Riashat Islam

About

Riashat Islam is a researcher specializing in deep reinforcement learning (RL), a field that sits at the intersection of machine learning and sequential decision-making. He is perhaps best known for co-authoring "An Introduction to Deep Reinforcement Learning" (2018), a foundational survey that has become one of the most widely read references in the field, accumulating over 1,700 citations across its versions. This work synthesized the rapidly evolving landscape of deep RL, making the field accessible to researchers and practitioners alike, and has served as a cornerstone resource for students entering the discipline worldwide. Beyond survey contributions, Islam has pursued more specialized research in goal-conditioned reinforcement learning, exploring how agents can effectively represent and pursue diverse objectives. His 2022 paper on discrete factorial representations as abstractions for goal-conditioned RL demonstrates a continued focus on improving the scalability and generalization of RL agents in complex, multi-task settings. Together, his body of work reflects a dual commitment to both broadening access to deep RL concepts and advancing the technical frontiers of the field, making him a meaningful contributor to the modern reinforcement learning research community.

Research Focus

Key Achievements

Top Papers

1. 1
   An Introduction to Deep Reinforcement Learning

   1,246 citations · 2018
2. 2
   An Introduction to Deep Reinforcement Learning

   539 citations · 2018
3. 3
   Discrete Factorial Representations as an Abstraction for Goal Conditioned Reinforcement Learning

   4 citations · 2022

Key Collaborators

Contact & Links