SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for April
-
AI Seminar-Experiments in Scaling Reinforcement Learning with Verifiable Rewards
Fri, Apr 04, 2025 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Nathan Lambert, Allen Institure
Talk Title: Experiments in Scaling Reinforcement Learning with Verifiable Rewards
Series: AI Seminar
Abstract: With the release of DeepSeek’s R1 reasoning model, interest in reinforcement learning may be at an all time high. Academics are pouring energy into the space, trying to replicate DeepSeek’s results and establish clear trade-offs and capabilities of this new era of reinforcement learning on language models. This talk discusses these new results with language models trained with Reinforcement Learning with Verifiable Rewards (RLVR), our efforts at scaling them for Ai2’s OLMo and Tülu language models, hints that we may have missed indicating that RL is more effective than people give credit for, and some history from my background in model-based RL/robotics. The goal of the talk is to present a mix of (recent) historical context on language modeling and cutting edge research with RL to forecast how the rapidly expanding industry of language models may change in the near future.
Biography: Nathan Lambert is a Senior Research Scientist and post-training lead at the Allen Institute for AI focusing on building open language models. At the same time he founded and operates Interconnects.ai to increase transparency and understanding of current AI models and systems.
Previously, he helped build an RLHF research team at HuggingFace. He received his PhD from the University of California, Berkeley working at the intersection of machine learning and robotics. He was advised by Professor Kristofer Pister in the Berkeley Autonomous Microsystems Lab and Roberto Calandra at Meta AI Research. He was lucky to intern at Facebook AI and DeepMind during his Ph.D. Nathan was was awarded the UC Berkeley EECS Demetri Angelakos Memorial Achievement Award for Altruism for his efforts to better community norms.
If speaker approves to be recorded for this seminar it will be posted on the USC/ISI YouTube page within 1-2 business days: https://www.youtube.com/user/USCISI.
Subscribe here to learn more about upcoming seminars: https://www.isi.edu/events/ .
Host: Eric Boxer and Justina Gilleland
More Info: https://www.isi.edu/events/5553/experiments-in-scaling-reinforcement-learning-with-verifiable-rewards/
Webcast: https://usc.zoom.us/j/94409584905?pwd=Sm5LVkd0bndUdEluM3piK0NWTUQrUT09Location: Information Science Institute (ISI) - Virtual Only
WebCast Link: https://usc.zoom.us/j/94409584905?pwd=Sm5LVkd0bndUdEluM3piK0NWTUQrUT09
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/events/5553/experiments-in-scaling-reinforcement-learning-with-verifiable-rewards/
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. -
CA DREAMS - Technical Seminar Series
.jpg)
Fri, Apr 04, 2025 @ 12:00 PM - 01:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Shaya Fainman, Professor, University of California, San Diego
Talk Title: Foundry Enabled Chip-scale Photonics Technology and Applications
Abstract: Dense photonic integration requires miniaturization of materials, devices, circuits and systems, including passive components (e.g., engineered composite metamaterials, filters, etc.), active components (e.g., modulators and nonlinear wave mixers) and integrated circuits (Fourier transform spectrometer, programmable phase modulator of free space modes, linear algebra processors, etc.). In this talk we will discuss recent progress in developing CMOS compatible nonlinear optical materials as well as examples of foundry enabled silicon photonic circuits and systems. Specifically, we will review silicon photonics-based Fourier transform spectrometer (Si-FTS) that can bring broadband operation and fine resolution to the chip scale. Here we will present the modeling and experimental demonstration of a thermally tuned Si-FTS accounting for dispersion, thermo-optic non-linearity, and thermal expansion. We show how these effects modify the relation between the spectrum and interferogram of a light source and we develop a quantitative correction procedure through calibration with a tunable laser. Providing design flexibility and robustness, the Si-FTS is poised to become a fundamental building block for on-chip spectroscopy. Moreover, taking advantage of nanofabrication we will discuss on-chip spectrometers using stratified waveguide filters and machine learning. Moving forward, we will discuss chip-scale integrated circuit/system that will allow to realize linear algebra accelerators with superior performance in speed, energy consumption and size compared to its electronic counterpart. Such system can be manufactured using monolithic CMOS process and impact such applications as 5G/6G and beyond wireless MIMO systems as well as deep learning and artificial intelligence.
