SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for January
-
SmartCare: A Discovery and Clinical Decision Support System for Personalized Healthcare
Fri, Jan 15, 2016 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Mihaela van der Schaar, UCLA
Talk Title: Ai Seminar: SmartCare: A Discovery and Clinical Decision Support System for Personalized Healthcare
Series: Artificial Intelligence Seminar
Abstract: Modern technology makes it possible to collect more and more data for each patient; modern record-keeping makes it possible to access more and more data from past patients. Unfortunately, it has so far not been possible to make use of all this data to create a system of personalized diagnosis and treatment.Instead, current diagnosis and treatment continues to rely on Clinical Practice Guidelines, which are largely based on experience and opinion rather than on scientific analysis and evidence, are geared towards the -lowest common
denominator-, ignore the strengths of a given institution (e.g. specialists, technology) and are targeted toward a representative patient rather than toward the unique characteristics of the current patient.
In this talk, I will present a novel framework SmartCare that integrates the (longitudinal, multi-modal) data of the current patient (demographic information, current medical condition, medical/family history, availability of home care, etc.) with what has been learned from previous patients to recommend personalized diagnosis and treatment. To do so,
SmartCare must overcome enormous conceptual, theoretical and practical challenges, some of which will
be discussed in the talk.
SmartCare has already recorded an important (even remarkable) success: a vastly improved procedure for breast cancer screening. Current clinical practice (BI-RADS: Breast Imaging
Report and Data System) results in an enormous number of false positives, leading to many further invasive and unnecessary procedures (including surgery) that involve needless risk,
suffering and expense. The SmartCare procedure reduces false positives by 39% while maintaining the same misdetection rate. In the U.S. alone, this means 80,000 fewer false positives per year.
Biography: Mihaela van der Schaar is Chancellor's Professor of Electrical Engineering at University of California, Los Angeles. She received an NSF CAREER Award (2004), the Okawa Foundation Award (2006), the IBM Faculty Award (2005, 2007, 2008), and several best paper awards, including the 2011 IEEE Circuits and Systems Society Darlington Award Best Paper Award. She holds 33 granted US patents. She is also the founding and managing director of the UCLA Center for Engineering Economics, Learning, and Networks (see http://netecon.ee.ucla.edu). Her research interests are in data science, medical informatics, machine learning, game theory, and network science. For more information about her research visit: http://medianetlab.ee.ucla.edu/
Host: Greg Ver Steeg
Webcast: http://webcasterms1.isi.edu/mediasite/Viewer/?peid=2cad8868c8314e248c72e1ba11c4c0e61dLocation: Information Science Institute (ISI) - 6th Flr Conf Rm # 689, Marina Del Rey
WebCast Link: http://webcasterms1.isi.edu/mediasite/Viewer/?peid=2cad8868c8314e248c72e1ba11c4c0e61d
Audiences: Everyone Is Invited
Contact: Peter Zamar
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-Learning Open Domain Knowledge From Text
Fri, Jan 15, 2016 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Gabor Angeli, Stanford University
Talk Title: Learning Open Domain Knowledge From Text
Series: Natural Language Seminar
Abstract: The increasing availability of large text corpora holds the promise of acquiring an unprecedented amount of knowledge from this text. However, current techniques are either specialized to particular domains or do not scale to large corpora. This dissertation develops a new technique for learning open-domain knowledge from unstructured web-scale text corpora. A first application aims to capture common sense facts: given a candidate statement about the world and a large corpus of known facts, is the statement likely to be true? We appeal to a probabilistic relaxation of natural logic -- a logic which uses the syntax of natural language as its logical formalism -- to define a search problem from the query statement to its appropriate support in the knowledge base over valid (or approximately valid) logical inference steps. We show a 4x improvement at retrieval recall compared to lemmatized lookup, maintaining above 90% precision. This approach is extended to handle longer, more complex premises by segmenting these utterance into a set of atomic statements entailed through natural logic. We evaluate this system in isolation by using it as the main component in an Open Information Extraction system, and show that it achieves a 3% absolute improvement in F1 compared to prior work on a competitive knowledge base population task. A remaining challenge is elegantly handling cases where we could not find a supporting premise for our query. To address this, we create an analogue of an evaluation function in game playing search: a shallow lexical classifier is folded into the search program to serve as a heuristic function to assess how likely we would have been to find a premise. Results on answering 4th grade science questions show that this method improves over both the classifier in isolation and a strong IR baseline, and achieves the best published results on the task.
Biography: Gabor is a new graduate from Chris Manning's natural language processing lab. He graduated with a BS in electrical engineering/computer science from UC Berkeley in 2010, and defended his Ph.D. in the fall of 2015. His research focuses on natural language understanding, ranging from relation extraction and knowledge base population, textual entailment, common-sense reasoning, and question answering. He has led the Stanford knowledge base population project for the past three years, with Stanford ranking 5th, 1st, and 1st (tied) among teams participating in the TAC-KBP competition over those three years. In addition to publications at ACL, EMNLP and NAACL, he co-authored an EMNLP best dataset paper on collecting a large dataset for textual entailment. Outside of academia, he was the NLP architect for Baarzo in 2014 (acquired by Google), and is currently a fellow at XSeed Capital. In his free time, Gabor enjoys hiking, board games, and binge-watching Netflix shows.
