*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Toward Grounded Spatio-Temporal Reasoning
Committee:
Dr. Ghassan AlRegib, ECE, Chair , Advisor
Dr. Zsolt Kira, CoC
Dr. Patricio Vela, ECE
Dr. Marcus Rohrbach, Facebook Research
Dr. Devi Parikh, CoC
Abstract:
To develop an Artificial Intelligence (AI) system that can understand the world around us, it needs to be able to interpret and reason about the world we see and the language we speak. In recent years, there has been a lot of attention to research at the intersection of vision, temporal reasoning, and language. One of the major challenges is how to ensure proper grounding and perform reasoning across multiple modalities given the heterogeneity resides in the data when there is no or weak supervision of the data. For example, (1) in Vision-and-Language Navigation, how to ensure the navigation agent to identify which part of the instruction has been completed or ongoing and which part is potentially needed for the next action selection, and how to identify which direction to go by finding the part of the instruction that corresponds to the observed images. (2) In visual understanding, how to efficiently leverage object-level features for downstream visual understanding tasks like action recognition and visual captioning, how to detect interactions/relationships when there is no or only weak supervision from classification labels or ground-truth image/video descriptions. (3) In visual captioning, how to enforce generated sentences to be properly grounded without ground-truth grounding annotations. In this talk, our goal is to leverage spatial, temporal, and language inputs for both visual and textual understanding. I will show (1) how to equip the concept of self-monitoring to a seq-to-seq model in order to develop a visual-textual co-grounded navigation agent that can follow human commands and perform backtracking when necessary, (2) how to efficiently achieve object-level fine-grained video understanding for both human action recognition and video captioning, and (3) how to enforce the visual captioning models to generate grounded descriptions without ground-truth annotations via a novel cyclical training regimen that adds no extra computation during inference.
