SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for April
-
NL Seminar - 30 Years of Perplexity
Thu, Apr 04, 2024 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Kevin Knight, Threeven Labs
Talk Title: 30 Years of Perplexity
Abstract: REMINDER: If you’re an outside visitor who wishes to attend in person, kindly send a message to nlg DASH seminar DASH host AT isi.edu at least 1 business day prior with your full name, job title and professional affiliation. Please arrive to ISI at least 15 minutes prior to the start of the seminar. If you do not have access to the 11th Floor, please check in at the 10th floor main reception desk to register as a visitor and someone will escort you to the conference room location. For more information on the NL Seminar series and upcoming talks, please visit: https://nlg.isi.edu/nl-seminar/ NLP scientists have been trying for decades to accurately predict the next word in running text. Why were we so determined to succeed at this strange task? How did we track our successes (and failures)? Why was word prediction at the center of early statistical work in text compression, machine translation, and speech recognition? Will it lead to artificial general intelligence (AGI) in the 2020s? I’ll attempt to answer these questions with anecdotes drawn from three decades of research in NLP, text compression, and code-breaking.
Biography: Dr. Kevin Knight served on the faculty of the University of Southern California (26 years), as Chief Scientist at Language Weaver, Inc. (9 years), and as Chief Scientist for Natural Language Processing at Didi Global (4 years). He received a PhD in computer science from Carnegie Mellon University and a bachelor's degree from Harvard University. Dr. Knight's research interests include machine translation, natural language generation, automata theory, decipherment of historical documents, and number theory. He has co-authored over 150 research papers on natural language processing, as well as the widely adopted textbook "Artificial Intelligence" (McGraw-Hill). Dr. Knight served as President of the Association for Computational Linguistics (ACL) in 2011, as General Chair for ACL in 2005, as General Chair for the North American ACL (NAACL) in 2016, and as co-program chair for the inaugural Asia-Pacific ACL (2020). He received an Outstanding Paper Award at NAACL 2018, and Test-of-Time awards at ACL 2022 and ACL 2023. He is a Fellow of the ACL, the Association for the Advancement of Artificial Intelligence (AAAI), and the USC Information Sciences Institute (ISI). Subscribe here to learn more about upcoming seminars: https://www.isi.edu/events/
Host: Jonathan May and Justin Cho
More Info: https://nlg.isi.edu/nl-seminar/
Location: Information Science Institute (ISI) - Live Only at ISI-Conf Rms #1135-37
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://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- Human and Machine Conditions Favoring Innovation in Science
Fri, Apr 12, 2024 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Brian Uzzi, Northwestern University.
Talk Title: Human and Machine Conditions Favoring Innovation in Science
Series: AI Seminar
Abstract: Innovation involves recombining past knowledge. The increasing rate at which scientific knowledge is expanding should bode well for innovation. Nevertheless, problems related to creating replicable science and appraising the merits of new ideas threaten innovation. In study 1, we use artificial intelligence to examine the replication problem in science. We estimate a paper’s replicability using ground truth, manual replication data, and then test the model’s accuracy on an extensive set of out-of-sample studies. The model’s accuracy is better than reviewer base rates and on par with prediction markets. We then conduct a discipline-wide census of replicability in psychology for the past 20 years. Replicability failures varies widely by subfield and is highest for experimental studies and papers receiving media coverage. In study 2, we investigate how the merits of innovative ideas communicated in science. Here we conduct semantic analyses of grant application success with a focus on scientific promotional language, which purportedly helps to convey an innovative idea’s originality. Our analysis examines the full text of tens of thousands of both funded and unfunded grants from three leading public and private funding agencies. We find promotional language in a grant proposal is associated with up to a doubling in its probability of being funded, with a grant’s intrinsic innovativeness, and with its predicted citation impact and productivity. Lastly, a computer experiments that substitute a grant’s promotional language with neutral synonyms indicates that promotional language may communicate the merits of ideas through cognitive activation.
