IDS Guest Seminar: Leveraging Generative Models to Understand the Visual Cortex

IDS Guest Seminar - by Mr. Andrew LUO

Title: Leveraging Generative Models to Understand the Visual Cortex

Speaker: Andrew Luo, Ph.D. candidate, Carnegie Mellon University

Date: May 24, 2024
Time: 10:00am – 11:00am
Venue: HKU IDS Office, P307, Graduate House / Zoom

Mode: Hybrid. Seats for on-site participants are limited. A confirmation email will be sent to participants who have successfully registered. 

Abstract

Understanding the functional organization of the higher visual cortex is a fundamental goal in neuroscience. Traditional approaches have focused on mapping the visual and semantic selectivity of neural populations using hand-selected, non-naturalistic stimuli, which require a priori hypotheses about visual cortex selectivity. To address these limitations, we introduce two data-driven methods: Brain Diffusion for Visual Exploration (‘BrainDiVE’) and Semantic Captioning Using Brain Alignments (‘BrainSCUBA’). BrainDiVE synthesizes images predicted to activate specific brain regions, having been trained on a dataset of natural images and paired fMRI recordings, thus bypassing the need for hand-crafted visual stimuli. This approach leverages large-scale diffusion models combined with brain-gradient guided image synthesis. We demonstrate the synthesis of preferred images with high semantic specificity for category-selective regions of interest. BrainSCUBA, on the other hand, generates natural language descriptions for images predicted to maximally activate individual voxels. This approach enables efficient fine-grained labeling of the entire higher visual cortex. Together, these two methods offer well-specified constraints for future hypothesis-driven examinations and demonstrate the potential of data-driven approaches in uncovering brain organization.

Speaker

Mr. Andrew Luo
Ph.D. candidate @ Carnegie Mellon University

Andrew Luo is a Ph.D. candidate in the joint program for Neural Computation and Machine Learning at Carnegie Mellon University, co-advised by Profs Michael Tarr and Leila Wehbe. His research focuses on understanding the functional organization of higher visual cortex using learnable generative models across modalities. Before joining CMU, he earned a B.S. degree in Computer Science from MIT in 2019.

For information, please contact:
Email: datascience@hku.hk

IDS Guest Seminar: A Bottom-Up Approach towards Generalizable Robot Learning

IDS Guest Seminar - by Dr. Xingyu Lin

Title: A Bottom-Up Approach towards Generalizable Robot Learning

Speaker: Xingyu LIN, Postdoctoral Scholar, University of California, Berkeley

Date: May 22, 2024
Time: 10:00am – 11:00am
Venue: HKU IDS Office, P307, Graduate House / Zoom

Mode: Hybrid. Seats for on-site participants are limited. A confirmation email will be sent to participants who have successfully registered. 

Abstract

The rise of data-driven methods in robotics has significantly enhanced a robot’s capacity for perception, reasoning, and acting. However, the challenge and expense of collecting a diverse dataset with robots prevent learning control policies that are generalizable across various settings and tasks. Alternatively, while data sources like videos and robot play data are scalable, they are often not directly applicable due to the domain gaps and the absence of optimal action labels. In this talk, I will discuss my research on learning visual representations, particle trajectory models, and particle dynamics models from these data to learn generalizable low-level policies. These structured representations enables the learned policies to generalize to novel objects and configurations. I will conclude by demonstrating how these low-level skills can be assembled to tackle long-horizon and novel tasks.

Speaker

Dr. Xingyu Lin
Postdoctoral Scholar @ University of California, Berkeley

Xingyu Lin is a postdoctoral researcher at the University of California Berkeley, working with Pieter Abbeel. His research lies at the intersection of computer vision, machine learning and robotics, with a focus on learning robust manipulation skills that generalize to novel objects, tasks and deformable objects. Xingyu holds a PhD from the Robotics Institute at Carnegie Mellon University, advised by David Held. Prior to that, he received his undergraduate degree in computer science from Peking University. His research has been published at top conferences, including CoRL, RSS, NeurIPS and ICLR. He was also selected as an RSS (Robotics Science and System) 2022 Pioneer.

