement and improvement in detection accuracy is proposed, which is highly efficient for deployment in intelligent transportation systems.
- 3
- 2
- ExpVG: Investigating the Design Space of Visual Grounding in Multimodal Large Language Model2
This paper presents a comprehensive study of various design choices that impact the VG performance of MLLMs using LLaVA-1.5, and explores different visual grounding paradigms in MLLMs, identifying the most effective design and providing insights.
- 2
- Copyright-Aware Incentive Scheme for Generative Art Models Using Hierarchical Reinforcement Learning2
This is the first technical work that introduces to incentive contributors and protect their copyrights by compensating them and a hierarchical budget allocation method based on reinforcement learning to determine the budget for each round and the remuneration of the data holder based on the data holder's contribution and copyright loss in each round.
- 2
- 2
- Rendering-Refined Stable Diffusion for Privacy Compliant Synthetic Data2
R Rendering-Refined Stable Diffusion is introduced, a pipeline that combines 3D-rendering with Stable Diffusion, enabling prompt-based control over human attributes while preserving posture, unlike standard diffusion models that fail to retain posture or GANs that lack realism and flexible attribute control.
- Towards Fundamentally Scalable Model Selection: Asymptotically Fast Update and Selection2
This work defines isolated model embedding, a family of model selection schemes supporting asymptotically fast update and selection, and presents Standardized Embedder, an empirical realization of isolated model embedding.
- 2
- 2
- 2
- 2
- 2
- PrivateRec: Differentially Private Training and Serving for Federated News Recommendation2
A unified federated news recommendation method for effective and privacy-preserving model training and online serving with differential privacy guarantees and a federated recommendation model training method with differential privacy, which can avoid the dimension-dependent noise for large models via label permutation and differentially private attention modules.
- 2
- 2
- Auditing unauthorized training data from AI generated content using information isotopes1
This work proposes an information-isotope tracing framework that selectively marks target data elements and detects their propagation in model outputs, providing concrete evidence of data utilization under black-box access.
- UniCompress: Token Compression for Unified Vision-Language Understanding and Generation1
This work proposes a unified token compression algorithm UniCompress that significantly reduces visual token count while preserving performance on both image understanding and generation tasks, and introduces a plug-in compression and decompression mechanism guided with learnable global meta tokens.
- Diagnosing Knowledge Conflict in Multimodal Long-Chain Reasoning1
This work formalizes these failures under a unified notion of knowledge conflict, distinguishing input-level objective conflict from process-level effective conflict and provides a mechanism-level view of multimodal reasoning under knowledge conflict and enable principled diagnosis and control of long-CoT failures.
- 1
- Seeing Further on the Shoulders of Giants: Knowledge Inheritance for Vision Foundation Models1
This paper unifies multiple pre-trained teacher models in a shared latent space to mitigate the ``imbalanced transfer''issue caused by their distributional gaps and introduces a knowledge preservation strategy to take a general-purpose teacher as a knowledge base for integrating knowledge from the remaining purpose-specific teachers using an adapter module.
- Sybil-based Virtual Data Poisoning Attacks in Federated Learning1
A sybil-based virtual data poisoning attack is proposed, where a malicious client generates sybil nodes to amplify the poisoning model’s impact, and a virtual data generation method based on gradient matching is developed to reduce neural network computational complexity.
- 1
- 1
- Byzantine-Robust Learning on Heterogeneous Data via Gradient Splitting1
GAIN, a gradient decomposition scheme that can help adapt existing robust algorithms to heterogeneous datasets, is proposed, a gradient decomposition scheme that can help adapt existing robust algorithms to heterogeneous datasets.
- 1
- 1
- –
- –
- –
- –
- Argus-Unified: Towards A Compact and Economical Unified Model for Image Understanding and Generation–
- 2nd Workshop on Frontiers in Graph Machine Learning for the Large Model Era–
This workshop explores how graph learning can support emerging challenges in knowledge reasoning, temporal and multi-hop inference, and AI systems and investigates how advances in representation learning, structure-aware generalization, and efficient graph processing can contribute to trustworthy and scalable AI systems.
- –
- –
- Empirical Recipes for Efficient and Compact Vision-Language Models–
An empirical end-to-end efficiency analysis and systematically profile inference to identify the dominant bottlenecks are conducted and optimization recipes tailored to compact VLMs are developed that substantially reduce latency while preserving accuracy.
- Flower Hub: A Reproducible Benchmarking Platform for Federated Learning in Simulation and Deployment–
- FEDEXCHANGE: Bridging the Domain Gap in Federated Object Detection for Free–
FEDEXCHANGE is proposed, a novel FOD framework that bridges domain gaps without introducing additional local computational overhead that enhances FOD performance, achieving 1.6X better mean average precision in challenging domains, such as rainy conditions, while requiring only 0.8X the computational resources compared to baseline methods.
- –
- –
- DECO-Bench: Unified Benchmark for Decoupled Task-Agnostic Synthetic Data Release–
This work tackles the question of how to systematically benchmark task-agnostic decoupling methods for privacy-preserving machine learning (ML), and integrates various decoupling techniques along with synthetic data generation and evaluation protocols within a unified system.
- –
- –
- –
- –
- Activity Recognition on Avatar-Anonymized Datasets with Masked Differential Privacy–
This work proposes an effective approach called masked differential privacy (MaskDP), which allows for controlling sensitive regions where differential privacy is applied, in contrast to applying DP on the entire input.
- –
- The 1st International Workshop on Federated Learning with Graph Data (FedGraph)–
The nascent field of federated learning (FL), which aims to enable individual clients to jointly train their models while keeping their local data decentralized and completely private, is a promising paradigm for large-scale distributed and private training of GNNs.
- –
- –
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 2026-10-12. 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.