Host: Xing Shi and Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - 6th Flr Conf Rm # 689, 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
Fri, Jan 22, 2016 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Yisong Yue, Cal Tech
Talk Title: A Decision Tree Framework for Data-Driven Speech Animation
Abstract: n many animation projects, the animation artist typically spends significant time animating the face, which involves many labor-intensive tasks that offer little potential for creative expression. One particularly tedious task is speech animation: animating the face to match spoken audio. Indeed, the often prohibitive cost of speech animation has limited the types of animations that are feasible, including localization to different languages.
In this talk, I will show how to view speech animation through the lens of data-driven sequence prediction. In contrast to previous sequence prediction settings, speech animation is an instance of contextual spatiotemporal sequence prediction, where the output is continuous and high-dimensional (e.g., a configuration of the lower face), and also depends on an input context (e.g., audio or phonetic input).
I will present a decision tree framework for learning to generate context-dependent spatiotemporal sequences given training data. This approach enjoys several attractive properties, including ease of training, fast performance at test time, and the ability to robustly tolerate corrupted training data using a novel latent variable approach. I will showcase this approach in a case study on speech animation, where our approach outperforms several competitive baselines in both quantitative and qualitative evaluations, and also demonstrates strong robustness to corrupted training data.
This is joint work with Taehwan Kim, Sarah Taylor, Barry-John Theobald, and Iain Matthews.
Biography: Yisong Yue is an assistant professor in the Computing and Mathematical Sciences Department at the California Institute of Technology. He was previously a research scientist at Disney Research. Before that, he was a postdoctoral researcher in the Machine Learning Department and the iLab at Carnegie Mellon University. He received a Ph.D. fromCornell University and a B.S. from the University of Illinois at Urbana-Champaign.
Yisong's research interests lie primarily in the theory and application of statistical machine learning. He is particularly interested in developing novel methods for spatiotemporal reasoning, structured prediction, interactive learning systems, and learning with humans in the loop. In the past, his research has been applied to information retrieval, recommender systems, text classification, learning from rich user interfaces, analyzing implicit human feedback, data-driven animation, sports analytics, policy learning in robotics, and adaptive routing & allocation problems.
Host: Ashish Vaswani
More Info: http://webcasterms1.isi.edu/mediasite/SilverlightPlayer/Default.aspx?peid=6147027e077e4570919a58730193abf91d
Location: Information Science Institute (ISI) - 11th floor large conference room
Audiences: Everyone Is Invited
Contact: Kary LAU
Event Link: http://webcasterms1.isi.edu/mediasite/SilverlightPlayer/Default.aspx?peid=6147027e077e4570919a58730193abf91d
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-EXTRACTING USER INFORMATION FROM ONLINE SOCIAL MEDIA
Fri, Jan 22, 2016 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Jiwei Li, Stanford University
Talk Title: EXTRACTING USER INFORMATION FROM ONLINE SOCIAL MEDIA
Series: Natural Language Seminar
Abstract: The overwhelming popularity of online social media creates an unprecedented opportunity to display aspects of oneself. Inferring information about these users has the potential to benefit many downstream applications such as recommendation engines and targeted advertising. In this talk I will show how to extract important personal information such as major life events and personal attributes (e.g., gender, education, job) from social evidence such as the text produced by users and their friends and from properties of their social network. I will describe algorithms making use of a variety of frameworks, including distant supervision, and a deep learning architecture that learns user representations by integrating many heterogeneous social signals.
Biography: Jiwei Li is a PH.D. student in the computer science department at Stanford University, working with Prof. Dan Jurafsky. His research interests include discourse, language generation, and social networks, with a focus on deep learning methods. Jiwei receives his B.S. from Peking University in 2012. He was rewarded the Facebook Fellowship in 2015.
Host: Xing Shi and Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Webcast: http://webcasterms1.isi.edu/mediasite/Viewer/?peid=6b5348f2f8dc4a4dbb595eca444410d51dLocation: Information Science Institute (ISI) - 6th Flr Conf Rm # 689, Marina Del Rey
WebCast Link: http://webcasterms1.isi.edu/mediasite/Viewer/?peid=6b5348f2f8dc4a4dbb595eca444410d51d
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. -
Special Seminar
Thu, Jan 28, 2016 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Natali Ruchansky, Postdoc - Boston University
Talk Title: Learn to query and query to learn.
Series: AI Seminar
Abstract: Working with data has never been easy: Netflix wishes to make movies recommendations to users, but their data is incomplete. Biologists have large interesting protein-protein interaction networks, but they don't know how to extract useful insight from them.
The above examples can be viewed as dealing with two components: the data and the inference process.