Biography: Yeshaiahu (Shaya) Fainman is an inaugural ASML/Cymer Chair of Advanced Optical Technologies and Distinguished Professor in Electrical and Computer Engineering (ECE) at the University of California, San Diego (UCSD). He received the M. Sc and Ph. D degrees from Technion-Israel Institute of Technology in 1979 and 1983, respectively. He is directing research of the Ultrafast and Nanoscale Optics group at UCSD and made significant contributions to near field optical phenomena, nanoscale science and engineering of ultra-small, sub-micrometer semiconductor light emitters and nanolasers, inhomogeneous and meta-materials, nanophotonics and plasmonics, non-conventional imaging and silicon photonics. His current research interests are in near field optical science and optical technology with applications targeting information technologies and biomedical sensing. He contributed over 360 manuscripts in peer review journals and over 560 conference presentations and conference proceedings. During his career he has led as Director and Deputy Director of numerous large size interdisciplinary projects and centers supported by BMDO, DARPA, NSF-ERC, and ONR. He is a Fellow of the OSA, IEEE, SPIE, and a recipient of the Miriam and Aharon Gutvirt Prize, Lady Davis Fellowship, Brown Award, SPIE Gabor Award, OSA Emmett N. Leith Medal, OSA Joseph Fraunhofer Award/Robert M. Burley Prize and OPTICA (former OSA) Nick Holonyak Jr Award.
Host: Dr. Steve Crago
More Info: https://www.isi.edu/events/5445/foundry-enabled-chip-scale-photonics-technology-and-applications/
Webcast: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addonWebCast Link: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addon
Audiences: Everyone Is Invited
Contact: Amy Kasmir
Event Link: https://www.isi.edu/events/5445/foundry-enabled-chip-scale-photonics-technology-and-applications/
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- An Investigation of Intermediate Representations in Spoken Language Models
Thu, Apr 10, 2025 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Tolulope Ogunremi, Stanford University
Talk Title: An Investigation of Intermediate Representations in Spoken Language Models
Abstract: 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. https://usc.zoom.us/j/93979709729?pwd=v8abin7zGE0E7jWy4cGoEj8vyyFlUT.1 Meeting ID: 939 7970 9729 Passcode: 804448
Spoken language models, large language models trained to process speech and audio inputs by leveraging speech encoder representations, have rapidly increased in popularity as a new modelling approach to speech processing tasks. These models train modality adapters to adapt speech encoder output into language model input.In this work, we use CommonVoice and FLEURS automatic speech recognition (ASR) data in several languages to investigate the output of the modality adapter of spoken language models. We introduce an algorithm to determine whether the modality adapter output resembles a transcription, transliteration or a semantic representation of the speech. We also find that the representation of a language in the language model affects the modality adapter output and transcription abilities of the spoken language models.
Biography: Tolúlá»ÂpẹàÒgúnrẹÃÂmí is a Computer Science PhD candidate at Stanford University in the Stanford NLP Group. Her work focusses on speech and language processing for low-resource languages, currently African languages. Her research combines linguistic investigations of these languages and community-based projects that integrate the concerns of local language communities with technological advances. Before, she did a Masters in Speech and Language Processing at the University Edinburgh.
