To Prospective Students

Last update: 8/2023


I want to be a great advisor and believe that this can only happen if and only if:

What We Do

Ultimately, our research mission is to 1) discover new sciences so we understand things that we previously don't and 2) build new engineering systems to do things that we previously couldn't.

Concretely, we do research at the intersection of imaging (optics + image sensors), cognition/perception (psychophysics and computational modeling), and computer systems (computer architecture and programming model).

Below is a partial list of things that we are currently exploring:


What We Are Looking For

We welcome students with a diverse background even if your formative education does not involve extensive programming. Over the year we've worked with students with backgrounds in physical sciences, math, psychology, archaeology, and arts.

In general, your experience will be much more positive if you are interested/had experience in at least one of the following, in addition to having solid programming skills (C/C++, Python, deep learning frameworks):

I encourage you to go over the slides of our visual computing course.


Mentoring Philosophy

As your advisor, I will help you adopt the IDEAL research methodology: Identification, Definition, Exploration, Assessment, and Learning.

IDENTIFICATION. Identifying a research topic is the first step for you. I believe good research topics should have both real-world impact and scientific merits. One effective approach to identifying such research topics is to spare some time everyday reading technology websites and news articles (e.g., TechCrunch, The Next Web). The focus is to identify issues that the general public care about (and thus have broad impact) and to think about what scientific and technical breakthroughs are needed to address those issues (your potential research topic!).

DEFINITION. Clearly defining the problem scope, after identifying a research topic, is vital to any research projects. In my observations, the single most expensive mistake that graduate students tend to make is skipping the scope definition step and jumping to the solution exploration step too quickly. Often, it leads to months of time wasted on searching solutions to an “undefined” problem. I will help you develop problem definition skills: if the problem identified is too specific, generalize it and scope the problem in a broader context; if the problem identified is too big, break it down into multiple smaller problems where each has a clear scope.

EXPLORATION. Exploring creative solutions is what we all enjoy as scientists and engineers. I believe creativity can be trained. Many psychology studies have shown that creativity comes from the reuse of ideas, or in other words, to take ideas that have worked well in one context and apply it by analogy to another context. A creative solution is nothing more than the juxtaposition of a particular idea in a particular context that people have never though of before.

To freely “reuse” ideas, however, one must master a lot of ideas. That is, to be able to "think out of the box", you first need to have a big box! I will encourage you to extensively read papers, particularly those that are not directly related to your current research, to create an idea repository that your future projects can borrow from.

ASSESSMENT. Assessing the practicality of a solution is particularly important to architecture and systems research. I will encourage you to assess solutions not only under the constraints of today’s technology, but also to consider future technology scaling so as to create systems that have a long-term impact.

LEARNING. The research loop is never closed without learning from the success and failure of the past. I will help you deliver good research talks as it is an important way for learning. A good talk usually means positive and insightful feedbacks, which motivate better research.


Send an Email

If you've made this far, feel free to send me an email telling us why your interests align with ours and how you can contribute. Ideally, you will have read one of our recent papers and tell us what you think, e.g., insights you learn from the paper, things that you are not convinced, how you would have done things differently, etc.