Irina Rish

Papers

Total Citations

222

H-Index

About

Irina Rish is a leading researcher at the intersection of artificial intelligence and neuroscience, with a primary focus on continual reinforcement learning and the scaling laws of deep neural networks. Her most influential work, "Towards Continual Reinforcement Learning: A Review and Perspectives," has accumulated over 200 citations across its versions, establishing her as a key voice in lifelong and non-stationary RL. In this comprehensive review, Rish argues that reinforcement learning provides a natural framework for studying continual learning, offering crucial perspectives on how AI systems can adapt to changing environments without catastrophic forgetting. She has also made significant contributions to understanding neural network behavior through her work on "Broken Neural Scaling Laws," which introduces a smoothly broken power law functional form that accurately models and extrapolates how evaluation metrics vary with computational resources. This work challenges traditional assumptions about neural scaling, providing researchers with more precise tools for predicting model performance. Rish's research bridges theoretical foundations with practical implications, making her work essential reading for anyone interested in building more adaptive, efficient, and scalable AI systems.