If speaker approves to be recorded for this seminar, it will be posted on the USC/ISI YouTube page within 1-2 business days: https://www.youtube.com/user/USCISI Subscribe here to learn more about upcoming seminars: https://www.isi.edu/events/ For more information on the NL Seminar series and upcoming talks, please visit: https://www.isi.edu/research-groups-nlg/nlg-seminars/ Hosts: Jonathan May and Katy Felkner
Host: Jonathan May and Katy Felkner
More Info: https://www.isi.edu/research-groups-nlg/nlg-seminars/
Webcast: https://usc.zoom.us/j/93979709729?pwd=v8abin7zGE0E7jWy4cGoEj8vyyFlUT.1Location: Information Science Institute (ISI) - Conf Rm#689
WebCast Link: https://usc.zoom.us/j/93979709729?pwd=v8abin7zGE0E7jWy4cGoEj8vyyFlUT.1
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- How I learned to stop worrying and love AI
Fri, Apr 11, 2025 @ 11:30 AM - 12:30 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: William Regli, University of Maryland
Talk Title: How I learned to stop worrying and love AI
Series: AI Seminar
Abstract: In Voltaire’s Candide, Dr. Pangloss is relentlessly optimistic in the face of novella’s unflinching portrait of the human condition; his opposite, Martin, is pessimistic and cynical. Today’s developments around Artificial Intelligence are being driven by similarly opposing forces. The Panglossian approach views AI as humanity’s grasping of Promethean fire whereas others see existential risk and threats to human safety, privacy, and wellbeing. We might hope that the reality is somewhere in between; and we might suspect that the reason for these extreme views is that we probably have the problems around AI framed incorrectly.This presentation attempts to summarize my personal views regarding AI that I have developed during my decade away from academia in various forms of public service. First, as a member of the Defense Advanced Research Projects Agency’s (DARPA) leadership team in the Defense Sciences Office (2014-2017); next as the founding director of the Applied Research Laboratory for Intelligence and Security, the Department of Defense’s university-affiliated research center (UARC) for the social sciences and AI at the University of Maryland (2018-2023); and lastly as Senior Advisor for AI Risk Modeling for Biden Administration’s Office of Science and Technology Policy (2023-2024). The bottom line, upfront:— Current AI narratives are techno-philic and need to be re-framed because the thorniest problems are decidedly non-technical—they are mostly about AI’s interaction with, and influence on, people and society;— Unlike physics and engineering we do not yet have the required level of scientific understanding about AI and its effects on people and society needed to establish rigorous engineering practices and manage its use; and, lastly— The impacts of AI, operating at various levels in our society (ranging from individuals to our planetary community as a whole), are going to be uneven in scale, speed, and impact. I would rather not merely admire these problems, hence I will try to re-frame them as inherently socio-technical. I will provide a practical methodology for identifying emerging scientific and engineering questions related to the ongoing integration of AI with humans and society. Using this approach, I will provide several examples of research questions that merit investigation. In the end, I hope to provide a unique perspective on recent developments in AI and a tangible means by which we might address these daunting emerging challenges.
Biography: Dr. Regli is a computer scientist who has focused his career on interdisciplinary and use-inspired problems spanning artificial intelligence, engineering and manufacturing, and computational modeling. Dr. Regli’s recent sponsored research activities include verification and validation of intelligent systems; intelligent computer networks; and the use of artificial intelligence in advanced manufacturing. He has published more than 250 technical articles, created two technology companies (one focused on mobile communications for public safety, the other on information management in edge networks), and produced five foundational U.S. Patents in the area of 3D CAD search.From 2014 to 2017 Regli served on the leadership team of the Defense Advanced Research Projects Agency (DARPA), as Deputy Director (9/14-12/16) and Acting Director (1/17-7/17) of the Defense Sciences Office (DSO); then as Special Assistant to the DARPA Director (8/17-12/17). During his tenure, DSO initiated programs in areas as diverse as artificial intelligence, design and manufacturing, social science, applied mathematics, physical sciences and advanced sensing technologies; in his role leading DSO he advanced the data management and retention plans for the agency, co-developed the “Disruptioneering” program