Machine Learning // Megha Bose

Logistic Regression - Classification Dressed Up as Regression Mar 2022 - 10 min read

Linear Regression - The Best Straight Line Through the Cloud of Points Mar 2022 - 12 min read

Machine Learning - What It Means to Generalize From Data Mar 2022 - 12 min read

The Modern Optimizer Zoo - Every Variant of Gradient Descent You Didn't Ask For Dec 2020 - 22 min read

Experimental Design & A/B Testing - Making Sure Your Data Answers Your Question Mar 2022 - 13 min read

Causal Inference - Correlation Isn't Wrong, You Just Asked the Wrong Question Mar 2022 - 11 min read

VC Dimension - How Complex a Hypothesis Class Can Be Mar 2022 - 6 min read

Backpropagation - The Circuit That Runs Gradients Backward Dec 2020 - 31 min read

RAG and Vector Databases - Teaching Models to Remember What They Were Never Trained On Jun 2025 - 17 min read

Graph Neural Networks - Learning When the Data Has Structure Jun 2024 - 13 min read

Process Reward Models - Supervising the Steps, Not Just the Answer Jun 2024 - 11 min read

Building LLM Agents - Giving Models the Ability to Act, Evaluate, and Iterate Jul 2024 - 12 min read

Multi-Agent Systems - Intelligence That Emerges From Interaction Apr 2024 - 11 min read

Variance Reduction - Getting More Signal From Every Sample Jun 2024 - 12 min read

Monte Carlo Tree Search - Planning by Sampling, Not Exhaustion Jun 2024 - 12 min read

Reinforcement Learning Theory - Learning to Act Without Being Told What Is Right Aug 2023 - 19 min read

The Manifold Hypothesis - High-Dimensional Data on Low-Dimensional Curves Jun 2024 - 14 min read

Score-Based Diffusion - Learning to Reverse Noise Jun 2024 - 14 min read

Normalizing Flows - Invertible Maps From Simple Noise to Complex Data Jun 2024 - 11 min read

Variational Autoencoders - Encoding Structure Into the Latent Space Jul 2024 - 19 min read

GANs - Two Networks Playing a Game Neither Can Win Alone Jul 2024 - 17 min read

Graphical Models - Probability Distributions With a Picture Jun 2024 - 11 min read

Neural Tangent Kernels - What Infinitely Wide Networks Actually Learn Jun 2024 - 9 min read

Universal Approximation - Why Neural Networks Can Fit Anything Jun 2024 - 10 min read

Convolutional Neural Networks - Translation Invariance by Design Jun 2024 - 11 min read