For information, please contact:
Email: datascience@hku.hk

IDS Distinguished Speaker Series #6: Towards Robust and Risk-aware Contextual Optimization

IDS Distinguished Speaker Series #6 - Professor Erick Delage from HEC Montréal

Title: Towards Robust and Risk-aware Contextual Optimization
Speaker: Prof. Erick Delage, Professor in Department of Decision Sciences, HEC Montréal
Moderator: Prof. Guodong Li, Associate Director of HKU IDS; Professor, Department of Statistics & Actuarial Science, HKU
Date: May 21, 2024
Time: 10:00am – 11:00am
Venue: Tam Wing Fan Inno Wing II / Zoom
Mode: Hybrid. Seats for on-site participants are limited. A confirmation email will be sent to participants who have successfully registered.
 

Abstract

Recently there has been a surge of interest in operations research (OR) and the machine learning (ML) community in combining prediction algorithms and optimization techniques to solve decision-making problems in the face of uncertainty. This gave rise to the field of contextual optimization, under which data-driven procedures are developed to prescribe actions to the decision-maker that make the best use of the most recently updated information. A large variety of models and methods have been presented in both OR and ML literature under a variety of names, including data-driven optimization, prescriptive optimization, predictive stochastic programming, policy optimization, (smart) predict/estimate-then-optimize, decision-focused learning, (task-based) end-to-end learning/forecasting/optimization, etc. The first part of the talk will identify three main frameworks for learning policies from data and sort out the literature that has pioneered this emerging field. The second part of the talk will present an overview of our groups efforts towards making contextual optimization methods more robust and risk-aware.

Speaker

Prof. Erick Delage
Professor @ Department of Decision Sciences, HEC Montréal

Professor Erick Delage is a professor in the Department of Decision Sciences at HEC Montréal, a chairholder of the Canada Research Chair in decision making under uncertainty, and a member of the College of New Scholars, Artists and Scientists of the Royal Society of Canada. His research interests span the areas of robust and stochastic optimization, decision analysis, reinforcement learning, and risk management with applications to portfolio optimization, inventory management, energy, and transportation problems.

Moderator

Prof. Guodong Li
Associate Director @ HKU IDS; Associate Head (Research) & Professor @ Department of Statistics & Actuarial Science

Professor Guodong Li joined the Department of Statistics & Actuarial Science, The University of Hong Kong, in 2009 as an Assistant Professor, and currently is a Professor. Prior to this, Professor Li had worked at the Division of Mathematical Sciences, Nanyang Technological University, Singapore, as an Assistant Professor since he received his PhD degree in statistics from the University of Hong Kong in 2007. He got his Bachelor and Master degrees in Statistics from Peking University.

For full biography of Professor Li, please refer to: https://datascience.hku.hk/people/professor-guodong-li/

For information, please contact:
Email: datascience@hku.hk

IDS Guest Seminar: Bridging the Representation Gap between Humans and Computers for Video Production

IDS Guest Seminar - by Dr. Anyi RAO

Title: Bridging the Representation Gap between Humans and Computers for Video Production

Speaker: Anyi Rao, Postdoctoral Scholar, Stanford University 

Date: May 14, 2024
Time: 9:30am – 10:30am
Venue: HKU IDS Seminar Room, P603, Graduate House / Zoom

Mode: Hybrid. Seats for on-site participants are limited. A confirmation email will be sent to participants who have successfully registered. 

Abstract

Videos are a beautiful way to share our lives, ideas, stories, and emotions. Recent generative models (e.g. SORA) can generate photorealistic short videos. However, they remain far from being able to create complicated artwork (e.g. films) that requires more human creativity due to the huge gap between human mental representations on control signals and computer representations on pixels, timestamps. To address this challenge, I design controllable and reliable tools to bridge this gap such that creators interact with the tool with conceptual-friendly control signals and produce desired content in a more efficient way. Each module in the tool is reliable and explainable, which allows the creators to input their intentions, get their expected outputs, and know what happened within it. This allows creators to make iterative improvements in the video creation process rather than numerous trial-and-errors.