template, expanded the DARPA Young Faculty Award, and was the Program Chair for the “DARPA 60” anniversary conference (9/2018). For his contributions, Regli received the Award for Excellence for Meritorious Service (2015) from the Undersecretary of Defense (AT&L) and DARPA Meritorious Public Service Medal (2017). Regli’s other government service includes as a National Research Council Postdoctoral Fellow at the National Institute of Standards and Technology (NIST) (1995-1997); as a Scientific Adviser to the U.S. Department of Energy (DoE) in the areas of information technology and manufacturing (2010-2014;2018-); and as a member of the US Air Force Scientific Advisory Board (2019-2021; 2022-). His community service currently includes a role on the Computing Research Association (CRA) Computing Community Consortium (CCC) Advisory Board (2021-) as well as several editorial boards. Regli recently completed service as the founding Executive Director (2018-2023) of the University of Maryland’s University-Affiliated Research Center for the Department of Defense: The Applied Research Laboratory for Intelligence and Security (ARLIS), the chartered DoD academic laboratory for the Intelligence and Security communities. For part of 2023-2024, Regli served in the Executive Offices of the President (EOP), White House Office of Science and Technology Policy (OSTP), as a Senior Advisor for AI Risk for the National AI Initiatives Office supporting a variety of activities.Dr. Regli holds a Ph.D. in Computer Science from the University of Maryland at College Park and Bachelor of Science degree in Mathematics from Saint Joseph's University. He is an elected Senior Member of both the Association of Computing Machinery (ACM) and of the Association for the Advancement of Artificial Intelligence (AAAI); and a Fellow of the Computer Society of the Institute of Electrical and Electronics Engineers (IEEE) for his “contributions to 3D search, design repositories and intelligent manufacturing”, and a Fellow of the American Association for the Advancement of Science (AAAS) for “work at the interface between science and government primarily at the Defense Advanced Research Projects Agency.”
Host: Abel Salinas and Maura Covaci
More Info: https://www.isi.edu/events/5723/how-i-learned-to-stop-worrying-and-love-ai/
Webcast: https://usc.zoom.us/j/98948507679?pwd=3j3zstL7xeFhfwELPJaJ8zHEbXBz4M.1Location: Information Science Institute (ISI) - Virtual Only
WebCast Link: https://usc.zoom.us/j/98948507679?pwd=3j3zstL7xeFhfwELPJaJ8zHEbXBz4M.1
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/events/5723/how-i-learned-to-stop-worrying-and-love-ai/
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. -
CA DREAMS - Technical Seminar Series
.jpg)
Fri, Apr 11, 2025 @ 12:00 PM - 01:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Jonathan Klamkin, Professor, UC Santa Barbara and Director of UCSB Nanofab
Talk Title: Integrated Photonics: It's Always About High Performance
Series: CA DREAMS - Technical Seminar Series
Abstract: Photonics integrated circuits (PICs) are now widely deployed for optical communications. And more recently, the demand for higher speed, lower latency, and lower power consumption interconnects has increased significantly to support AI infrastructure. These optical interconnects are also a viable approach to move data from array-based systems such as focal plane arrays and RF phased arrays. This presentation will describe highly complex PICs for RF photonics, optical interconnects, and sensors, as well as heterogeneous integration methods to bridge high-performance active components with silicon photonics. While the community tends to think of silicon as the means to low cost, integrated photonics has always been about high performance.
Biography: Jonathan Klamkin received the B.S. degree from Cornell University, and the M.S. and Ph.D. degrees from the University of California Santa Barbara (UCSB). From 2008-2011 he was a member of the Technical Staff at MIT Lincoln Laboratory. From 2011-2013 he was an Assistant Professor at the Institute of Communication, Information and Perception Technologies (TeCIP), Scuola Superiore Sant’Anna, Pisa, Italy. From 2013-2015 he was an Assistant Professor of Electrical and Computer Engineering (ECE) and Materials at Boston University. In 2015 Professor Klamkin joined the ECE Department at UCSB where he leads the Integrated Photonics Laboratory (iPL) and serves as Director of the UCSB Nanofab. He has published 250 journal and conference papers, more than 30 issued and pending patents, and has delivered more than 120 invited, keynote and plenary presentations. Professor Klamkin is the recipient of the NASA Early Career Faculty Award, the DARPA Young Faculty Award, the DARPA Director's Fellowship, and the PIERS Young Scientist Award. He is a Fellow of Optica.
