Deep Neural Network Performance

Deep neural network (DNN) performance research focuses on optimizing DNN efficiency and robustness across diverse applications and hardware platforms. Current efforts concentrate on improving inference speed and energy efficiency on resource-constrained devices like mobile phones and embedded systems, often employing techniques like dynamic neural networks and specialized hardware accelerators, as well as mitigating vulnerabilities to privacy attacks and addressing the impact of hardware imperfections. These advancements are crucial for deploying DNNs in safety-critical applications and expanding their use in areas like healthcare and autonomous systems, where reliable and efficient performance is paramount.

Papers

September 2, 2024

CARIn: Constraint-Aware and Responsive Inference on Heterogeneous Devices for Single- and Multi-DNN Workloads
Ioannis Panopoulos, Stylianos I. Venieris, Iakovos S. Venieris
Convolutional Neural Network Deep Neural Network Deep Learning Application Logical Constraint Heterogeneous Device DNN Workload Deep Neural Network Performance

February 2, 2024

Double-Dip: Thwarting Label-Only Membership Inference Attacks with Transfer Learning and Randomization
Arezoo Rajabi, Reeya Pimple, Aiswarya Janardhanan, Surudhi Asokraj, Bhaskar Ramasubramanian, Radha Poovendran
Transfer Learning Supervised ImageNet Black Box Membership Inference Attack Randomization Program Deep Neural Network Performance Label Only Membership Inference Attack

January 17, 2024

Dynamic DNNs and Runtime Management for Efficient Inference on Mobile/Embedded Devices
Lei Xun, Jonathon Hare, Geoff V. Merrett
Neural Network Deep Neural Network Efficient Inference Embedded System Mobile User Runtime Monitoring Dynamic Neural Network Deep Neural Network Performance

December 11, 2023

Compute-in-Memory based Neural Network Accelerators for Safety-Critical Systems: Worst-Case Scenarios and Protections
Zheyu Yan, Xiaobo Sharon Hu, Yiyu Shi
Worst Case DNN Accelerator Safety Critical System Neural Network Accelerator Compute in Memory Deep Neural Network Performance Worst Case Performance

August 2, 2023

Using ScrutinAI for Visual Inspection of DNN Performance in a Medical Use Case
Rebekka Görge, Elena Haedecke, Michael Mock
Model Performance Root Cause Visual Inspection Visual Analytics Clinical Application Deep Neural Network Performance

July 13, 2023

A Scenario-Based Functional Testing Approach to Improving DNN Performance
Hong Zhu, Thi Minh Tam Tran, Aduen Benjumea, Andrew Bradley
Machine Learning Deep Neural Network Machine Learning Model Transfer Learning Scenario Based Testing Deep Neural Network Performance

February 23, 2023

Practical Knowledge Distillation: Using DNNs to Beat DNNs
Chung-Wei Lee, Pavlos Athanasios Apostolopulos, Igor L. Markov
Knowledge Distillation DNN Framework Gradient Boosting Model Distillation Deep Neural Network Performance Deep Neural Network Ensemble

November 18, 2022

SAMSON: Sharpness-Aware Minimization Scaled by Outlier Normalization for Improving DNN Generalization and Robustness
Gonçalo Mordido, Sébastien Henwood, Sarath Chandar, François Leduc-Primeau
Native Robustness Sharpness Aware Minimization Neural Network Generalization Deep Neural Network Performance Sharpness Aware Training

August 29, 2022

Survey: Exploiting Data Redundancy for Optimization of Deep Learning
Jou-An Chen, Wei Niu, Bin Ren, Yanzhi Wang, Xipeng Shen
Deep Learning Deep Neural Network Timely Survey Optimization Purpose DNN Framework Information Redundancy Deep Neural Network Performance

July 15, 2022

Computing-In-Memory Neural Network Accelerators for Safety-Critical Systems: Can Small Device Variations Be Disastrous?
Zheyu Yan, Xiaobo Sharon Hu, Yiyu Shi
DNN Accelerator Safety Critical System Non Volatile Memory Deep Neural Network Performance Device Variation

February 3, 2022

Mapping DNN Embedding Manifolds for Network Generalization Prediction
Molly O'Brien, Julia Bukowski, Mathias Unberath, Aria Pezeshk, Greg Hager
Deep Neural Network DNN Framework Class Manifold Neural Network Generalization Deep Neural Network Performance

Deep Neural Network Performance

Papers

CARIn: Constraint-Aware and Responsive Inference on Heterogeneous Devices for Single- and Multi-DNN Workloads

Double-Dip: Thwarting Label-Only Membership Inference Attacks with Transfer Learning and Randomization

Dynamic DNNs and Runtime Management for Efficient Inference on Mobile/Embedded Devices

Compute-in-Memory based Neural Network Accelerators for Safety-Critical Systems: Worst-Case Scenarios and Protections

Using ScrutinAI for Visual Inspection of DNN Performance in a Medical Use Case

A Scenario-Based Functional Testing Approach to Improving DNN Performance

Practical Knowledge Distillation: Using DNNs to Beat DNNs

SAMSON: Sharpness-Aware Minimization Scaled by Outlier Normalization for Improving DNN Generalization and Robustness

Survey: Exploiting Data Redundancy for Optimization of Deep Learning

Computing-In-Memory Neural Network Accelerators for Safety-Critical Systems: Can Small Device Variations Be Disastrous?

Mapping DNN Embedding Manifolds for Network Generalization Prediction