Ben Eysenbach is an Assistant professor of computer science at Princeton University.
His lab is currently focusing on methods that learn from unlabeled data. In the same way that unsupervised learning has revolutionized other areas of artificial intelligence, his research studies how unsupervised reinforcement learning methods might not only allow us to solve decision-making problems with less data, but also open the door to applications in sciences and engineering.
His graduate research focused on machine learning algorithms for sequential decision making. Combining tools from statistics, probabilistic inference and deep learning, his work has resulted in reinforcement learning algorithms that not only achieve a high degree of performance, but also carry strong theoretical guarantees and are typically simpler than prior methods.
Before his PhD, Ben spent a year conducting robotics research at Google Brain, teaching simulated robots to do backflips and learning how to avoid breaking themselves. Ben studied mathematics as an undergraduate at the Massachusetts Institute of Technology where he was elected to Phi Beta Kappa. As an undergraduate researcher, he taught computers to understand images and videos using machine learning and computer vision. For this work Ben received the annual award for Outstanding Undergraduate Research Project in Artificial Intelligence from the MIT Computer Science and Artificial Intelligence Laboratory. Outside of class, Ben contributed to augmented reality devices, drones for measuring water quality, and a sensor for the rocket team. Ben plans to stay in academia after he completes his PhD, designing the next generation of safe machine learning algorithms and teaching the next generation of scientists.
Growing up outside San Francisco instilled in Ben a love of the outdoors. If he’s not at his desk, he’s probably running in the mountains.