Biography: Brian Uzzi the Richard L. Thomas Distinguished Professor of Leadership at the Kellogg School of Management, Northwestern University. He also is Co-Director of the Northwestern University Institute on Complex Systems and Data Science (NICO), holds professorship in sociology and the McCormick School of Engineering, and writes a column on AI and business for Forbes. Brian’s work focuses on the link between social networks and human achievement and the role of AI in mind + machine partnerships. Brian has been awarded over 30 teaching and research prizes worldwide, including the Euler Award. He has been on the faculties of Harvard, INSEAD, University of Chicago, and Berkeley, and is a Fellow of the Network Science Society. His work has been funded by DARPA, NSF, and other foundations, is widely cited, and appears frequently in major media outlets worldwide. Before entering science, Brian worked as a carpenter and a musician. His PhD is from Stony Brook University in sociology.
Host: Zhuoyu Shi and Karen Lake
More Info: https://www.isi.edu/events/4641/ai-seminar-human-and-machine-conditions-favoring-innovation-in-science/
Webcast: https://usc.zoom.us/j/95888595423?pwd=VHBLa041dUJWcWx0NEhuYmQrV29ZQT09Location: Information Science Institute (ISI) - Virtual Only
WebCast Link: https://usc.zoom.us/j/95888595423?pwd=VHBLa041dUJWcWx0NEhuYmQrV29ZQT09
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/events/4641/ai-seminar-human-and-machine-conditions-favoring-innovation-in-science/
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 - DeLLMa: A Framework for Decision Making Under Uncertainty with Large Language Models
Thu, Apr 18, 2024 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Ollie Liu, USC, USC
Talk Title: DeLLMa: A Framework for Decision Making Under Uncertainty with Large Language Models
Abstract: REMINDER: Meeting hosts only admit on-line guests that they know to the Zoom meeting. Hence, you’re highly encouraged to use your USC account to sign into Zoom. If you’re an outside visitor, please inform us at (nlg-seminar-host(at)isi.edu) to make us aware of your attendance so we can admit you. Specify if you will attend remotely or in person at least one business day prior to the event Provide your: full name, job title and professional affiliation and arrive at least 10 minutes before the seminar begins. If you do not have access to the 6th Floor for in-person attendance, please check in at the 10th floor main reception desk to register as a visitor and someone will escort you to the conference room location. Abstract: Large language models (LLMs) are increasingly used across society, including in domains like business, engineering, and medicine. These fields often grapple with decision-making under uncertainty, a critical yet challenging task. In this paper, we show that directly prompting LLMs on these types of decision-making problems yields poor results, especially as the problem complexity increases. To overcome this limitation, we propose DeLLMa (Decision-making Large Language Model assistant), a framework designed to enhance decision-making accuracy in uncertain environments. DeLLMa involves a multi-step scaffolding procedure, drawing upon principles from decision theory and utility theory, to provide an optimal and human-auditable decision-making process. We validate our framework on decision-making environments involving real agriculture and finance data. Our results show that DeLLMa can significantly improve LLM decision-making performance, achieving up to a 40% increase in accuracy over competing methods.
Biography: Ollie Liu is second-year Ph.D student in Computer Science at University of Southern California, co-advised by Prof. Dani Yogatama and Prof. Willie Neiswanger. In life, I usually go by Oliver. My current research interests lie in (multimodal) foundation models, especially their algorithmic reasoning capabilities and applications in sciences.
Host: Jonathan May and Justin Cho
More Info: https://www.isi.edu/research-groups-nlg/nlg-seminars/
Webcast: https://www.youtube.com/watch?v=XSTIFr9J0koLocation: Information Science Institute (ISI) - Conf Rm#689
WebCast Link: https://www.youtube.com/watch?v=XSTIFr9J0ko
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/research-groups-nlg/nlg-seminars/
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 in Structural Biology: Roots of Success
Fri, Apr 19, 2024 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Dr. Helen Berman, USC
Talk Title: AI in Structural Biology: Roots of Success
Abstract: The successful prediction of protein structures competitive with experimentally determined models was announced in 2020. The achievement by by AlphaFold 2 using AI methods was the culmination of years of effort to solve the protein folding problem. In this talk I will describe the history of this effort and the unique characteristics of the training data that were derived from the Protein Data Bank. I will also discuss how this breakthrough will open the doors to new research in structural and cell biology.
