BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//AISI - ECPv6.6.2//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:AISI
X-ORIGINAL-URL:https://ai-science.uci.edu
X-WR-CALDESC:Events for AISI
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:UTC
BEGIN:STANDARD
TZOFFSETFROM:+0000
TZOFFSETTO:+0000
TZNAME:UTC
DTSTART:20250101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=UTC:20251208T130000
DTEND;TZID=UTC:20251208T140000
DTSTAMP:20260501T035717
CREATED:20251202T203527Z
LAST-MODIFIED:20251202T203527Z
UID:2970-1765198800-1765202400@ai-science.uci.edu
SUMMARY:Monte Carlo Tree Diffusion and Loopholing Diffusion
DESCRIPTION:Professor Sungjin Ahn\nKAIST/New York University\nMonday\, December 8\, 2025\n1:00 pm to 2:00 pm\n4013 Donald Bren Hall \n\nAbstract\nDiffusion models have rapidly advanced to become a central generative engine in modern AI. Yet two major challenges remain: enhancing test-time compute scalability and developing diffusion-based language models that can rival or ultimately replace autoregressive approaches. In this talk\, I present two recent advances addressing these directions. To enable scalable test-time reasoning and planning\, I introduce Monte Carlo Tree Diffusion\, a method that integrates diffusion models with Monte Carlo Tree Search for substantial performance gains under increased computation. To advance diffusion-based language modeling\, I introduce Loopholing Discrete Diffusion Models\, a new framework that overcomes key limitations of discrete diffusion and offers a promising path toward competitive alternatives to autoregressive language models. \n\nSpeaker Bio\nSungjin Ahn (CV) is currently an Associate Professor in the School of Computing and the Graduate School of AI at KAIST\, and also holds a joint appointment at New York University. Before joining KAIST\, he was an Assistant Professor of Computer Science at Rutgers University\, where he was affiliated with the Center for Cognitive Science. At KAIST\, he directs the Machine Learning and Mind Lab and the KAIST-Mila Prefrontal AI Research Center. He received his Ph.D. from the University of California\, Irvine under the supervision of Prof. Max Welling\, and subsequently completed a postdoctoral fellowship at MILA\, conducting deep learning research under the mentorship of Prof. Yoshua Bengio.
URL:https://ai-science.uci.edu/event/professor-sungjin-ahn/
LOCATION:On Campus\, Donald Bren Hall\, Room 4011\, Irvine\, CA\, 92697\, United States
CATEGORIES:Seminar
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20251120T143000
DTEND;TZID=UTC:20251120T153000
DTSTAMP:20260501T035717
CREATED:20251118T193453Z
LAST-MODIFIED:20251128T204114Z
UID:1951-1763649000-1763652600@ai-science.uci.edu
SUMMARY:From Models to Systems: The Next Frontier of Generative AI in Life and Molecular Sciences
DESCRIPTION:Jakub Tomczak\nChan Zuckerberg Initiative\nThursday\, November 20\, 2025\n2:30 pm to 3:30 pm\n4011 Donald Bren Hall\n \nCoffee and light refreshments will be served \nSpeaker Bio\nJakub Tomczak is a Generative AI leader with over 15 years of experience in machine learning\, deep learning\, and Generative AI. He has led extensive research across academia and industry\, contributing three patents\, numerous publications at top conferences (including NeurIPS\, ICML\, and CVPR)\, and securing EUR 2.3M in direct funding while contributing to consortia that have collectively obtained over EUR 100M. Jakub has managed teams for more than a decade and has served as a fractional AI leader for organizations such as eBay\, Qualcomm\, and multiple startups. He was Program Chair of NeurIPS 2024 and is the author of Deep Generative Modeling\, the first comprehensive textbook on Generative AI. He is also the founder of Amsterdam AI Solutions. \nAbstract\nGenerative Artificial Intelligence (GenAI) has revolutionized how we model complex phenomena\, yet its true potential in life and molecular sciences remains largely untapped. In this talk\, I will discuss the transition from model-centric to system-centric generative AI — moving beyond isolated deep models toward integrated\, interpretable\, and scientifically grounded generative systems. I will present key advances from my research spanning probabilistic and deep generative modeling\, including Variational Autoencoders with VampPrior and diffusion-based extensions\, attention-based and mixed models for multi-instance and multi-modal biological data\, and joint generative–predictive diffusion frameworks that unify representation learning and explainability. These developments enable principled modeling of genomic\, molecular\, and biomedical image data under uncertainty\, symmetry\, and multi-scale constraints. Finally\, I will introduce the concept of Generative AI Systems (GenAISys) — architectures that connect multiple foundation models with simulators and domain knowledge — and outline their role as the next frontier for agentic\, reliable AI in scientific discovery and healthcare innovation. \nHosted by\nAI in Science Institute\nCenter for Machine Learning and Intelligent Systems
URL:https://ai-science.uci.edu/event/ai-in-science-seminar-jakub-tomczak/
LOCATION:On Campus\, Donald Bren Hall\, Room 4011\, Irvine\, CA\, 92697\, United States
CATEGORIES:Seminar
END:VEVENT
END:VCALENDAR