Christopher Wang

I'm an M.Eng student advised by Boris Katz at the InfoLab . I also work with Andrei Barbu. I am interested in NLP, Machine Learning, and Linguistics. My current projects focus on grounded language acquisition and semantic parsing. In my free time, I enjoy reading and drawing.

Learning a natural-language to LTL executable semantic parser for grounded robotics [ video ] [ pdf ]

Christopher Wang (MIT) · Candace Ross (MIT) · Yen-Ling Kuo (MIT) · Boris Katz (MIT) · Andrei Barbu (MIT) in in CoRL 2020

ObjectNet: A large-scale bias-controlled dataset for pushing the limits of object recognition models [ pdf ]

Andrei Barbu (MIT) · David Mayo (MIT) · Julian Alverio (MIT) · William Luo (MIT) · Christopher Wang (MIT) · Dan Gutfreund (IBM Research) · Josh Tenenbaum (MIT) · Boris Katz (MIT) in NeurIPS 2019

6.s897 ML for Healthcare Final Project

Worked with Sooraj Boominathan and Kevin Weng for 6.s897 Final Project

For our final project, we worked on predicting patient re-hospitalization for psychiatric patients. We learned a low-dimensional representation for each patient and found that this improved prediction. Writeup and poster here.

6.837 Computer Graphics Final Project

Worked with Emily Weng for 6.837 Final Project

We implemented Craig Reynold's algorithm for simulated flocking in C++. Writeup and video here.

6.854 Grad Algorithms Final Project

Worked with Oliver Ren for 6.854 Final Project

We synthesize the results from On the Quickest Flow Problem in Dynamic Networks – A Parametric Min-Cost Flow Approach (SIAM, 2015) and Cancel-and-tighten algorithm for quickest flow problems (Networks Volume 69, 2017). The goal of our paper is to summarize and clearly explain both Lin and Jaillet's cost-scaling algorithm as well as Saho and Shigeno's improvements.

Grafitti

Collaborated with Kevin Weng for 6.178 IAP web competition 2016

Used Meteor, MongoDB, and D3.js to build "Graffiti," a web-app which allows users to post disappearing messages to an unstructured message board. Users can vote on posts to keep them from fading.
Won 2nd place in 6.148 Rookie Division!

DBShare

Collaborated with Sooraj Boominathan, Nathan Yu, and Nan Hu for Dartmouth Hackday 2016

Used Node.js and MongoDB to build "DBShare," a web-app which allows Dartmouth students to share their spare dining passes with other students.

MercuryRice

Institute for Data, Systems, and Society & Joint Program on the Science and Policy of Global Change 2016

Built static HTML pages to display mercury levels in China using Python Bokeh. The content of the website is still in progress. The site, as it is now, can be viewed here.

Clippd

Collaborated with Sooraj Boominathan and Kevin Weng for HACK MIT 2016

Built a web app which can take a YouTube url and produce a 'sound board' of words which are spoken in the video. Users can use buttons from this sound board to form strings. Our app can then use this string to cut-and-paste an edited version of the video which only includes the desired words.