Machine Learning & Applied Artificial Intelligence
Comprehensive statistical and applied machine learning curriculum: from mathematical learning principles and feature engineering to supervised/unsupervised algorithms, ensemble methods, deep neural networks, and scalable production MLOps deployment.
Introduction to Machine Learning
What is ML, Tom Mitchell's framework, AI vs ML vs DL taxonomy, Traditional programming vs ML, four learning paradigms, 7-stage MLOps lifecycle, and Python data science tooling.
Mathematical Foundations
Linear algebra, vectors & matrices, matrix operations & multiplication, eigenvalues & eigenvectors, multivariable calculus, partial derivatives, gradient descent, Bayes' theorem, probability distributions, variance, covariance, and correlation.
Data Preprocessing
Features & labels, missing values imputation, outlier handling, Min-Max normalization, Z-score standardization, categorical encoding, feature selection vs extraction, stratified splitting, and imbalanced resampling (SMOTE).
Supervised Learning
Meaning of supervised learning, labeled data, continuous regression vs. discrete classification, empirical risk minimization, model training & inference, and the Bias-Variance tradeoff governing generalization.
Regression Algorithms
Simple & multiple linear regression, OLS Normal Equations, polynomial basis expansions, gradient descent, Ridge (L2), Lasso (L1) sparsity, Elastic Net, and evaluation metrics (MAE, MSE, RMSE, R², Adjusted R²).
Classification Algorithms
Logistic regression, K-Nearest Neighbors, Naive Bayes, Decision Trees, Random Forest bagging, Support Vector Machines (SVM), Kernel trick, and diagnostic metrics (Confusion Matrix, Precision, Recall, F1, ROC-AUC).
Ensemble Learning
What is ensemble learning, Condorcet theorem, Bagging & Random Forest, Boosting paradigms (AdaBoost, Gradient Boosting), modern boosters (XGBoost, LightGBM, CatBoost), Voting classifiers, and Stacking with OOF firewalls.
Unsupervised Learning
Meaning of unsupervised learning, unlabeled data, clustering (K-Means, DBSCAN), density estimation (GMM, KDE), dimensionality reduction (PCA), and association rule learning (Apriori, Support, Confidence, Lift).
Clustering Algorithms
Distance metrics, K-Means & K-Means++, Elbow method, Silhouette score, K-Medoids (PAM), Hierarchical agglomeration & dendrograms, DBSCAN density discovery, Spectral graph Laplacians, and customer persona segmentation.
Dimensionality Reduction
Need for reduction, Curse of dimensionality, Principal Component Analysis (PCA steps & explained variance), Linear Discriminant Analysis (LDA), t-SNE, UMAP basics, and feature reduction vs. selection.
Model Evaluation & Improvement
Overfitting, underfitting, bias-variance tradeoff decomposition, train-validation-test partitions, Stratified K-Fold cross-validation, Grid vs. Random hyperparameter search, regularization, feature importance, and class imbalance remediation (SMOTE).
Neural Networks & Deep Learning
Biological vs. artificial neurons, perceptron, multilayer perceptrons, universal approximation, activations (Sigmoid, Tanh, ReLU, Softmax), backpropagation calculus, Adam optimization, and foundations of CNNs, RNNs, and Transformers.
Natural Language Processing
Text preprocessing, tokenization (word, character, subword BPE), stop-words, stemming vs. lemmatization, Bag-of-Words, TF-IDF, distributed word embeddings (Word2Vec, GloVe), text classification, sentiment analysis, and large language model foundations.
Recommender Systems
What is a recommender system, utility matrices, content-based filtering, user-based and item-based collaborative filtering, low-rank matrix factorization (SVD), and ranking evaluation metrics (Precision@K, NDCG).
Reinforcement Learning
Agent and environment interaction loops, Markov Decision Processes (MDP), policies, state and action value functions, Bellman equations, exploration vs. exploitation (epsilon-greedy), Q-learning temporal difference control, and real-world applications.