Fall 2019 Schedule
| 2019.08.28 | Welcome Meeting, Plan fall schedule |
| 2019.09.04 | Jon ACL best paper nominee: https://arxiv.org/abs/1907.06679.
Xiaolei ACL paper: https://www.aclweb.org/anthology/P19-1403. Lunch |
| 2019.09.05 (Thurs 3:30 pm) | CS Colloquium - Dan Jurafsky |
| 2019.09.11 | Yoshi Presenting a NAACL paper https://www.aclweb.org/anthology/N19-1162 |
| 2019.09.18 | Akanksha presenting ACL paper https://www.aclweb.org/anthology/P19-1568 |
| 2019.09.25 | AIDA virtual site visit |
| 2019.10.02 | Rakuten overview, Abhidip presenting BERT[1]and XLNet [2] |
| 2019.10.09 | Ghazaleh, Situated Open World Reference Resolution for Human-Robot Dialogue[3] |
| 2019.10.16 (10 am - 1 pm) | CLASIC/CLEAR Open House & Industry day (lunch at Fleming). 10 min talks and posters by PhD/MS students |
| 2019.10.17 (Thurs 3:30 pm) | CS Colloquium: Jinho Choi |
| 2019.10.23 | Ali Almelhem (Economics department) presenting own research |
| 2019.10.30 | Rehan presenting his research. Slides[4]
Kristin EMNLP-LOUHI practice talk Vivian and Jon EMNLP posters |
| 2019.11.06 | Michael Reagan: Extraction of force-dynamic image schemas for event structure representation of procedural text,
Abstract: Event structure decomposition is integral to machine reading comprehension, dialog state tracking, and other natural language understanding tasks where models of entity states and temporality as well as of event-event, participant-event, and participant-participant relations are desirable. Previous work in event representation has yet to show how organizing conceptual structures into well-defined, flexible units representing both background knowledge of the world and how that knowledge is expressed in a certain language might improve AI reasoning capabilities. In preparation for my dissertation research into this question, in this talk I will argue for a fine-grained approach to event representation based on theoretical work done in cognitive semantics: force dynamics (Talmy 1988; Croft 2012), a image schematic model of causation that my research may show can support common-sense reasoning applications. Event representation is often done at the macro-level, with a focus on sequences of events in narratives, procedures, etc.; in contrast, at a micro-level we examine subevents of clausal events by employing the entity-centric, fine granularity of a force-dynamic approach to characterize participant interaction as a function of time, qualitative change, and transmission of force. As a proof of concept linking micro- and macro-level analyses, I propose designing, implementing, and evaluating a computational model for the extraction of participant histories as storylines from scientific, procedural text. One task will be to compare how well force-dynamic event structure can be extracted using a symbolic approach based on a model of argument structure and verb classes versus a neural approach using pre-trained language models. A second task will be to examine the use of dynamic knowledge graphs as a representation for evolving fine-grained storyline event structure and the tracking of entity states. A third task will be to examine the applicability of image schemas for effect prediction via a generative process with force-dynamic structures as priors. The primary hypotheses of the study will be examined along with a tentative timeline for how my research may progress.
|
| 2019.11.13 | Chelsea prelim. |
| 2019.11.20 | Parth Jawale, presenting Tao Li, Vivek Srikumar, Augmenting Neural Nets with First Order Logic, [5], ACL 2019, follow-on paper, Tao Li, Vivek Gupta, Maitrey Mehta, Vivek Srikumar, A Logic-Driven Framework for Consistency of Neural Models, [6], EMNLP 2019. |
| 2019.11.27 | Fall break |
| 2019.12.04 | Sam: Extractive Adversarial Networks: High-Recall Explanations for Identifying Personal Attacks in Social Media Posts [7]
Abstract: We introduce an adversarial method for producing high-recall explanations of neural text classifier decisions. Building on an existing architecture for extractive explanations via hard attention, we add an adversarial layer which scans the residual of the attention for remaining predictive signal. Motivated by the important domain of detecting personal attacks in social media comments, we additionally demonstrate the importance of manually setting a semantically appropriate `default' behavior for the model by explicitly manipulating its bias term. We develop a validation set of human-annotated personal attacks to evaluate the impact of these changes. |
| 2019.12.11 | Sarah Moeller - Lorelei denouement |
| 2019.12.18 | Final exams |