Host: Dr. Steve Crago
More Info: https://www.isi.edu/events/5730/integrated-photonics-its-always-about-high-performance/
Webcast: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addonWebCast Link: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addon
Audiences: Everyone Is Invited
Contact: Amy Kasmir
Event Link: https://www.isi.edu/events/5730/integrated-photonics-its-always-about-high-performance/
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-Ushering Agents to an Open Social World
Thu, Apr 17, 2025 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Hao Zhu, Stanford University
Talk Title: Ushering Agents to an Open Social World
Abstract: 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 JOIN VIA ZOOM: https://usc.zoom.us/j/98699643447?pwd=59bYaPQunEwvO3kiZM8jel8s2efWnu.1 Meeting ID: 986 9964 3447 Passcode: 804448 Unlike frontier AI models trained on static datasets, humans learn through dynamic interactions with other people and the world. This fundamental difference in learning methodology not only makes language agents less sample-efficient than humans but also introduces significant risks when these agents are deployed to interact with real humans in the real world. Building agents that can efficiently learn through interaction with other agents, humans and the world is a challenging problem. In this presentation, I will outline three foundational approaches we've developed to address this challenge: (1) Learning through exploration on the internet (NNetNav-live) — We deploy an open-ended agent (without explicit task instructions) to explore the web, gather experience and retroactively label and train on the data. (2) Learning from human normative decision-making (EgoNormia) — We explore methods for agents to observe and internalize social norms in physical interactions through crowd-sourced annotation with context perturbation. (3) Learning to build metrics from human feedback (AutoLibra, in prep) — We present a framework for automatically building behavior evaluation metric systems that help both humans understand agent performance, and agents improve the policy based on human feedback. These complementary approaches offer a path toward creating AI agents that can more effectively learn, adapt, and integrate into our open social world." Hao Zhu is a postdoctoral researcher in the Computer Science Department at Stanford University. He finished his PhD from CMU. He is interested in AI agents, human-agent interaction, robotics and embodied AI, and what AI agents tell us about human social and embodied cognition.
Biography: Hao Zhu is a postdoctoral researcher in the Computer Science Department at Stanford University. He finished his PhD from CMU. He is interested in AI agents, human-agent interaction, robotics and embodied AI, and what AI agents tell us about human social and embodied cognition.
Host: Jonathan May and Katy Felkner
More Info: https://www.isi.edu/events/5715/ushering-agents-to-an-open-social-world/
Webcast: https://www.youtube.com/watch?v=6jdaj0yOHGs&list=PLknXvJJeEDaK5yBaaGfBhwCEKd7be-d19Location: Information Science Institute (ISI) - Conf Rm#689
WebCast Link: https://www.youtube.com/watch?v=6jdaj0yOHGs&list=PLknXvJJeEDaK5yBaaGfBhwCEKd7be-d19
Audiences: Everyone Is Invited
Contact: Pete Zamar
Event Link: https://www.isi.edu/events/5715/ushering-agents-to-an-open-social-world/
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. -
CA DREAMS - Technical Seminar Series
.jpg)
Fri, Apr 18, 2025 @ 12:00 PM - 01:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Hamidreza Aghasi, Professor, UC Irvine
Talk Title: CMOS Circuits and Systems for Coherent Multi-Band Millimeter-Wave Radars
Series: CA DREAMS - Technical Seminar Series
Abstract: This presentation discusses recent advancements in CMOS-based circuit and system design for coherent multi-band millimeter-wave (mm-wave) FMCW radars, enabling enhanced range and angular resolution across multiple frequency bands. We begin by presenting block-level strategies for achieving broadband radiation coverage and phase noise reduction in mm-wave radar systems. Next, we introduce a 49–63â¯GHz dual-PLL stepped-chirp radar transceiver in 22â¯nm FD-SOI, which employs phase-locked sub-chirps to achieve a 14â¯GHz effective bandwidth and 1.4â¯cm range resolution, while addressing challenges related to chirp linearity and phase/frequency synchronization. We then present a dual-band 23–27â¯GHz and 69–81â¯GHz MIMO radar in 65â¯nm CMOS featuring synchronized LO generation and frequency tripling, enabling a 16â¯GHz total bandwidth and sub-centimeter range resolution. The presentation concludes with recent measurement results and highlights from ongoing designs targeting further improvements in range and angular resolution for next-generation mm-wave radar systems.
