Picture for Tom Melham

Tom Melham

Exposing Previously Undetectable Faults in Deep Neural Networks

Add code
Jun 01, 2021
Figure 1 for Exposing Previously Undetectable Faults in Deep Neural Networks
Figure 2 for Exposing Previously Undetectable Faults in Deep Neural Networks
Figure 3 for Exposing Previously Undetectable Faults in Deep Neural Networks
Figure 4 for Exposing Previously Undetectable Faults in Deep Neural Networks
Viaarxiv icon

Semantic Adversarial Perturbations using Learnt Representations

Add code
Jan 29, 2020
Figure 1 for Semantic Adversarial Perturbations using Learnt Representations
Figure 2 for Semantic Adversarial Perturbations using Learnt Representations
Figure 3 for Semantic Adversarial Perturbations using Learnt Representations
Figure 4 for Semantic Adversarial Perturbations using Learnt Representations
Viaarxiv icon

DeepSynth: Program Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning

Add code
Nov 22, 2019
Figure 1 for DeepSynth: Program Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning
Figure 2 for DeepSynth: Program Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning
Figure 3 for DeepSynth: Program Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning
Figure 4 for DeepSynth: Program Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning
Viaarxiv icon

Generating Realistic Unrestricted Adversarial Inputs using Dual-Objective GAN Training

Add code
May 07, 2019
Figure 1 for Generating Realistic Unrestricted Adversarial Inputs using Dual-Objective GAN Training
Figure 2 for Generating Realistic Unrestricted Adversarial Inputs using Dual-Objective GAN Training
Figure 3 for Generating Realistic Unrestricted Adversarial Inputs using Dual-Objective GAN Training
Figure 4 for Generating Realistic Unrestricted Adversarial Inputs using Dual-Objective GAN Training
Viaarxiv icon