Andrew (Andy) EhrenbergI am a Masters student in Machine Learning at UCL with an interest in improving generalization and exploration efficiency of embodied agents. In particular, I'm keen on (Meta) Reinforcement Learning; Offline RL; and embodied agent based models of language learning that enable language grounding, and the formation of reliable beliefs about tasks and environments that allow for effective exploration. I envision that the use of language and knowledge/memory will improve the zero-shot generalization capacities of agents, which will allow for deployment in real-world scenarios. In 2020, I received my Bachelors in Statistics (with a Math Minor) at UCLA, where I took courses on Deep Learning, Machine Learning, Probability Theory, Monte Carlo Methods, and Programming. I have gained experience in Computer Vision research, in which I developed software for eye-gaze tracking, researched generative models for image colorization, worked with a team on models for DeepFake detection, and experimented with implementations of Imitation Learning. Aside from my academic interests, I competed for the NCAA Division I Track and Field and Cross Country teams at UCLA, enjoy writing/playing music, and spend lots of time climbing, backpacking, and cycling. |
|
AI ResearchI'm interested advancing RL informed by natural language, Multi-Agent Reinforcement Learning, (Unsupervised) Meta RL and Offline RL to improve exploration, generalization, zero-shot coordination, emergent communication, and long-horizon robotic planning. |
|
|