Speaker

Dr. Anyi Rao
Postdoctoral Scholar @ Stanford University  

Anyi Rao is a Postdoctoral Scholar at Stanford. He studies reliable human-centered AI for creativity and film, focusing on intelligent media editing and creation, semantic and cinematic analysis, aiming to build connections between AI and humans for collaborative intelligence and unleash human creativity and productivity. His works include ControlNet, AnimateDiff, MovieNet, Virtual Studio, Shoot360, and CityNeRF, with a Marr Prize (ICCV best paper award). He leads the organization of the Creative Video Editing and Understanding Workshop at CVPR24, ICCV23, the Generative Models Course at SIGGRAPH24, and the 2023 Paris AI Short Film Festival.

For information, please contact:
Email: datascience@hku.hk

HKU IDS e-Newsletter – Interview on Teachers (Volume 2, Issue 1 @ May 2024)

HKU IDS e-Newsletter - Volume 2, Issue 1 (May 2024)

Welcome our new HKU IDS scholars!

Interdisciplinary Synergy – Interviews with HKU IDS Scholars with Joint Appointments

Our HKU IDS Scholar –
Professor Qingpeng ZHANG, Associate Professor, Co-hosted at the Department of Pharmacology and Pharmacy

Our HKU IDS Scholar –
Professor Boris BABIC, Associate Professor, Co-hosted at the Department of Philosophy

HKU IDS e-Newsletter – Interview on Students (Volume 2, Issue 1 @ May 2024)

HKU IDS e-Newsletter - Volume 2, Issue 1 (May 2024)

Interview on students on course feedback

The Future of Data Science – Feature Story about Teaching & Learning at HKU IDS:
Listen to how students enrolling in our graduate courses feel!

Our HKU IDS students are motivated to become better researchers and supervisors after taking the graduate courses last semester.

Listen to how our Year 1 PhD students, Mr Yunchao ZHANG and Mr Anupam PANI, feel about DATA8003 “Theoretical Foundation of Deep Learning“, by Professor Difan ZOU, and DATA8010 ” Embodied AI: Perception, Representation and Action“, by Professor Yanchao YANG.

We welcome both HKU IDS and non-HKU IDS students to enrol in our research postgraduate programme.

Let’s check out how the two non-IDS students, Mr Mengjin ZHANG from School of Biological Sciences, and Mr Likai PENG from Education, found a pleasant and rewarding experience in our highly collaborative research environment by taking part in DATA8002 “Statistical Inference and Machine Learning for Network Data” by Professor Alec KIRKLEY and DATA8005 “Advanced Natural Language Processing” by Professor Tao YU.

AFAC2024 – Advanced FinTech AI Competition

Organized by Alibaba, and Supported by HKU IDS

AFAC2024 - Advanced FinTech AI Competition Promotion Seminar

Title: Promotional Seminar of AFAC2024 – Advanced FinTech AI Competition by Alibaba Ant Group
Date: May 7, 2024 (Tue)
Time: 3:00pm – 4:30pm HKT 
Venue: HKU IDS Seminar Room (P603), Graduate House

Remarks: The event will feature seasoned FinTech experts who will provide insights into the competition details. Gifts are prepared for interactive attendees!

About AFAC2024

Centered on the themes of large-language-model technology and entrepreneurial innovation, the AFAC2024 focus on real industry challenges and provides vast amounts of genuine industry data, and it also offers over 1.3 million (CNY) prize pool. The competition comprises three categories -Challenge Group, Start-up Group, and Enterprise Group – forming an integrated contests of algorithm, creative application and business practice. The three groups target individuals with technical backgrounds from academic institutions and corporations, startup teams, and SMEs, encouraging contestants to delve into continuous exploration and breakthroughs in AI algorithms, tackling complex technological challenges.

For information, please contact:
Email: datascience@hku.hk

Centre for Information Technology in Education Research Symposium (CITERS) 2024

Centre for Information Technology in Education Research Symposium 2024 (CITERS 2024)

Theme

Digital Transformation: Innovating for Future Learning 

 

The Centre for Information Technology in Education (CITE) of the Faculty of Education of The University of Hong Kong (HKU) is holding its research symposium, that is, CITERS 2024, on 3 and 4 May, 2024 at HKU. The theme of this year’s symposium is ‘Digital Transformation: Innovating for Future Learning’. CITE, which aims to strive for excellence in the use of IT in education, provides a platform for individuals and institutions, dreamers and experts to come together to build new knowledge about learning and transformative use of technology. CITERS 2024 welcomes researchers, practitioners and people of related fields to share ideas, research findings and good practices, and join our discussions on various topics. This year’s sub-themes include:

  1. Digital Equity in Education
  2. Student-Driven Innovation
  3. Technological Boundaries and Education Transformation
  4. Community Engagement in Learning
  5. Artificial Intelligence, Open Data, and Digital Resistance
  6. Learning and Teaching with Digital Technologies
  7. Digital Literacy for Future Readiness
  8. Learning Design and Learning Analytics

Panel Discussion

What role can data science play in transforming learning for a digital future?

