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.
hub
Imagenet: A large-scale hierarchical image database
17 Pith papers cite this work. Polarity classification is still indexing.
17
Pith papers citing it
hub tools
citation-role summary
dataset 1
citation-polarity summary
roles
dataset 1polarities
use dataset 1representative citing papers
Floating-point weight formats in embedded neural networks suffer near-total accuracy loss from a single electromagnetic fault injection, while 8-bit integer formats retain substantially higher accuracy on the same hardware.
OD-TTA enables resource-efficient test-time adaptation on edge devices by triggering updates only on detected domain shifts, achieving comparable accuracy with lower energy and computation costs for embodied visual systems.
AIR amortizes 2D Gaussian splatting into a self-supervised feed-forward network via residual stages, explicit stage control, and Predict-Optimize-Distill training.
ELSA is a near-SRAM dataflow architecture realizing elastic inference in SNNs via fine-grained spine/token pipelines, bundled AER, and mini-batch Gustavson products, delivering up to 3.4x speedup and 22.1x energy gains over SOTA accelerators on ResNet-50.
This work provides the first systematic study of transferring direct-coded spiking neural networks to event-based representations while aiming to preserve accuracy and reduce energy use.
A new chain of lightweight neural predictors with information inheritance achieves near state-of-the-art lossless compression ratios while delivering 1.2-6.3x faster encoding and 2.8-12.3x faster decoding than PAC on GPUs.
LOCALUT delivers 1.82x geometric mean speedup for quantized DNN inference on real UPMEM DRAM-PIM devices by using operation-packed LUTs with canonicalization, reordering, and slice streaming.
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.
Fed-TaLoRA uses task-agnostic low-rank residual adaptation with post-aggregation calibration to enable efficient federated continual fine-tuning across sequential tasks under non-IID conditions.
AAND is a two-stage anomaly detection method that advances a pre-trained teacher via residual anomaly amplification and applies hard knowledge distillation in reverse distillation to achieve SOTA results on MVTecAD, VisA, and MVTec3D-RGB.
Replacement Learning replaces selected blocks in CNNs and ViTs with learnable parameter-fusion surrogates derived from adjacent layers to reduce full-depth backpropagation redundancy.
AnyUser translates free-form sketches on images plus optional language into executable robot actions for domestic tasks using multimodal fusion and a hierarchical policy.
GAPL anchors text prompts to second-order Gram matrix statistics to improve vision-language model adaptation across domains.
TEA is a new targeted adversarial attack that incorporates edge information from the target image to reduce query count and improve performance in low-query black-box hard-label settings.
Multi-horizon forecasting with deep learning on sky images and PV data improves prediction accuracy across multiple future time steps and architectures by jointly optimizing sequences of outputs.
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.
citing papers explorer
-
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.
-
The Weight of a Bit: EMFI Sensitivity Analysis of Embedded Deep Learning Models
Floating-point weight formats in embedded neural networks suffer near-total accuracy loss from a single electromagnetic fault injection, while 8-bit integer formats retain substantially higher accuracy on the same hardware.
-
EmbodiTTA: Resource-Efficient Test-Time Adaptation for Embodied Visual Systems
OD-TTA enables resource-efficient test-time adaptation on edge devices by triggering updates only on detected domain shifts, achieving comparable accuracy with lower energy and computation costs for embodied visual systems.
-
AIR: Amortized Image Reconstruction Framework for Self-Supervised Feed-Forward 2D Gaussian Splatting
AIR amortizes 2D Gaussian splatting into a self-supervised feed-forward network via residual stages, explicit stage control, and Predict-Optimize-Distill training.
-
ELSA: An ELastic SNN Inference Architecture for Efficient Neuromorphic Computing
ELSA is a near-SRAM dataflow architecture realizing elastic inference in SNNs via fine-grained spine/token pipelines, bundled AER, and mini-batch Gustavson products, delivering up to 3.4x speedup and 22.1x energy gains over SOTA accelerators on ResNet-50.
-
Direct-to-Event Spiking Neural Network Transfer
This work provides the first systematic study of transferring direct-coded spiking neural networks to event-based representations while aiming to preserve accuracy and reduce energy use.
-
Lossless Compression via Chained Lightweight Neural Predictors with Information Inheritance
A new chain of lightweight neural predictors with information inheritance achieves near state-of-the-art lossless compression ratios while delivering 1.2-6.3x faster encoding and 2.8-12.3x faster decoding than PAC on GPUs.
-
LOCALUT: Harnessing Capacity-Computation Tradeoffs for LUT-Based Inference in DRAM-PIM
LOCALUT delivers 1.82x geometric mean speedup for quantized DNN inference on real UPMEM DRAM-PIM devices by using operation-packed LUTs with canonicalization, reordering, and slice streaming.
-
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.
-
Task-agnostic Low-rank Residual Adaptation for Efficient Federated Continual Fine-Tuning
Fed-TaLoRA uses task-agnostic low-rank residual adaptation with post-aggregation calibration to enable efficient federated continual fine-tuning across sequential tasks under non-IID conditions.
-
Advancing Pre-trained Teacher: Towards Robust Feature Discrepancy for Anomaly Detection
AAND is a two-stage anomaly detection method that advances a pre-trained teacher via residual anomaly amplification and applies hard knowledge distillation in reverse distillation to achieve SOTA results on MVTecAD, VisA, and MVTec3D-RGB.
-
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.
-
AnyUser: Translating Sketched User Intent into Domestic Robots
AnyUser translates free-form sketches on images plus optional language into executable robot actions for domestic tasks using multimodal fusion and a hierarchical policy.
-
Gram-Anchored Prompt Learning for Vision-Language Models via Second-Order Statistics
GAPL anchors text prompts to second-order Gram matrix statistics to improve vision-language model adaptation across domains.
-
Accelerating Targeted Hard-Label Adversarial Attacks in Low-Query Black-Box Settings
TEA is a new targeted adversarial attack that incorporates edge information from the target image to reduce query count and improve performance in low-query black-box hard-label settings.
-
Learning Long-Term Temporal Dependencies in Photovoltaic Power Output Prediction Through Multi-Horizon Forecasting
Multi-horizon forecasting with deep learning on sky images and PV data improves prediction accuracy across multiple future time steps and architectures by jointly optimizing sequences of outputs.
-
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.