Indian students interested in NLP, looking to do some solid research before starting graduate school, should consider: 1. Pre-doctoral program at Google w/ Partha Talukdar and team. 2. MSR RF program w/ Monojit Choudhury kalikabali, Sunayana and others. (tinyurl.com/y5otnw8v)

New large-scale dataset of American Sign Language, How2sign (how2sign.github.io) to be presented at #CVPR2021. Significantly broader vocabulary and larger size than previous public ASL datasets!

For complex sequence tasks, natural decomposition in cascaded models yields superior inference over E2E models. In our #NAACL2021 paper arxiv.org/abs/2105.00573, we present Searchable Hidden Intermediates to decompose sequence models while maintaining E2E differentiability. (1/N)

We're organizing the 1st workshop on Crossmodal Social Animation ICCV2021 ! We welcome submission on all areas and applications of animation and interaction in social settings. Deadlines Main Track: July 24, 2021 Late Breaking: Sep 7, 2021 webpage: sites.google.com/view/xs-anim

We're hosting an AMA on with Amanda Duarte, who presented her work with Shruti Palaskar Lucas Ventura Deepti Ghadiyaram @jorditorresai Xavi Giró et al. on the How✌Sign dataset at #CVPR2026 this year. 🔗: wandb.me/ama-amanda-dua… It's video data to train DNNs to understand ASL!

I defended my PhD! 🥳 I am incredibly fortunate to have found the best advisors in Profs Florian Metze and Alan W Black. I am also thankful to my committee members Yonatan Bisk - not atttending any conferences and Cordelia Schmid for their guidance; and mentors, friends, family for being by my side throughout!

Thrilled to share that I’ve joined Apple Siri Understanding as an ML Scientist. Many thanks to Saurabh Adya and team for giving me this opportunity to work on super challenging and impactful areas, and to my PhD advisors Florian Metze and Alan Black for their continued guidance!

If you work on text generation from images+text, you (ironically) probably wonder what “general-purpose” multimodal model to use. In our #EMNLP2022 Findings paper, we study this with self-rationalization and show that the answer is not simple. 1/ arxiv.org/abs/2205.11686

Dear friends, please spread this tweet! If you're interested, please do not hesitate to contact me or David. We're looking forward to your applications!