PCDM uses a poisoning-oriented conditional diffusion model with an adjustable vector and jumping strategy to create stealthier and more effective poisoned data than GAN-based attacks against federated learning.
hub
Learning multiple layers of features from tiny images
32 Pith papers cite this work. Polarity classification is still indexing.
32
Pith papers citing it
hub tools
citation-role summary
dataset 2
background 1
citation-polarity summary
representative citing papers
QLAM extends state-space models with quantum superposition in the hidden state for linear-time long-sequence modeling and reports consistent gains over RNN and transformer baselines on sequential image tasks.
FedSAF shifts prototype alignment in heterogeneous federated learning from coordinate matching to inter-class structural relations and reports up to 3.52% gains over prior methods.
Backdoor poisoning triggers in contrastive learning can be repurposed as statistical watermarks for dataset IP protection via a multi-level scheme and density-based verification.
Unlearnable examples fail under pretraining-finetuning due to semantic filtering by frozen layers, but Shallow Semantic Camouflage restores effectiveness by confining perturbations to semantically valid subspaces.
AdaLoc keeps a model locked to authorized users by confining all post-deployment updates to a chosen subset of weights, preserving both task performance for authorized use and near-random accuracy for unauthorized use across vision and language models.
FedCoE proposes a coordinated dual-level MoE framework for federated learning that improves global and personalized accuracy while enabling strong cold-start performance for new clients.
WePE encodes 2D patch positions in Vision Transformers via Weierstrass elliptic functions on the complex plane to exploit double periodicity and derive relative positions algebraically.
DFBScanner detects backdoors by combining anomaly indicators from final-layer parameters into a Trojan clue score, reporting 97.17% true-positive rate, 0.95% false-positive rate, and 1 ms average detection time on a benchmark of over 5,000 models.
SC-DN establishes a global first-order stationary point per round and solves a mixed-integer signomial program to optimize four control variables for VFL, yielding better classification performance and lower resource use than greedy baselines on image and multi-modal data.
EDL learns a transferable classification loss from unlimited synthetic data via evolutionary optimization and a ranking-consistency objective, serving as a competitive drop-in replacement for cross-entropy on CIFAR-10 with ResNet models.
MANOJAVAM unifies matrix multiplication and SVD for PCA on FPGA with block-streaming systolic arrays and pipelined Jacobi-CORDIC, delivering up to 22.75x SVD speedup and 42.14x lower energy than an NVIDIA A6000 GPU.
The FastAT Benchmark standardizes evaluation of over twenty fast adversarial training methods under unified conditions, showing that well-designed single-step approaches can match or exceed PGD-AT robustness at lower training cost on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
CroSatFL cuts ground station communications by over 100x and transmission energy by 6x in satellite federated learning compared to baselines, while keeping competitive accuracy.
BID-LoRA uses bi-directional low-rank adapters with retain/new/unlearn pathways and escape unlearning to enable continual learning and unlearning while minimizing knowledge leakage and parameter updates.
HIL-CBM is a hierarchical label-free concept bottleneck model that improves classification accuracy and explanation quality over prior single-level CBMs using a visual consistency loss and dual heads.
FedACT schedules devices across concurrent FL jobs via alignment scoring and fairness to reduce average job completion time by up to 8.3x and raise accuracy by up to 44.5% versus baselines.
BicKD introduces a bilateral contrastive loss in knowledge distillation that strengthens class-wise orthogonality and intra-class consistency in predictive distributions, outperforming prior logit-based methods.
Packed Shamir secret sharing yields up to 11x lower communication and 2.6x faster online runtime for secure deep learning inference versus prior Shamir-based methods.
FLARE uses adaptive multi-dimensional reputation scores and soft exclusion to improve Byzantine robustness in federated learning by up to 16% over prior methods while handling a new Statistical Mimicry attack.
Hierarchical layer-grouped prompt tuning coordinates prompts across layers via shared group prompts and a single root prompt per task to reduce catastrophic forgetting in continual learning.
IAdaPID-ADG integrates non-increasing effective learning rates from AMSGrad and gradient-difference modulation from DiffGrad into AdaPID, yielding better convergence and stability than prior optimizers on MNIST, CIFAR10, IARC, and AnnoCerv.
