GRAM Workshop @ ICML 2024
The Geometric-Grounded Representation Learning and Generative Modelling GRaM workshop aims to provide a platform that fosters learning, collaboration, and the advancement of the geometry-grounded methods in machine learning. In this Blogpost-Tutorial track, we intend to encourage transparent discussions and opinions in the field, and make geometric machine learning more accessible.
This blogpost track is directly inspired by the amazing ICLR blogpost track. 
We are pleased to announce the blogposts and tutorials accepted at our workshop!
Accepted Blogposts:
- Do Transformers Really Perform Bad for Graph Representation?
- Applications of TopoX to Topological Deep Learning
- Learning Embedding Spaces with Metrics via Contrastive Learning
- Equivariant Diffusion for Molecule Generation in 3D using Consistency Models
- Accelerating Equivariant Graph Neural Networks with JAX
- Equivariant Neural Fields - continuous representations grounded in geometry
- Correct, Incorrect and Extrinsic Equivariance?
- Effect of equivariance on training dynamics
- Towards Equivariant Adaptation of Large Pretrained Models
Accepted Tutorials:
We recommend using Safari for accurately rendering MathJax formulas on Google Colab
- Manifold Free-Form Flows
- Data Representations on the Bregman Manifold
- PHATE Representations Can Effectively Capture Continuous Population Structure in Human Genomic Data
- LaB-GATr: Detailed Model Reference And Usage Tutorial
This is just the beginning!
Do you feel like writting a blogpost or a tutorial that is linked with geometry? We would love to host you!
Check our instructions to submit your blogpost as a pull request. You can also write a colab notebook. Don’t forget to ping our GRaM co-organiser: alison.pouplin [at] aalto [dot] fi 