In the case of Netflix the latter is known and algorithms are needed to complete the missing data; I will call this Learning to Query. In the biology example the data is readily available and an algorithm is needed to query the graph for discoveries; I will call this Querying to Learn. In my work I addressed these two converse problems and proposed simple, but effective algorithms to solve them.
In this talk I will first present my work on the Learning to Query problem through the lens of matrix completion. I will discuss the new problem of Active Matrix Completion which asks to first analyzes the quality of the available data, such as movie ratings on Netflix, then perform the completion and inference, or movie recommendation. I will then present a new algorithm called Order&Extend that tackles the Active Completion problem. By framing the problem in terms of linear systems, Order&Extend identifies which portions of the data do or do not have enough information, suggests how the data can be augmented, and finally produces a completion.
In the second half of my talk I will present my work on the Querying to Learn through the lens of graph mining. Here there is a data set available, and in particular there are some query-nodes of interest; the biologists have a protein-protein-interaction network and wish to study the interactions between three particular proteins. I will present the new notion of a Wiener-Connector that isolates interesting connections among the query-nodes by utilizing the simple relationship of shortest paths. I will then discuss the algorithm for finding the Wiener-Connector along with its applicability and utility, for example, in identifying possible protein-disease associations and providing outputs that are easy to interpret and visualize, making it useful across different domains.
Biography: I am a PhD student in the Computer Science Department at Boston University.
I am a member of the Data Managment Group, and I work with Professor Evimaria Terzi and Professor Mark Crovella.
While 'passionately curious' about (too) many things, my research focus is algorithmic data mining, mathematics, and networks. In particular I am currently working on problems in graph mining, and in matrix and tensor completion through the lense of linear algebra.
Host: Kristina Lerman
Webcast: http://webcasterms1.isi.edu/mediasite/SilverlightPlayer/Default.aspx?peid=0ac4e800c18744fcbac14781671b6d481dLocation: Information Science Institute (ISI) - 11th floor Large CR
WebCast Link: http://webcasterms1.isi.edu/mediasite/SilverlightPlayer/Default.aspx?peid=0ac4e800c18744fcbac14781671b6d481d
Audiences: Everyone Is Invited
Contact: Alma Nava / Information Sciences Institute
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-Leveraging the Social Web to Enable Open-Domain Interactive Storytelling
Fri, Jan 29, 2016 @ 03:00 PM - 04:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Reid Swanson, USC/ICT
Talk Title: Leveraging the Social Web to Enable Open-Domain Interactive Storytelling
Series: Natural Language Seminar
Abstract: Storytelling is an integral part of human interaction and critical to nearly all forms of entertainment. Since the introduction of TALE-SPIN over thirty years ago, automating the process of storytelling has been an active area of research. However, despite the incredible advances in other areas of computer science, such as 3D graphics and computational physics, that have enabled dazzling immersive interactive environments, there has been little progress in delivering automated *stories* that have the richness and complexity we expect in this genre of discourse.
In this talk I will primarily discuss work done during my thesis that leverages the vast amounts of knowledge hidden implicitly in the social web in order to enable a text-based open-domain interactive storytelling system. In this system the human and computer take turns writing sentences of an emerging fictional story on any topic the author chooses. The system uses an architecture inspired by case-based reasoning with a knowledge base of over a million personal stories about the daily lives and experiences of ordinary people. At each turn the system selects a sentence from the corpus that tries to maximize the semantic and discourse coherence given the text of the story so far.
I will also describe how crowd-sourcing communities were used to collect thousands of collaborative stories with the system and tens of thousands of ratings from hundreds of participants on several subjective evaluation criteria. The best models show significant improvements over the baseline and are judged to be indistinguishable from entirely human written weblog stories from a held out part of the collection.
I will conclude with some more recent and ongoing research that examines additional methods of evaluation and new models of narrative generation based on Recurrent Neural Networks.
Biography: Reid Swanson received his PhD in Computer Science from the University of Southern California in 2010 where he focused on a large-scale text-based interactive storytelling system. His primary research interest is in large-scale open-domain interpretation and generation of interactive narratives.
After graduating he spent a year at the Walt Disney Imagineering Research & Development lab in Glendale, CA. At Disney he worked with an interdisciplinary team of industry engineers, academics, artists and performers to develop technologies for bringing persistent interactive storytelling to select groups of guests at their theme parks and resorts.
From 2011 until 2015, Reid worked as a postdoc at UC Santa Cruz where he participated in a range of different projects. As part of the SIREN project, with Arnav Jhala, he investigated games for teaching conflict resolution management. On the SSIM project, with Michael Mateas, he helped research and develop virtual training environments targeting the military and law enforcement agencies to help prevent conflict escalation in unknown social environments. With Marilyn Walker, he also investigated automated methods for analyzing and mining prototypical arguments on internet debate forums about controversial topics such as gun control, gay marriage and evolution.
In August of 2015 he rejoined the Institute of Technologies as a Research Scientist where he is researching the role of narrative structure in the persuasiveness of an intended message embedded in the story across different cultures.
Host: Xing Shi and Kevin Knight
More Info: http://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - 6th Flr Conf Rm # 689, 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.