Replacement Learning replaces selected blocks in CNNs and ViTs with learnable parameter-fusion surrogates derived from adjacent layers to reduce full-depth backpropagation redundancy.
iGSP uses implicit gradient subspace projection in two phases to enable efficient continual adaptation of vision-language models, claiming SOTA accuracy with 42.7% fewer trainable parameters and 86.9% less total parameter growth.
citing papers explorer
-
PCDM: A Diffusion-Based Data Poisoning Attack Against Federated Learning Systems
PCDM uses a poisoning-oriented conditional diffusion model with an adjustable vector and jumping strategy to create stealthier and more effective poisoned data than GAN-based attacks against federated learning.
-
QLAM: A Quantum Long-Attention Memory Approach to Long-Sequence Token Modeling
QLAM extends state-space models with quantum superposition in the hidden state for linear-time long-sequence modeling and reports consistent gains over RNN and transformer baselines on sequential image tasks.
-
From Coordinate Matching to Structural Alignment: Rethinking Prototype Alignment in Heterogeneous Federated Learning
FedSAF shifts prototype alignment in heterogeneous federated learning from coordinate matching to inter-class structural relations and reports up to 3.52% gains over prior methods.
-
Repurposing and Evaluating the (In)Feasibility of Dataset Poisoning enabled Watermarking for Contrastive Learning
Backdoor poisoning triggers in contrastive learning can be repurposed as statistical watermarks for dataset IP protection via a multi-level scheme and density-based verification.
-
Channel-Level Semantic Perturbations: Unlearnable Examples for Diverse Training Paradigms
Unlearnable examples fail under pretraining-finetuning due to semantic filtering by frozen layers, but Shallow Semantic Camouflage restores effectiveness by confining perturbations to semantically valid subspaces.
-
Re-Key-Free, Risky-Free: Adaptable Model Usage Control
AdaLoc keeps a model locked to authorized users by confining all post-deployment updates to a chosen subset of weights, preserving both task performance for authorized use and near-random accuracy for unauthorized use across vision and language models.
-
FedCoE: Bridging Generalization and Personalization via Federated Coordinated Dual-level MoEs
FedCoE proposes a coordinated dual-level MoE framework for federated learning that improves global and personalized accuracy while enabling strong cold-start performance for new clients.
-
Weierstrass Positional Encoding for Vision Transformers
WePE encodes 2D patch positions in Vision Transformers via Weierstrass elliptic functions on the complex plane to exploit double periodicity and derive relative positions algebraically.
-
Lightweight and Fast Backdoor Model Detection
DFBScanner detects backdoors by combining anomaly indicators from final-layer parameters into a Trojan clue score, reporting 97.17% true-positive rate, 0.95% false-positive rate, and 1 ms average detection time on a benchmark of over 5,000 models.
-
Optimizing Server Placement for Vertical Federated Learning in Dynamic Edge/Fog Networks
SC-DN establishes a global first-order stationary point per round and solves a mixed-integer signomial program to optimize four control variables for VFL, yielding better classification performance and lower resource use than greedy baselines on image and multi-modal data.
-
Distribution-Free Pretraining of Classification Losses via Evolutionary Dynamics
EDL learns a transferable classification loss from unlimited synthetic data via evolutionary optimization and a ranking-consistency objective, serving as a competitive drop-in replacement for cross-entropy on CIFAR-10 with ResNet models.
-
MANOJAVAM: A Scalable, Unified FPGA Accelerator for Matrix Multiplication and Singular Value Decomposition in Principal Component Analysis
MANOJAVAM unifies matrix multiplication and SVD for PCA on FPGA with block-streaming systolic arrays and pipelined Jacobi-CORDIC, delivering up to 22.75x SVD speedup and 42.14x lower energy than an NVIDIA A6000 GPU.
-
FastAT Benchmark: A Comprehensive Framework for Fair Evaluation of Fast Adversarial Training Methods
The FastAT Benchmark standardizes evaluation of over twenty fast adversarial training methods under unified conditions, showing that well-designed single-step approaches can match or exceed PGD-AT robustness at lower training cost on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
-
CroSatFL: Energy-Efficient Federated Learning with Cross-Aggregation for Satellite Edge Computing
CroSatFL cuts ground station communications by over 100x and transmission energy by 6x in satellite federated learning compared to baselines, while keeping competitive accuracy.
-
BID-LoRA: A Parameter-Efficient Framework for Continual Learning and Unlearning
BID-LoRA uses bi-directional low-rank adapters with retain/new/unlearn pathways and escape unlearning to enable continual learning and unlearning while minimizing knowledge leakage and parameter updates.
