Machine Learning

Tactic2Vec: Deep Metric Learning for Similar Soccer Plays Retrieval
Tactic2Vec is a Deep Metric Learning Framework. It solves the problem of play retrieval by learning a dense vector representation where geometrically and tactically similar scenes are close to each other in the embedding space.
Tech: Python
Code
Tactic2Vec: Deep Metric Learning for Similar Soccer Plays Retrieval
Physics-Informed Neural Network (PINN) for Cryosurgery Simulations
PINN to solve the nonlinear PDE governing temperature distribution during cryosurgery simulations.
Tech: Python
Code
Physics-Informed Neural Network (PINN) for Cryosurgery Simulations
Why does Andrej Karpathy Use SVMs for Embedding Lookups?
In this blog post, we’ll explore the Exemplar-SVM technique promoted by Andrej Karpathy. We’ll discuss the general embedding lookup process, what is exemplar-SVM, explain why this technique is effective, compare it to K-Nearest Neighbors (KNN), and walk through a brief code implementation.
Mohamed Gad
Jun 12, 2026 10 min read
Why does Andrej Karpathy Use SVMs for Embedding Lookups?
Introduction to Physics-informed Neural Networks
Neural networks are well-known for their ability to learn patterns from data, helping us make predictions and decisions in a variety of fields. But what if we could make these networks even smarter by teaching them to follow the rules that govern the world around us?
Mohamed Gad
Oct 19, 2024 6 min read
Introduction to Physics-informed Neural Networks