Projects

This list includes research areas I am interested in and potential areas for student projects. The projects are divided into two categories: available include research areas we are currently working and/or where I have several projects ready-to-go available for students; coming soon include areas we have explored in the past or are interested in but where there are not ready-to-go projects (these projects will be highly exploratory in their initial stages.!). Each project page includes a brief summary of the goals, domains being explored, and reference papers related to the projects in that area.

available

Interpretable Deep Neural Networks

This research direction investigates designs of self-interpretable deep neural networks capable of providing explanations alongside their predictions. This involves either generalizing existing designs to novel domains and tasks or creating new ones from scratch.

Explaining Neurons

This research area aims to understand what deep neural networks learn during the training process. Specifically, we are interested on analyzing the behavior of individual neurons and groups of neurons and discovering the concepts they learn to recognize and the relationships between these concepts.

Evaluating Explanations

The goal of this research area is to understand the motivations behind specific predictions of a deep neural network. understanding what deep neural networks learn during the training process. Recently this area has been moved under the umbrella name of "Mechanistic Interpretability"

Visual Analytics for Explainable AI

The goal of this research area is to develop the backbone AI systems of Visual Analytics systems that facilitate the interpretation of deep neural networks by integrating explanation methods.

coming soon

Explaining Predictions

The goal of this research area is to understand the motivations behind specific predictions of a deep neural network (e.g., biased decisions).

Explaining Reinforcement Learning

The goal of this research area is to explore the application of explanation methods in the area of deep reinforcement learning (e.g., games).