The HK Tech Forum on Data Science and AI
(DSAI) gathers world-renowned scholars in data science and AI to address
challenging issues in driving data science and AI technology for the benefit of
the society. DSAI is co-organized by Hong Kong Institute for Advanced Study (HKIAS),
School of Data Science (SDS) and Hong Kong Institute for Data Science (HKIDS). The
forum aims to exchange new ideas and technological development among DSAI
scholars in Hong Kong and the rest of the world.
Day 1
Opening Ceremony – Prof. Way Kuo gave
opening speech.
Then
Prof. Joe Qin (Dean, School of Data Science; Director, HKIDS) introduced the
School of Data Science. He mentioned different programmes in Data Science
including PhD, MSc and BSc (Data and System Engineering; Data Science) in 2021.
Prof.
John Hopcroft (Cornell University, USA) was the first speaker and his
presentation entitled “Math for the Big Data Revolution”. Firstly, Prof.
Hopcroft briefed background of graphs, random walks, expected value, etc.
And
then He also introduced a book for data science. Finally, he introduced the project
101 to improve computer science education at the top 33 universities in China.
The
second speaker was Prof. Yi Ma (University of California, Berkeley, USA) and
his topic named “CTRL: Closed-Loop Data Transcription via Rate Reduction”. He proposed
to learn a closed-loop transcription between the distribution of a
high-dimensional multi-class dataset and an arrangement of multiple independent
subspaces, known as a linear discriminative representation (LDR).
Finally,
he recommended some references and books for further study (e.g. High-Dim
Analysis with Low-Dim Models, 2022 & Cybernetics, 1948).
The
third speaker was Prof. Yiran Chen (Duke University, USA) and his presentation
title was “Scalable, Heterogeneity-Aware and Trustworthy Federated Learning”. Firstly,
he briefed two major bottlenecks that hinder applying federated learning in
practice. They are statistical heterogeneity and communication limitation.
Finally, he introduced FedMask – joint computation and communication efficient
personalized Federated Learning via Heterogeneous Masking.
The
fourth speaker was Prof. Yingying Fan (University of Southern California, USA)
and her topic entitled “Asymptotic Properties of High-Dimensional Random
Forests”. She discussed their derived
the consistency rates for the random forests algorithm associated with the
sample CART (Classification & Regression Trees) splitting criterion.
Prof.
Dacheng Tao (JD Explore Academy, China) was the fifth speaker and his
presentation was “More Is Different: ViTAE elevates the art of computer vision”.
Firstly, he briefed the history of neural networks and AI.
Then
Prof. Tao briefed his recent work on transformers named ViTAE which can be
easily adapted to larger-scale parallel computing resources to achieve faster
training. He also introduced his book related to deep learning.
The
sixth speaker was Prof. Kay Chen Tan (The Hong Kong Polytechnic University,
China) and his presentation entitled “Advances in Evolutionary Transfer
Optimization”. Prof. Tan overviewed of evolutionary transfer optimization (ETO)
including Multi-task, GPU-based and Dynamic Optimization.
Finally,
Prof. Tan briefed his future research topics and they are large-scale
optimization, multi-form optimization, complex environment and theoretical
study. Lastly, he also introduced us to study some reference and books.
The
seventh speaker was Dr. Qingpeng Zhang (City University of Hong Kong, China) and
his topic named “GraphSynergy: A Network-inspired Deep Learning Model for
Anticancer Drug Combination Prediction”. Dr. Zhang introduced an end-to-end
deep learning framework based on a protein–protein interaction (PPI) network to
make synergistic anticancer drug combination predictions.
Prof.
Ruth Misener (Imperial College London, China) was the last speaker and his
presentation title named “OMLT: Optimization and Machine Learning Toolkit”. She
introduced OMLT (https://github.com/cog-imperial/OMLT),
an open-source software package incorporating surrogate models, which have been
trained using machine learning, into larger optimisation problems.
Lightning Talks
Clint Ho, Xinyue Li, Linyan Li, Yu Yang,
Xiao Qiao & Xiangyu Zhao (City University of Hong Kong, China)
Firstly, Dr. Clint Chin Pang Ho presented a
topic “Robust Dynamic Decision-Making under Uncertainty”.
Then Dr. Xinyue Li presented her topic
named “Sensor Device and its Novel Application in Digital Health”.
And then Dr. Linyan Li presented “Urbanization
and Health – Using Deep Learning to Study the Built Environment and its Health
Impacts”.
After that Dr. Yu Yang presented “Generative
Choice Models for Subset Selection”.
Finally, Dr. Xiao Qiao presented his topic entitled
“Portfolio Choice for Online Loans”.
Lastly, Dr. Xiangyu Zhao
presented his study “Adaptive and Automated Recommender Systems”.
Reference:
