ML Curriculum

Master the fundamental data structure of numerical computing in Python. Learn to create, index, and manipulate n-dimensional arrays efficiently.

Unlock the performance of vectorized operations. Understand broadcasting rules that eliminate explicit loops in numerical code.

The mathematical backbone of machine learning. Dot products, matrix multiplication, and eigendecomposition form the language of models.

ML is fundamentally about uncertainty. Bayes theorem, Gaussian distributions, and maximum likelihood estimation give models a rigorous probabilistic foundation.

The choice of loss function defines what your model optimizes. MSE, cross-entropy, hinge loss, and contrastive losses each encode different assumptions.

Raw data is messy. Learn normalization, standardization, and feature engineering — the critical steps that happen before any model sees data.

The foundation of supervised learning. Fit a line through data using the least squares criterion and understand the geometry of prediction.

The optimization engine behind all of machine learning. Learn how parameters update iteratively by following the direction of steepest descent.

Build your intuition for deep learning from scratch. Understand layers, activations, and why depth enables hierarchical feature learning.

The chain rule made computational. Learn how gradients flow backward through a network to train every parameter simultaneously.

Prevent your model from memorizing noise. L1, L2 penalties, dropout, and early stopping each impose different inductive biases.

The simplest unsupervised algorithm: assign points to the nearest centroid and update centroids until convergence. Deceptively powerful in practice.

High-dimensional data is sparse and hard to visualize. PCA finds the directions of maximum variance, compressing without losing what matters.

Agents learn by interacting with an environment. Rewards shape behavior through the Markov Decision Process framework — the basis of game-playing AI.

Training from scratch is expensive. Pre-trained models encode rich representations that transfer across tasks — the backbone of modern applied ML.

Real-world data arrives with outliers, duplicates, missing values, and evolving schemas. Mastering cleaning pipelines is the unglamorous work that makes or breaks model performance.

Raw features rarely speak the language of your model. One-hot encoding, target encoding, standardization, and binning transform data into representations that amplify signal and suppress noise.