Zoom meeting ID: 944 0958 4905Passcode: 822247
It will be posted on our USC/ISI YouTube page within 1-2 business days: https://www.youtube.com/user/USCISI.
Biography: Dr. Berman received her AB from Barnard College in 1964. She trained in crystallography with George Jeffrey at the University of Pittsburgh where she received her Ph.D. in 1967. After her postdoctoral training, she went to the Institute for Cancer Research, Fox Chase Cancer Center where she rose through the ranks from Research Associate to Senior Member. In 1989, she joined the faculty of Rutgers University and is currently a Board of Governors Distinguished Professor Emerita of Chemistry and Chemical Biology. She is also an Research Professor at the University of Southern California. Her research has focused on nucleic acids, protein-nucleic acid interactions, and collagen. She has published more than 300 scholarly articles.
Helen was a co-founder of the Protein Data Bank (PDB) archive that was launched in 1971 and has been committed to the continued development and maintenance of this community resource. Major accomplishments on this journey include taking leadership roles in establishing the Nucleic Acid Database, the Research Collaboratory for Structural Bioinformatics Protein Data Bank, the worldwide Protein Data Bank, the Structural Biology Knowledgebase, and the Unified Data Resource for 3D Electron Microscopy.
She is also working on ways to use film and digital arts to communicate to a broader audience about the importance of structural biology in medicine and health. She was the Executive Producer of Target Zero – a documentary about HIV prevention in which high-quality molecular animations illustrate how the anti-HIV drugs work. As part of ongoing collaborations between the USC Bridge Institute and USC School of Cinematic Arts, she is working on the World in a Cell VR experience that provides a view of the inside of a pancreatic beta cell.
Helen is a member of the National Academy of Science, the American Academy for Arts and Sciences and a Fellow of the Biophysical Society, the American Association for the Advancement of Science, the American Crystallographic Association, and the International Society for Computational Biology. She is the recipient of several awards including the Benjamin Franklin Award for Open Access in the Life Sciences, the DeLano Award for Computational Biosciences, the ACA Martin Buerger and David Rognlie Awards, the Distinguished Service Award from the Biophysical Society, and the Carl Brändén Award from the Protein Society.
Visit links below to subscribe and for details on upcoming seminars:
https://www.isi.edu/isi-seminar-series/
https://www.isi.edu/events/
Host: Hosted by: Fred Morstatter POC: Justina Gilleland
More Info: https://www.isi.edu/events/4648/ai-in-structural-biology-roots-of-success/
Webcast: https://usc.zoom.us/j/94409584905?pwd=Sm5LVkd0bndUdEluM3piK0NWTUQrUT09Location: Information Science Institute (ISI) - CR#1014
WebCast Link: https://usc.zoom.us/j/94409584905?pwd=Sm5LVkd0bndUdEluM3piK0NWTUQrUT09
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/events/4648/ai-in-structural-biology-roots-of-success/
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-How to Steal ChatGPTs Embedding Size, and Other Low-rank Logit Tricks
Thu, Apr 25, 2024 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Matt Finlayson, USC
Talk Title: How to Steal ChatGPTs Embedding Size, and Other Low-rank Logit Tricks
Series: NL Seminar
Abstract: The commercialization of large language models (LLMs) has led to the common practice of restricting access to proprietary models via a limited API. In this work we show that, with only a conservative assumption about the model architecture, it is possible to learn a surprisingly large amount of non-public information about an API-protected LLM from a relatively small number of API queries (e.g., costing under $1000 USD for OpenAI’s gpt-3.5-turbo). Our findings are centered on one key observation: most modern LLMs suffer from a softmax bottleneck, which restricts the model outputs to a linear subspace of the full output space. We exploit this fact to unlock several capabilities, including (but not limited to) obtaining cheap full-vocabulary outputs, auditing for specific types of model updates, identifying the source LLM given a single full LLM output, and even efficiently discovering the LLM’s hidden size. Our empirical investigations show the effectiveness of our methods, which allow us to estimate the embedding size of OpenAI’s gpt-3.5-turbo to be about 4096. Lastly, we discuss ways that LLM providers can guard against these attacks, as well as how these capabilities can be viewed as a feature (rather than a bug) by allowing for greater transparency and accountability. *Meeting hosts only admit on-line guests that they know to the Zoom meeting. Hence, you’re highly encouraged to use your USC account to sign into Zoom. If you’re an outside visitor, please inform us at (nlg-seminar-host(at)isi.edu) to make us aware of your attendance so we can admit you. Specify if you will attend remotely or in person at least one business day prior to the event Provide your: full name, job title and professional affiliation and arrive at least 10 minutes before the seminar begins. If you do not have access to the 6th Floor for in-person attendance, please check in at the 10th floor main reception desk to register as a visitor and someone will escort you to the conference room location.
Biography: Matthew Finlayson is a PhD student studying NLP at the University of Southern California. Previously he was a predoctoral researcher at the Allen Institute for AI (AI2) after completing his bachelors degree in computer science and linguistics at Harvard University. Matthew is interested in the practical consequences of the architectural design of language models, from security to generation, as well as understanding how language models learn and generalize from data.
Host: Jon May and Justin Cho
More Info: https://www.isi.edu/research-groups-nlg/nlg-seminars/
Webcast: https://www.youtube.com/watch?v=3U9nA-l2YAsLocation: Information Science Institute (ISI) - Conf Rm#689
WebCast Link: https://www.youtube.com/watch?v=3U9nA-l2YAs
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/research-groups-nlg/nlg-seminars/
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: Staring into the Abyss and Eating Glass
Fri, Apr 26, 2024 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Rajiv Maheswaran, Former Second Spectrum CEO, USC Faculty
Talk Title: Staring into the Abyss and Eating Glass
Abstract: Rajiv spent a decade as faculty at USC and then a decade building a startup. He will briefly outline the path that took him from one to the other and answer questions about the adventures and misadventures and the similarities and differences in the journeys.
Biography: Rajiv Maheswaran was the CEO and Co-Founder of Second Spectrum, a startup that used computer vision and machine learning to automate next-generation data collection and real-time video augmentation in sports, creating software for teams, leagues and media. Second Spectrum clients have spanned the top leagues in sports including the NBA (where all teams are clients), NFL, English Premier League and top media partners such as Amazon (winning an Emmy for NFL Prime Vision), CBS (winning an Emmy for RomoVision), ESPN (NBA Marvel Arena of Heroes), and NBC (Premier League DataCast). Prior to Second Spectrum, Rajiv was a Research Assistant Professor at the University of Southern California’s Department of Computer Science and Project Leader at the Information Sciences Institute where he and Second Spectrum COO / Co-Founder Yu-Han Chang co-directed the Computational Behavior Group. Rajiv has written over 100 publications in artificial intelligence, control theory, data visualization, decision theory and game theory. Rajiv and Yu-Han received the USC Viterbi School of Engineering Use-Inspired Research Award, and they are the only two-time winners of the Alpha Award for Best Research Paper at the MIT Sloan Sports Analytics Conference. Rajiv received a B.S. degree in Applied Mathematics, Engineering and Physics from the University of Wisconsin-Madison and received M.S. and Ph.D. degrees in Electrical and Computer Engineering from the University of Illinois at Urbana-Champaign.
Host: Craig Knoblock
More Info: https://www.isi.edu/events/4889/ai-seminar-staring-into-the-abyss-and-eating-glass/
Webcast: https://usc.zoom.us/j/99736587279?pwd=TnZXTzdYOWFNZlJmQ0dYYlF2QURDZz09Location: Information Science Institute (ISI) - 1135/1137
WebCast Link: https://usc.zoom.us/j/99736587279?pwd=TnZXTzdYOWFNZlJmQ0dYYlF2QURDZz09
Audiences: Everyone Is Invited
Contact: Tricia Olmedo
Event Link: https://www.isi.edu/events/4889/ai-seminar-staring-into-the-abyss-and-eating-glass/
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.
