Post-Doc in Machine Learning/Computer Vision/Language Processing applied to Sign Language Analysis

Applications are sought for a post-doctoral position as part of an ongoing project on automatic sign language analysis and recognition. The post-doc will work primarily with TTI-Chicago (TTIC) faculty Greg Shakhnarovich and Karen Livescu. The work is part of a broad interdisciplinary project including both recognition and linguistics of American Sign Language (ASL), in collaboration with U. Chicago linguistics faculty Diane Brentari and Jason Riggle. The post-doc will also have opportunities to collaborate with other TTIC and U. Chicago researchers as relevant, and should be able to interact with colleagues in a broad range of disciplines.

The goal of the work is to advance automatic recognition and analysis of American Sign Language (ASL). The work includes both development of core learning/vision/language methods and application to ASL. Some recent publications related to the project are below.

The specific focus of the post-doc's research will be determined by mutual agreement based on the post-doc's background and interests, and may include any of: new graphical models, deep networks, and other learning techniques; techniques inspired by speech recognition and NLP; hand pose tracking/estimation constrained by the grammar of sign; domain adaptation/normalization for different subjects and signing styles; and work with new data sources, including depth sensors (e.g. Kinect) and motion capture. Prior knowledge of/interest in ASL is preferred but not required.

The post-doc will be based at TTI-Chicago, a computer science graduate institute located on the University of Chicago campus, with vibrant research groups in machine learning, computer vision, and speech and language processing. The position is for 1-2 years.

Advising and teaching opportunities are likely to be available if desired, but this is not a requirement of the position.

Applicants are expected to have strong qualifications in machine learning or related fields, and ideally also computer vision and/or speech and language processing. Applicants should have a PhD or expected PhD in computer science, electrical engineering, statistics, computational linguistics, or a related field.

The start date is flexible. Compensation includes a competitive salary and benefits plan. There are no citizenship requirements for the post-doc. The position will remain open until filled.

Please direct questions and interest in the position to Complete applications should be sent to the same address and consist of:

1) brief cover letter, describing your interest in and qualifications for the position
2) curriculum vitae including publication list
3) contact information for two or more researchers who can provide letters of reference

Related publications/preprints:


T. Kim, W. Wang, H. Tang, and K. Livescu
"Signer-independent fingerspelling recognition with deep neural network adaptation"

M. Mostajabi, P. Yadollahpour, and G. Shakhnarovich
"Feedforward semantic segmentation with zoom-out features"
CVPR 2015.

S. Trivedi, D. McAllester, and G. Shakhnarovich
"Discriminative metric learning by neighborhood gerrymandering"
NIPS 2014.

H. Tang, K. Gimpel, and K. Livescu
"A comparison of training approaches for discriminative segmental models"
Interspeech 2014.

T. Kim, G. Shakhnarovich, and K. Livescu
"Fingerspelling recognition with semi-Markov conditional random fields"
ICCV 2013.

T. Kim, K. Livescu, and G. Shakhnarovich
"American Sign Language fingerspelling recognition with phonological feature-based tandem models"
SLT 2012.

U. Chicago:

J. Keane and D. Brentari
"Fingerspelling: Beyond Handshape Sequences" The Oxford Handbook of Deaf Studies in Language: Research, Policy, and Practice Oxford, 2015.

J. Keane
Towards an articulatory model of handshape: What fingerspelling tells us about the phonetics and phonology of handshape in American Sign Language Ph.D. Dissertation, U. Chicago, 2014.