-
Hierarchical, Interpretable, Label-Free Concept Bottleneck Model
HIL-CBM is a hierarchical label-free concept bottleneck model that improves classification accuracy and explanation quality over prior single-level CBMs using a visual consistency loss and dual heads.
-
FedACT: Concurrent Federated Intelligence across Heterogeneous Data Sources
FedACT schedules devices across concurrent FL jobs via alignment scoring and fairness to reduce average job completion time by up to 8.3x and raise accuracy by up to 44.5% versus baselines.
-
BicKD: Bilateral Contrastive Knowledge Distillation
BicKD introduces a bilateral contrastive loss in knowledge distillation that strengthens class-wise orthogonality and intra-class consistency in predictive distributions, outperforming prior logit-based methods.
-
High-Throughput and Scalable Secure Inference Protocols for Deep Learning with Packed Secret Sharing
Packed Shamir secret sharing yields up to 11x lower communication and 2.6x faster online runtime for secure deep learning inference versus prior Shamir-based methods.
-
FLARE: Adaptive Multi-Dimensional Reputation for Robust Client Reliability in Federated Learning
FLARE uses adaptive multi-dimensional reputation scores and soft exclusion to improve Byzantine robustness in federated learning by up to 16% over prior methods while handling a new Statistical Mimicry attack.
-
Teaching Prompts to Coordinate: Hierarchical Layer-Grouped Prompt Tuning for Continual Learning
Hierarchical layer-grouped prompt tuning coordinates prompts across layers via shared group prompts and a single root prompt per task to reduce catastrophic forgetting in continual learning.
-
An Improved Adaptive PID Optimizer with Enhanced Convergence and Stability for Deep Learning
IAdaPID-ADG integrates non-increasing effective learning rates from AMSGrad and gradient-difference modulation from DiffGrad into AdaPID, yielding better convergence and stability than prior optimizers on MNIST, CIFAR10, IARC, and AnnoCerv.
-
Replacement Learning: Training Neural Networks with Fewer Parameters
Replacement Learning replaces selected blocks in CNNs and ViTs with learnable parameter-fusion surrogates derived from adjacent layers to reduce full-depth backpropagation redundancy.
-
iGSP:Implicit Gradient Subspace Projection for Efficient Continual Learning of Vision-Language Models
iGSP uses implicit gradient subspace projection in two phases to enable efficient continual adaptation of vision-language models, claiming SOTA accuracy with 42.7% fewer trainable parameters and 86.9% less total parameter growth.
-
MoASE++: Mixture of Activation Sparsity Experts with Domain-Adaptive On-policy Distillation for Continual Test Time Adaptation
MoASE++ combines activation sparsity experts with domain-adaptive on-policy distillation to achieve state-of-the-art continual test-time adaptation on image classification and segmentation benchmarks.
-
Breaking Global Self-Attention Bottlenecks in Transformer-based Spiking Neural Networks with Local Structure-Aware Self-Attention
LSFormer uses local structure-aware spiking self-attention and spiking response pooling to cut global attention bottlenecks, delivering 4.3% and 8.6% accuracy gains on Tiny-ImageNet and N-CALTECH101 over prior transformer-based SNNs.
-
Sparsity Hurts: Simple Linear Adapter Can Boost Generalized Category Discovery
LAGCD inserts residual linear adapters into each ViT block plus a distribution alignment loss to improve generalized category discovery by increasing model flexibility while reducing bias between seen and novel classes.
-
Unveiling the Backdoor Mechanism Hidden Behind Catastrophic Overfitting in Fast Adversarial Training
Catastrophic overfitting in fast adversarial training is reinterpreted as a weak-trigger variant of unlearnable tasks, allowing backdoor-inspired recalibration and outlier suppression to restore robustness.
-
Cooperate to Compete: Strategic Data Generation and Incentivization Framework for Coopetitive Cross-Silo Federated Learning
CoCoGen+ models each federated learning round as a weighted potential game with strategic synthetic data generation and payoff redistribution incentives, showing improved efficiency over baselines under non-IID data and competition.
-
Neural Network Optimization Reimagined: Decoupled Techniques for Scratch and Fine-Tuning
DualOpt decouples optimization by using real-time layer-wise weight decay for scratch training and weight rollback for fine-tuning to improve convergence, generalization, and reduce knowledge forgetting.
- DARTIC: Decentralized Anonymous Reputation at Scale for Trustworthy Crowdsourcing
- EmbodiTTA: Resource-Efficient Test-Time Adaptation for Embodied Visual Systems