Conferences, Lectures, & Seminars
Events for February
-
AI Seminar
Fri, Feb 03, 2017 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Elias Bareinboim, Assistant Professor, Purdue University
Talk Title: The Data-Fusion Problem: Causal Inference and Reinforcement Learning
Abstract: Machine Learning is usually dichotomized into two categories, passive and active which, by and large, are studied separately. Reality is more demanding. Passive and active modes of operation are but two extremes of a rich spectrum of data collection modes that generate the bulk of the data available in practical, large scale situations. In typical medical explorations, for example, data from multiple observations and experiments are collected, coming from distinct experimental setups, different sampling conditions, and heterogeneous populations. Similarly, in a more basic setting, a baby learns from its environment by both passively observing others and interacting with its environment by actively performing interventions. In this task, I will review the theory of structural causality and use it to explain the relationship between causal inference and reinforcement learning . Further, I will formulate and discuss a collection of inference tasks that lie in the intersection of RL and causal inference, including personalized decision making.
Biography: Elias Bareinboim is an assistant professor in the Department of Computer Science at Purdue University. His research focuses on causal and counterfactual inference and their applications to data driven fields. Bareinboim received a Ph.D. in Computer Science from UCLA advised by Judea Pearl. His doctoral thesis was the first to propose a general solution to the problem of data fusion and to provide practical methods for combining datasets generated under different experimental conditions. Bareinboims recognitions include IEEE AIs 10 to Watch, the Dan David Prize Scholarship, the Yahoo! Key Scientific Challenges Award, and the 2014 AAAI Outstanding Paper Award.
Host: Mayank Kejriwal
Audiences: Everyone Is Invited
Contact: Kary Lau
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
NL Seminar Recurrent Neural Networks for Narrative Prediction
Fri, Feb 03, 2017 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Melissa Roemmele, USC ICT
Talk Title: Recurrent Neural Networks for Narrative Prediction
Series: Natural Language Seminar
Abstract: Narrative prediction involves predicting what happens next in a story. This task has a long history in AI research but is now getting more recognition in the NLP community. In this talk I will describe three different evaluation schemes for narrative prediction, one of which the Story Cloze Test is the shared task for this years LSDSem workshop at EACL. I will present my ongoing efforts to develop Recurrent Neural Network based models that succeed on these evaluation frameworks, and discuss the particular challenges posed by each of them.
Biography: I am a PhD candidate at the USC Institute for Creative Technologies advised by Andrew Gordon in the Narrative Group. My thesis research explores machine learning approaches to automatically generating text based stories. I am interested in using this research to stimulate creativity in writing. More broadly, I am excited by any opportunity to use automated analysis of text data to give people new insights and ideas.
Host: Marjan Ghazvininejad and Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - 11th Flr Conf Rm # 1135, Marina Del Rey
Audiences: Everyone Is Invited
Contact: Peter Zamar
Event Link: http://nlg.isi.edu/nl-seminar/
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
NL Seminar-THE LIMITS OF UNSUPERVISED SYNTAX AND THE IMPORTANCE OF GROUNDING IN LANGUAGE ACQUISITION
Fri, Feb 10, 2017 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Yonatan Bisk, USC/ISI
Talk Title: THE LIMITS OF UNSUPERVISED SYNTAX AND THE IMPORTANCE OF GROUNDING IN LANGUAGE ACQUISITION
Series: Natural Language Seminar
Abstract: The future of self driving cars, personal robots, smart homes, and intelligent assistants hinges on our ability to communicate with computers. The failures and miscommunications of Siri style systems are untenable and become more problematic as machines become more pervasive and are given more control over our lives. Despite the creation of massive proprietary datasets to train dialogue systems, these systems still fail at the most basic tasks. Further, their reliance on big data is problematic. First, successes in English cannot be replicated in most of the six thousand plus languages of the world. Second, while big data has been a boon for supervised training methods, many of the most interesting tasks will never have enough labeled data to actually achieve our goals. It is, therefore, important that we build systems which can learn from naturally occurring data and grounded, situated interactions.
In this talk I will discuss work from my thesis on the unsupervised acquisition of syntax which harnesses unlabeled text in over a dozen languages. This exploration leads us to novel insights into the limits of semantics free language learning. Having isolated these stumbling blocks I will then present my recent work on language grounding where we attempt to learn the meaning of several linguistic constructions via interaction with the world.
Biography: Yonatan Bisk has research that focuses on Natural Language Processing from naturally occurring data unsupervised and weakly supervised data. He is a postdoc researcher with Daniel Marcu at USCs Information Sciences Institute. Previously, he received his PhD from the University of Illinois at Urbana Champaign under Julia Hockenmaier and his BS from the University of Texas at Austin.
Host: Marjan Ghazvininejad and Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - 6th Flr -CR#689 (ISI/Marina Del Rey)
Audiences: Everyone Is Invited
Contact: Peter Zamar
Event Link: http://nlg.isi.edu/nl-seminar/
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor. -
AI Seminar - Interview Talk
Fri, Feb 17, 2017 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Jay Pujara, University of California, Santa Cruz
Talk Title: Probabilistic models for large, noisy, and dynamic data
Abstract: We inhabit a vast, uncertain, and dynamic universe. To succeed in such an environment, artificial intelligence approaches must handle massive amounts of noisy, changing evidence. My research addresses the problems of building scalable, probabilistic models amenable to online updates. To illustrate the potential of such models, I present my work on knowledge graph identification, which jointly resolves the entities, attributes, and relationships in a knowledge graph by combining statistical NLP signals and semantic constraints. Using probabilistic soft logic, a statistical relational learning framework I helped develop, I demonstrate how knowledge graph identification can scale to millions of uncertain candidate facts and tens of millions of semantic dependencies in real-world data while achieving state-of-the-art performance. My work further extends this scalability by adopting a distributed computing approach, reducing the inference time of knowledge graph identification from two hours to ten minutes. Updating large, collective models like those used for knowledge graphs with new information poses a significant challenge. I develop a regret bound for probabilistic models and use this bound to motivate practical algorithms that support low-regret updates while improving inference time over 65%. Finally, I highlight several active projects in sustainability, bioinformatics, and mobile analytics that provide a promising foundation for future research.
Biography: Jay Pujara is a postdoctoral researcher at the University of California, Santa Cruz whose principal areas of research are machine learning, artificial intelligence, and data science. He completed his PhD at the University of Maryland, College Park and received his MS and BS at Carnegie Mellon University. Prior to his PhD, Jay spent six years at Yahoo! working on mail spam detection, user trust, and contextual mail experiences, and he has also worked at Google, LinkedIn and Oracle. Jay is the author of over twenty peer-reviewed publications and has received three best paper awards for his work. He is a recognized authority on knowledge graphs, and has organized the Automatic Knowledge Base Construction (AKBC) workshop, recently presented a tutorial on knowledge graph construction, and has had his work featured in AI Magazine. For more information, visit https://www.jaypujara.org
Host: Craig Knoblock
More Info: http://webcastermshd.isi.edu/Mediasite/Play/1ed0700540864caabaedfc675e89543e1d
Location: Information Science Institute (ISI) -
Audiences: Everyone Is Invited
Contact: Kary LAU
Event Link: http://webcastermshd.isi.edu/Mediasite/Play/1ed0700540864caabaedfc675e89543e1d
This event is open to all eligible individuals. USC Viterbi operates all of its activities consistent with the University's Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation, or any other prohibited factor.
