BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:CS Colloquium Series - Student/Postdoc Session
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260817T052431Z
UID:tag:localist.com\,2008:EventInstance_51930565282027
DTSTART:20260205T193000Z
DTEND:20260205T202000Z
DESCRIPTION:First Talk\n\nTItle: Adaptive Resource Allocation for Improving
  HIV Testing Processes\n\nSpeaker: Davin Choo\, Postdoctoral Fellow at Har
 vard SEAS\n\nAbstract: HIV testing programs face severe resource constrain
 ts while operating under uncertainty and evolving information. In this tal
 k\, I present some recent work on adaptive resource allocation for improvi
 ng HIV testing processes\, framed as sequential decision-making problems w
 here actions reveal new information and shape future opportunities. I will
  introduce algorithmic models that capture frontier-based testing\, partia
 l observability\, and multi-round stochastic arrivals. I then show how too
 ls such as Gittins indices\, branching bandits\, and diffusion models can 
 be used to design principled adaptive policies. Beyond HIV testing\, these
  methods offer broader insights into AI-driven approaches for resource-con
 strained decision-making beyond public health.\n\nSpeaker Bio: Davin is a 
 postdoctoral fellow at Teamcore\, Harvard University. He earned his PhD in
  Computer Science from the National University of Singapore (NUS) as an AI
 SG PhD fellow\, a Master's degree in Computer Science from ETH Zürich\, a
 nd two undergraduate degrees in Computer Science and Applied Mathematics f
 rom NUS. Between his undergraduate and Masters\, he also worked for a whil
 e as an applied research scientist at DSO National Laboratories on project
 s that lie in the intersection of AI and security. During his PhD at NUS\,
  he focused on the foundations of AI and machine learning\, working on sta
 tistical models\, causal inference\, and the design of resource-efficient 
 algorithms. His current postdoctoral research at Harvard explores how prin
 cipled algorithmic and AI techniques can be applied to real-world problems
  with the goal of achieving meaningful social impact.\n\n \n\nSecond Talk\
 n\n \n\n \n\nTitle: Perception as Generation: Navigating Ambiguity with Di
 ffusion Models\n\nSpeaker: Xinran (Nicole) Han \, Ph.D Candidate at Harvar
 d SEAS\n\nAbstract: Recovering 3D structure from 2D images is a central pr
 oblem in computer vision\, yet it is fundamentally ambiguous: many differe
 nt 3D worlds can give rise to the same image. In this talk\, I argue that 
 instead of seeking a single “best” estimate\, vision systems should be
  generative and model the distribution of plausible interpretations\, akin
  to how humans respond to visual illusions that induce multiple distinct i
 nterpretations. I demonstrate how this behavior emerges from training a pa
 tch-based diffusion model on everyday objects. I further show that small m
 otions\, together with architectural inductive biases that encourage botto
 m-up and top-down integration\, enable joint reasoning over shape and mate
 rial. More broadly\, this line of work points toward a deeper understandin
 g of human perception and suggests new directions for building more robust
  embodied systems.\n\n \n\nSpeaker Bio: Xinran (Nicole) Han is a PhD stude
 nt at Harvard University\, working with Prof. Todd Zickler. Previously\, s
 he graduated from the University of Pennsylvania\, advised by Prof. Jianbo
  Shi. Her research interests span computer vision and human perception\, w
 ith an emphasis on 3D understanding. Her work focuses on combining physics
 -based insights and learning-based neural priors to build data- and comput
 e-efficient models that generalize to unseen scenarios.\n\n \n\nThere will
  be pretzels and coffee before the talk at 2:15pm outside of LL2.224
GEO:42.363197;-71.127278
LOCATION:Science and Engineering Complex (SEC)\, LL2.224
SUMMARY:CS Colloquium Series - Student/Postdoc Session
URL;VALUE=URI:https://events.seas.harvard.edu/event/cs-colloquium-series-st
 udentpostdoc-session
CATEGORIES:Colloquia / Seminar / Lecture
END:VEVENT
END:VCALENDAR