Biography: Hamidreza Aghasi received his B.Sc. in Electrical Engineering from Sharif University of Technology, Tehran, Iran, in 2011, and his M.S. and Ph.D. in Electrical Engineering from Cornell University in 2015 and 2017, respectively. In the summer of 2014, he was an intern at Samsung Research America’s Display Lab in San Jose, California. From 2017 to 2018, he was a postdoctoral research fellow at the University of Michigan, and from 2018 to 2019, he was a mm-Wave research scientist at Acacia Communications Inc. in Holmdel, New Jersey. He is currently an Assistant Professor in the Electrical Engineering and Computer Science Department at the University of California, Irvine (UC Irvine), where he is Director of the High-Speed Integrated Electronics (HIE) Laboratory. His research interests include RF, mm-wave, and terahertz circuit design for imaging, sensing, and communication applications. Dr. Aghasi is a senior member of IEEE, member of Solid-State Circuits Society (SSCS) and the Microwave Theory and Techniques Society (MTT-S), a TPC member of the IEEE Custom Integrated Circuit Conference (CICC) and the IEEE Radio Frequency Integrated Circuits Symposium (RFIC), and an associate editor of IEEE Transactions on Very Large Scale Integration (TVLSI). He also serves as the IEEE MTT-S/EDS Chapter Chair of Orange County, CA. He has reviewed for the IEEE Journal of Solid-State Circuits, IEEE Open Journal of Solid-State Circuits, IEEE Solid-State Circuits Letters, IEEE Transactions on Microwave Theory and Techniques, IEEE Transactions on Very Large Scale Integration (VLSI) Systems, IEEE Electron Device Letters, IEEE Transactions on Circuits and Systems, and IEEE Microwave and Wireless Components Letters. He received the Cornell Graduate Fellowship in 2011, the Jacobs Fellowship in 2012, the Cornell ECE Innovation Award in 2013, the Cornell Scale-up and Prototyping Award in 2017, Best Invited Paper Award at the 2019 IEEE CICC, the NeurIPS ML4PS Reproducibility Award in 2024, and the NSF CAREER Award in 2025.
Host: Dr. Steve Crago
More Info: https://www.isi.edu/events/5737/cmos-circuits-and-systems-for-coherent-multi-band-millimeter-wave-radars/
Webcast: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addonWebCast Link: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addon
Audiences: Everyone Is Invited
Contact: Amy Kasmir
Event Link: https://www.isi.edu/events/5737/cmos-circuits-and-systems-for-coherent-multi-band-millimeter-wave-radars/
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 Surprising Effectiveness of Membership Inference with Simple N-Gram Coverage
Thu, Apr 24, 2025 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Skyler Hallinan, USC
Talk Title: The Surprising Effectiveness of Membership Inference with Simple N-Gram Coverage
Series: NL Seminar
Abstract: 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. Join Via Zoom: https://usc.zoom.us/j/96791099940?pwd=6kov3zTLAnD4JU49d1VtX4XNAZMcvs.1Meeting ID: 967 9109 9940Passcode: 840282
Membership inference attacks serves as useful tool for fair use of language models, such as detecting potential copyright infringement and auditing data leakage. However, many current state-of-the-art attacks require access to models' hidden states or probability distribution, which prevents investigation into more widely-used, API-access only models like GPT-4. In this work, we introduce N-Gram Coverage Attack, a membership inference attack that relies solely on text outputs from the target model, enabling attacks on completely black-box models. We leverage the observation that models are more likely to memorize and subsequently generate text patterns that were commonly observed in their training data. Specifically, to make a prediction on a candidate member, N-Gram Coverage Attack first obtains multiple model generations conditioned on a prefix of the candidate. It then uses n-gram overlap metrics to compute and aggregate the similarities of these outputs with the ground truth suffix; high similarities indicate likely membership. We first demonstrate on a diverse set of existing benchmarks that N-Gram Coverage Attack outperforms other black-box methods while also impressively achieving comparable or even better performance to state-of-the-art white-box attacks --- despite having access to only text outputs. Interestingly, we find that the success rate of our method scales with the attack compute budget --- as we increase the number of sequences generated from the target model conditioned on the prefix, attack performance tends to improve. Having verified the accuracy of our method, we use it to investigate previously unstudied closed OpenAI models on multiple domains. We find that more recent models, such as GPT-4o, exhibit increased robustness to membership inference, suggesting an evolving trend toward improved privacy protections.
