On this page, I delve into various research directions that I have explored through the papers I have authored. While there is some overlap among them, the categorized sections below provide a helpful overview of most (though not all) of my research interests. They also shed light on the connections between some of the papers I have written.

I strongly believe that addressing sequential decision making problems entails tackling three primary challenges: representation learning (or generalization), exploration, and credit-assignment. To overcome these hurdles, diverse techniques can be employed, including temporal abstractions, model-based reinforcement learning, and policy optimization. These approaches can be explored within various contexts, such as continual learning problems, real-world applications, and computer games.

When it comes to my research style, I primarily focus on empirical investigations, placing significant emphasis on proper empirical evaluation and best practices for conducting such evaluations. Additionally, I am particularly excited in unifying algorithms that seemingly lack a connection at first glance but can be shown to be deeply connected.

Continual learning

Temporal abstractions


Representation learning

Credit assignment

Real-world applications

Unifying algorithms

Empirical evaluation and best practices

Model-based reinforcement learning

Policy optimization

Computer games

© Copyright 2023 Marlos C. Machado. Inspired by al-folio theme.