Hi! I am a second year Ph.D. student at the Computer Vision Lab, Virginia Tech. My advisor is Devi Parikh. I also work closely with Larry Zitnick from Facebook AI Research on various problems. I spend my time thinking about problems in Deep Learning/Machine Learning, Computer Vision (CV), and ultimately, Artifical Intelligence (AI).
I am interested in topics such as vision and language, unsupervised learning and common sense reasoning. A lot of my work focuses on leveraging abstract scenes made from clipart for high-level scene understanding, and language grounding.
I also care about issues of how we evaluate our models, as we edge towards higher-level AI-complete tasks. In my first project in grad school, I worked on a (now popularly used) evaluation metric for image captioning called CIDEr.
I am always interested in how we can elicit more human-like behaviour from machine learning models. One concrete direction that really excites me is building language models to solve a task (say convincing a human) as opposed to necessarily being correct or accurate. The latter could in many cases be a strictly harder thing to do!
- Two papers accepted to CVPR 2017 as Spotlight presentations!
- I will be back at Google Research in Winter 2017, working with Kevin Murphy on generative models for images!
- Serving as reviewer for ECCV 2016, BMVC 2017, CVPR 2017, ICCV 2017
- I interned at Google Research in Summer 2016, with Gal Chechik and Samy Bengio!
- I interned at the Center for Visual Computing at INRIA in Summer, 2014 with Iasonas Kokkinos
- MSCOCO Caption Evaluation code
- Codes from MSCOCO Caption Evaluation for metrics (BLEU, ROUGE, CIDEr-D and METEOR), independent of the COCO annotations
- Code for our CVPR'16 paper on Learning Visually Grounded Word Embeddings
If you like this layout/page, see this to build your own using github+jekyll