I'm interested in large scale machine learning that lies at the intersection of both system and theory. So far lots of "big data" researches have only focused on one or the other. For example, MapReduce is a nice system framework but generally not suitable for many ML tasks, and theoretical works often makes poor assumptions from the system design perspective. I'm hoping to reduce the gap of the two communities in my graduate research. My main project these days is Petuum.
Let me begin by clarifying my name. I go by "Dai Wei," even though my official name is Wei Dai, where Wei is my first name. I grew up in Taiwan and came to the States for undergrad at Wesleyan Univ. under generous financial support from Freeman Asian Scholarship. I later transferred to Caltech through the 3-2 program to pursue computer science and subsequently came to Carnegie Mellon Univ. for graduate studies.
Besides research, I'm part of a wonderful church called Oakland International Fellowship, where I help out with small group, tech team, bulletin, and worship. I also like to jog, play basketball, and getting crushed in squash by my friends.
My paper on Eager Stale Synchronous Parallel (ESSP) communication model for parameter server is accepted to AAAI (2015).
The alpha release of a project I'm working on, Petuum, is officially on GitHub!