Biography: Skyler Hallinan is a Ph.D. student in Computer Science at the University of Southern California where he is advised by Xiang Ren. His research aims to build trustworthy AI systems with robust reasoning capabilities via data-centric approaches. His work spans three core areas: understanding how data impacts downstream model behavior, safeguarding user data and privacy, and advancing model capabilities with better data. Previously, he was a research intern at Apple and Amazon, and received a B.S./M.S. in Computer Science from the University of Washington, where he was advised by Yejin Choi.
Host: Jonathan May and Katy Felkner
More Info: https://www.isi.edu/research-groups-nlg/nlg-seminars/
Webcast: https://usc.zoom.us/j/96791099940?pwd=6kov3zTLAnD4JU49d1VtX4XNAZMcvs.1Location: Information Science Institute (ISI) - Conf Rm#689
WebCast Link: https://usc.zoom.us/j/96791099940?pwd=6kov3zTLAnD4JU49d1VtX4XNAZMcvs.1
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- Texera: An Open-Source System for Cloud-Based Collaborative Data Science and AI/ML Using Workflows
Fri, Apr 25, 2025 @ 11:00 AM - 12:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Chen Li, UC Irvine
Talk Title: Texera: An Open-Source System for Cloud-Based Collaborative Data Science and AI/ML Using Workflows
Series: AI Seminar
Abstract: Since 2016 our team at UC Irvine has been developing the Texera open-source system (texera.io), with the goal of enabling a cloud-based platform to support collaborative data science, AI, and ML. It allows users with various backgrounds, including those with limited coding skills, domain scientists, and ML experts, to conduct AI-centric data science with a collaboration experience similar to Google Docs. After eight years of development, the system has a rich set of features, such as shared editing, shared execution, version control, commenting, debugging, user-defined functions in multiple languages (e.g., Python, R, Java), and support of state-of-the-art AI/ML techniques. Its backend parallel engine enables scalable computation on large data sets using computing clusters. It allows bioinformaticians to elastictally request resources from AWS to form a cluster to run computationally intensive jobs. It also supports community-based sharing of resources including datasets and workflows. In this talk, we will give an overview of the system, and focus on research challenges encountered in the development and our solutions. We will show use cases in both education and scientific communities.
Biography: Prof. Chen Li is a professor in the Department of Computer Science at UC Irvine. He received his Ph.D. degree in Computer Science from Stanford University, and his M.S. and B.S. in Computer Science from Tsinghua University, China. His research interests are in the fields of data management, data science, AI/ML, databases, data-intensive computing, search, and visualization. He was a co-founder and CTO of a startup to commercialize his research. He was a recipient of an NSF CAREER award, an ACM Distinguished Member, and an IEEE fellow. If speaker approves to be recorded for this seminar it will be posted on the USC/ISI YouTube page within 1-2 business days: https://www.youtube.com/user/USCISI. Subscribe here to learn more about upcoming seminars: https://www.isi.edu/events/ .
