I recently graduated from the University of Waterloo, and I’m building Palatial. In my free time I like to make music on SoundCloud.
Writing
Mixture Density Network (MDN)
MDNs output parameters of a probability distribution instead of point predictions. They excel at modeling data with multiple possible outputs for a single input.
Physics-Informed Neural Network (PINN)
PINNs incorporate physical laws directly into the neural network training process, enabling them to solve differential equations with limited data.
V-JEPA: Abstract Representation Learning
V-JEPA learns abstract representations by predicting masked video patches in representation space, not pixel space. The model learns to "understand" dynamics rather than "paint" pixels, demonstrating true world understanding through joint embedding predictive architecture.
30-Year Evolution of Physics Understanding in AI
MDNs - Memorization Era
Probabilistic predictions, massive data needs
PINNs - Rules Era
Physics equations embedded, reduced data
Early World Models
Self-discovered physics from videos
V-JEPA - Understanding Era
Abstract concepts, not pixels
Hybrid Robotics Architecture
Physics Layer
Integration Layer
Intelligence Layer
When to Use Each Approach
• Manufacturing
• Safety testing
• Exploration
• Human interaction
• Warehouse robotics
• Humanoid robots