PDE Layer Innovations
The discussion highlights the innovative use of a PDE layer, which arose from a blend of intuition and mathematical formalism. Mark's proposal emphasizes how diffusion can maintain translation invariance, while also exploring the potential for equivariance under various transformations like rotation. This approach addresses challenges in discretization on lattices, revealing deeper insights into symmetry in machine learning architectures.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Probabilistic Numeric CNNs with Roberto Bondesan - #482
Related Questions