Host: Zhuoyu Shi and Pete Zamar + Maura Covaci
More Info: https://www.isi.edu/events/5507/texera-an-open-source-system-for-cloud-based-collaborative-data-science-and-ai-ml-using-workflows/
Webcast: https://usc.zoom.us/j/95166882238?pwd=id334Bxxz7ZULMFpYWuHEppmFKlfUd.1Location: Information Science Institute (ISI) - Virtual Only
WebCast Link: https://usc.zoom.us/j/95166882238?pwd=id334Bxxz7ZULMFpYWuHEppmFKlfUd.1
Audiences: Everyone Is Invited
Contact: Pete 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. -
CA DREAMS - Technical Seminar Series
.jpg)
Fri, Apr 25, 2025 @ 12:00 PM - 01:00 PM
Information Sciences Institute
Conferences, Lectures, & Seminars
Speaker: Brian Thibeault and Steve Zamek, UCSB and PDF Solutions
Talk Title: Analytics for Semiconductors: From Megafabs to Nanofabs
Series: CA DREAMS - Technical Seminar Series
Abstract: This joint talk will provide an overview of advanced data analytics for semiconductor manufacturing and application thereof in R&D environment. The first part of the talk, by Steve Zamek from PDF Solutions, will provide an overview of end-to-end data analytics used by 100+ semiconductor companies. Big Data analytics enables faster yield ramp, improved efficiency, lower manufacturing cost and facilitates root cause analysis in a timely manner. The second part of the talk, by Brian Thibeault, the UCSB Nanofab Director, will discuss the small-business enabling R&D Nanofab environment. The UCSB Nanofab, a leading and accessible university nanofabrication facility, has been at the center of a semiconductor startup ecosystem for over 20 years. Brian will introduce Nanofab’s operation in the context of this vibrant ecosystem that bridges university research and commercialization. Several examples of successful lab-to-fab transition stories will be provided. Brian will also cover the recent collaboration between UCSB and PDF Solutions to improve process reproducibility, leveraging the Exensio® software within a technology-diverse, multi-user fabrication environment.
Biography: Brian Thibeault, UCSB Brian Thibeault has been UCSB Nanofab Technical/Operational Director for 6 years, where he is responsible for all aspects of facility advancements and operations, and the senior Nanofab staff scientist for 24 years. From 1996 to 2000, Brian worked on GaN-based LED and RF HEMT development for WiTech, LLC, founded by Steve DenBaars and Umesh Mishra and purchased by CREE Inc. in June of 2000. Brian holds a PhD from UCSB in Electrical and Computer Engineering (1997), where his research focused on Vertical Cavity Surface Emitting Laser design, growth, and fabrication. Steve Zamek, PDF Solutions Steve Zamek is a Director of Product Management at PDF Solutions Inc. Steve is responsible for the Big Data Analytics platform deployed in 100+ leading customers – foundries, IDM’s, OSAT’s and fabless companies. Prior to his current role, Steve held a variety of roles at KLA, a leading provider of inspection and metrology equipment for the semiconductor industry. Steve holds a PhD from UCSD, MSc from BGU, BSc from the Technion – all in Electrical Engineering. He had Internships with Cymer (now ASML) and Sun (now Oracle).
Host: Dr. Steve Crago
More Info: https://www.isi.edu/events/5740/analytics-for-semiconductors-from-megafabs-to-nanofabs/
Webcast: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addonWebCast Link: https://usc.zoom.us/j/97017422125?pwd=Dbrt8MNMrmBV3xalKQJcAiNsggFJjJ.1&from=addon
Audiences: Everyone Is Invited
Contact: Amy Kasmir
Event Link: https://www.isi.edu/events/5740/analytics-for-semiconductors-from-megafabs-to-nanofabs/
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.
.jpg)
.jpg)
.jpg)
.jpg)