Picture for Ehsan Abbasnejad

Ehsan Abbasnejad

ETAGE: Enhanced Test Time Adaptation with Integrated Entropy and Gradient Norms for Robust Model Performance

Add code
Sep 14, 2024
Viaarxiv icon

Rethinking State Disentanglement in Causal Reinforcement Learning

Add code
Aug 24, 2024
Viaarxiv icon

Bayesian Low-Rank LeArning (Bella): A Practical Approach to Bayesian Neural Networks

Add code
Jul 30, 2024
Viaarxiv icon

Knowledge Composition using Task Vectors with Learned Anisotropic Scaling

Add code
Jul 03, 2024
Viaarxiv icon

Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling

Add code
May 27, 2024
Figure 1 for Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling
Figure 2 for Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling
Figure 3 for Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling
Figure 4 for Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling
Viaarxiv icon

BruSLeAttack: A Query-Efficient Score-Based Black-Box Sparse Adversarial Attack

Add code
Apr 08, 2024
Viaarxiv icon

Premonition: Using Generative Models to Preempt Future Data Changes in Continual Learning

Add code
Mar 12, 2024
Figure 1 for Premonition: Using Generative Models to Preempt Future Data Changes in Continual Learning
Figure 2 for Premonition: Using Generative Models to Preempt Future Data Changes in Continual Learning
Figure 3 for Premonition: Using Generative Models to Preempt Future Data Changes in Continual Learning
Figure 4 for Premonition: Using Generative Models to Preempt Future Data Changes in Continual Learning
Viaarxiv icon

Do Deep Neural Network Solutions Form a Star Domain?

Add code
Mar 12, 2024
Viaarxiv icon

Neural Redshift: Random Networks are not Random Functions

Add code
Mar 05, 2024
Viaarxiv icon

Unveiling Backbone Effects in CLIP: Exploring Representational Synergies and Variances

Add code
Dec 22, 2023
Viaarxiv icon