Geometric deep learning has important applications in the fields of quantum computing, 3D perception, molecular designs, and the discovery of mathematical theorems. It takes account of properties such as invariance and equivariance. Many existing structure-aware deep networks lack rigorous theoretical foundations of desired properties in modelling, such as network stability, interpretability, and efficient computation. This workshop will gather researchers from mathematics and computer sciences to provide a forum to establish diverse mathematical theories for geometric deep learning, such as harmonic analysis, algebraic topology, algebraic geometry, combinatorics, differential geometry, differential equations, graph theory, approximation theory, statistics, and theoretical computer science.
Photos can be found at Google Drive: https://drive.google.com/drive/folders/1jJ4eDP1UF1jVhe-AIqhgcfi5uTbm9Z96?usp=sharing
The congress schedule can be found at https://iciam2023.org/3051
For our MS, the schedule can be seen below, and also https://iciam2023.org/registered_data?id=00211
Our minisymposium is in hybrid format of in person talks and online talks.
Zoom for online talks:
https://iciam2023.org/3219#Zoom_Events
For non-registered participants, use the following link:
Join from PC, Mac, Linux, iOS or Android: https://unsw.zoom.us/j/83673273013?pwd=Q2hpalVUY3pwdlU4VFhHdmVCa25aZz09
Meeting ID: 83673273013
Password: 152671
Our session is in the campus of Waseda University E812 of Building 11.
Please log in the submission system at https://review.iciam2023.org/
Then choose New Submission with Minisymposium Identifier Code [MIC] as VS6WWB55
There you can submit your talk and author information to be reviewed.
The talk title and abstract and your bio will also be put on the this website.
Please make submission at least one week before the submission deadline April 20, 2023.
The ICIAM 2023 Conference sessions will be held on the Tokyo local afternoon (UTC+9) over two days,
August 21 and August 22.
Keynote speakers are allotted 30 minutes and invited speakers 20 minutes including Q&A.
On August 21, sessions will be held in Room 1C (13:30-15:00) and Room 1D (15:30-17:10).
On August 22, sessions will be in Room 1E (13:30-15:00) and Room 2C (15:30-17:10). * indicates keynote speakers.
Zoom for online talks:
https://iciam2023.org/3219#Zoom_Events
For non-registered participants, use the following link:
Join from PC, Mac, Linux, iOS or Android: https://unsw.zoom.us/j/83673273013?pwd=Q2hpalVUY3pwdlU4VFhHdmVCa25aZz09
Meeting ID: 83673273013
Password: 152671
Date | Time (Tokyo) | Room | Speaker | Title | |
---|---|---|---|---|---|
Day 1 | |||||
Aug 21 | 13:20 | E812 | Yuguang Wang (SJTU) | Opening Remarks | |
Aug 21 | 13:30 | E812 | Xiaowen Dong (Oxford)* | On the stability of spectral graph filters and beyond | |
Aug 21 | 14:00 | E812 | Xiaosheng Zhuang (CityU HK) | Spherical Framelets with Directionality for Spherical Neural Networks | |
Aug 21 | 14:20 | E812 | Huan Xiong (HIT & MBZUAI) | Some Applications of Hyperplane Arrangements in Deep Learning | |
Aug 21 | 14:40 | E812 | Yiqing Shen (JHU) | Graph-Structured Knowledge Enhanced Large Language Model | |
Aug 21 | 15:00 | Lobby | Tea break | ||
Aug 21 | 15:30 | E812 | Qi Ye (SCNU) | Machine Learning in Banach Spaces: A Black-box or White-box Method? | |
Aug 21 | 15:50 | E812 | Dongmian Zou (DKU) | Stable Hyperbolic Neural Networks for Graph Generation and Classification | |
Aug 21 | 16:10 | E812 | Teresa Huang (JHU) | Spectral-Inspired Graph Neural Networks | |
Aug 21 | 16:30 | Online | Cristian Bodnar (Microsoft)* | Geometric Deep Learning from a Topological Viewpoint | |
Aug 21 | 17:10 | Lobby | Tea break | ||
Aug 21 | 17:40 | E812 | Junyu Xuan (UTS) | Negative sampling for graph neural networks based on determinantal point processes | |
Aug 21 | 18:00 | E812 | Yuanhong Jiang (SJTU) | Scattering Message Passing | |
Aug 21 | 18:20 | E812 | Kai Yi (UNSW) | Geometric Diffusion Generative model for protein sequence design | |
Aug 21 | 18:40 | E812 | Sho Sonoda (RIKEN) | Ridgelet Transforms of Neural Network on Manifolds and Hilbert Spaces | |
Aug 21 | 19:00 | Online | Francesco Di Giovanni (Cambridge)* | On oversquashing and expressivity: can GNNs mix variables? | |
Aug 21 | 19:40 | Online | Challenger Mishra (Cambridge) | Mathematical Conjecture Generation and Machine Intelligence | |
Day 2 | |||||
Aug 22 | 13:20 | E812 | Guido Montufar (UCLA & MPI)* | FoSR: First-order spectral rewiring for addressing oversquashing in GNNs | |
Aug 22 | 13:50 | E812 | Mengjia Xu (BrownU & MIT) | DynG2G: An Efficient Stochastic Graph Embedding Method for Temporal Graphs | |
Aug 22 | 14:10 | E812 | Lin Liu (SJTU) | Root-n consistent semiparametric learning under minimal sparsity conditions | |
Aug 22 | 14:30 | E812 | Yuelin Wang (SJTU) | Dynamic systems for neural message passing on graphs |