Date & Time: 10:00 – 11:00, 3 May 2023 (Friday)
Venue: CPD-3.28, 3/F. The Jockey Club Tower, Centennial Campus, The University of Hong Kong 
Chair: Professor Gary WONG, Faculty of Education, The University of Hong Kong

Panelists:

Enquiries

For enquiries, please contact the Organiser:

Phone
☎️ +852-2241 5325

Email
📧 citers@cite.hku.hk

IDS Seminar: Principles of Reasoning: Designing Compositional and Collaborative Generative AIs

Host: HKU Musketeers Foundation Institute of Data Science
Co-host: Department of Computer Science, HKU

IDS Seminar - by Prof. William Wang

Title: Principles of Reasoning: Designing Compositional and Collaborative Generative AIs

Speaker: William Wang, Mellichamp Professor of Artificial Intelligence; Director of the UCSB Center for Responsible Machine Learning

Date: Apr 18, 2024
Time: 11:00am – 12:00pm
Venue: HKU IDS Seminar Room (P603), Graduate House / Zoom

Mode: Hybrid. Seats for on-site participants are limited. A confirmation email will be sent to participants who have successfully registered. 

Abstract

A majority of existing research in large language models and generative AI systems focuses on scaling and engineering. In this talk, I argue that we need a principled understanding of the science of generative AI, in particular, to understand the reasoning ability of large language models. First, I present a Bayesian latent variable approach to enhancing in-context learning in large language models (LLMs) through optimal demonstration selection, demonstrating substantial improvements across various text classification tasks. Second, I argue that modern generative AI systems must be modular and collaborative to solve complex reasoning problems. I will introduce Logic-LM, a locally grounded neuro-symbolic framework that synergizes LLMs with symbolic solvers, significantly boosting logical problem-solving abilities. We will also briefly elaborate on how to build neuro-symbolic solutions to improve the compositionality in text-to-image systems. Our observations indicate that the future of generative AI is modular and collaborative, as opposed to a single-model system.

Speaker

Prof. William Wang 
Mellichamp Professor of Artificial Intelligence; Director of the Center for Responsible Machine Learning @ UC Santa Barbara

William Wang is Mellichamp Professor of Artificial Intelligence and Director of the UCSB Center for Responsible Machine Learning, UCSB Mind and Machine Intelligence Initiative, and the UCSB Natural Language Processing Group. His Ph.D. was from Carnegie Mellon University. His interests include the science of large language models and generative AIs, vision and language, neuro-symbolic reasoning, and responsible AI. He was recognized with several awards, including the Pierre-Simon Laplace Award by IEEE SPS (2024), the CRA Undergraduate Research Faculty Mentoring Award (2023), the British Computer Society – Karen Spärck Jones Award (2022), and the NSF CAREER Award in 2021. Dr. Wang was also listed among IEEE AI’s 10 to Watch in 2020 and has received accolades for his research, including the CVPR Best Student Paper Award in 2019 and the DARPA Young Faculty Award in 2018.

Moderator

Professor Yi Ma
Director; Professor, Chair of Artificial Intelligence @ HKU IDS & Department of Computer Science

Professor Yi Ma is a Chair Professor in the Musketeers Foundation Institute of Data Science (HKU IDS) and Department of Computer Science at the University of Hong Kong. He took up the Directorship of HKU IDS on January 12, 2023. He is also a Professor at the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. He has published about 60 journal papers, 120 conference papers, and three textbooks in computer vision, generalized principal component analysis, and high-dimensional data analysis.

Professor Ma’s research interests cover computer vision, high-dimensional data analysis, and intelligent systems. For full biography of Professor Ma, please refer to: https://datascience.hku.hk/people/yi-ma/

For information, please contact:
Email: datascience@hku.hk

IDS Distinguished Speaker Series #5: The Future of AI & Entrepreneurship – Direct from Silicon Valley

IDS Distinguished Speaker Series #5 - Captain Hoff from Founders Space

Host: HKU Musketeers Foundation Institute of Data Science
Co-Host: HKU Techno-Entrepreneurship Core

Title: The Future of AI & Entrepreneurship – Direct from Silicon Valley
Speaker: Mr. Steven Hoffman (Captain Hoff), Chairman & CEO, Founders Space
Moderator: Prof. Yi Ma, Director of HKU IDS; Professor, Chair of Artificial Intelligence, HKU
Date: April 12, 2024
Time: 5:00pm – 6:15pm HKT
Venue: Rayson Huang Theatre / Zoom

Mode: Hybrid. Seats for on-site participants are limited. A confirmation email will be sent to participants who have successfully registered.

Abstract

See the latest breakthroughs in AI from Silicon Valley and how entrepreneurs are leveraging the power of generative AI to transform the world. Steve Hoffman (Captain Hoff) will show you how AI is set to upend every aspect of our lives and society, as he takes us on a journey into the future.

Speaker

Mr Steven Hoffman (Captain Hoff)
Chairman & CEO @ Founders Space
 

Steven Hoffman (Captain Hoff) is an expert on AI business strategy and author of three award-winning books: “The Five Forces” (AI) published by BenBella and distributed by Penguin Random House; “Make Elephants Fly” (innovation) published by Hachette; “Surviving a Startup” (entrepreneurship) published by HarperCollins.

 

Hoffman is also a venture investor and Chairman & CEO of Founders Space, a global startup accelerator and innovation hub, with over 50 partners in 22 countries. Founders Space was ranked #1 accelerator for international startups by Forbes and Entrepreneur Magazines.

Hoffman has helped hundreds of entrepreneurs launch their companies in Silicon Valley and around the globe. He has also advised corporations on innovation strategy, including Qualcomm, Huawei, Bosch, Intel, Disney, Warner Brothers, Ad Age, MetLife, NBC, A&E, Siemens, Viacom, Turner, Gulf Oil, etc.

Prior to this, Hoffman worked as a TV development executive at Fries Entertainment, which produced over a hundred TV shows, movies, and mini-series that were acquired by MGM. He went on to pioneer interactive television with his venture-funded startup, Spiderdance, which produced interactive TV shows with NBC, MTV, Turner, Warner Brothers, History Channel, Game Show Network, and others.

While in Hollywood, Hoffman served on the Board of Governors of the Producers Guild of America’s New Media Council. He was also founder and Chairman of the Producers Guild Silicon Valley Chapter and a founding member of the Academy of Television’s Interactive Media Group.

As a serial entrepreneur, Hoffman combined new technologies with entertainment, producing groundbreaking games and interactive shows, like WebRIOTNo BoundariesWeakest LinkYumby, and RocketOn. He was also Head of Infospace’s Mobile Games Studio, whose play-for-prizes games included TetrisWheel of FortuneTomb RaiderThiefHitmanSkee-BallX-Files, and more.

Hoffman earned a bachelor’s degree in computer engineering from the University of California and a master’s degree in film and television from the University of Southern California. He currently resides in California but spends most of his time helping entrepreneurs and innovators around the world bring their dreams to life.

Moderator

Professor Yi Ma
Director; Professor, Chair of Artificial Intelligence @ HKU IDS & Department of Computer Science 

Professor Yi Ma is a Chair Professor in the Musketeers Foundation Institute of Data Science (HKU IDS) and Department of Computer Science at the University of Hong Kong. He took up the Directorship of HKU IDS on January 12, 2023. He is also a Professor at the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. He has published about 60 journal papers, 120 conference papers, and three textbooks in computer vision, generalized principal component analysis, and high-dimensional data analysis. 

Professor Ma’s research interests cover computer vision, high-dimensional data analysis, and intelligent systems. For full biography of Professor Ma, please refer to: https://datascience.hku.hk/people/yi-ma/

For information, please contact:
Email: datascience@hku.hk