Howard Chen
I recently received my PhD in CS from Princeton, where I was co-advised by Danqi Chen and Karthik Narasimhan.
My research goal is to build systems that can continually learn.
Specifically, I aim to build lifelong agents that can operate reliably over long interactive horizons and evolve through self-update.
I approch it through the lens of memory in both its in-context and parametric forms:
designing challenging evaluation to stress test long-horizon agentic capabilities (WebShop),
developing algorithms that let models self-organize and manage its long-term memory (MemWalker),
examining reliability under context accumulation (Tracking Belief Shift),
and understading how model parameters can be updated without drastic forgetting of old knowledge/capabilities (Continual Memorization, RL Mitigates Forgetting).
I am also broadly interested in topics like interpretability and safety from the memory perspective.
During my PhD, I have interned at Meta (FAIR) working with Asli Celikyilmaz and Jason Weston. Prior to Princeton, I was an ML researcher at ASAPP working with Tao Lei. I was also a research assistant at Cornell Tech working with Yoav Artzi.
I recieved my M.Eng. in CS from Cornell and B.S. in EE from
National Taiwan University.
