AI & ML Course

Lesson 1 • Types of Machine Learning

Covers supervised, unsupervised, and reinforcement learning paradigms with real-world examples. Connects each paradigm to the problem types addressed later in the course.

Lesson 2 • Setting Up Your ML Environment

Guides students through installing Python, key libraries, and notebook tools. Ensures every student has a working environment before hands-on exercises begin.

Lesson 3 • The ML Workflow End to End

Maps the full pipeline from raw data to deployed model. Provides a mental model students will revisit in every subsequent chapter.

Lesson 4 • What AI and ML Actually Mean

Clarifies the distinction between artificial intelligence, machine learning, and deep learning. Establishes precise vocabulary used throughout the course.

Lesson 1 • Data Splitting and Leakage Prevention

Explains train, validation, and test splits and the critical risk of data leakage. Establishes rigorous evaluation habits that carry through all modeling chapters.

Lesson 2 • Understanding Data Sources

Surveys structured, unstructured, and semi-structured data sources and their trade-offs. Grounds students in realistic data landscapes before any cleaning begins.

Lesson 3 • Handling Missing and Noisy Data

Covers imputation strategies, outlier treatment, and noise reduction methods. Prepares students to make defensible decisions when data is incomplete or corrupted.

Lesson 4 • Feature Engineering and Transformation

Demonstrates encoding, scaling, and creating new features from raw variables. Shows how engineered features directly improve model accuracy in later chapters.

Lesson 1 • Tree-Based Regression Models

Introduces decision tree, random forest, and gradient boosting regressors. Compares their accuracy and interpretability against linear models on the same datasets.

Lesson 2 • Regression Evaluation Metrics

Introduces MAE, MSE, RMSE, and R-squared as complementary performance measures. Teaches students to select the right metric based on business context.

Lesson 3 • Regularization Techniques

Covers Ridge, Lasso, and Elastic Net regularization to combat overfitting in regression. Connects regularization strength to the bias-variance trade-off introduced conceptually earlier.

Lesson 4 • Polynomial and Nonlinear Regression

Extends linear regression to capture curved relationships through polynomial features. Demonstrates when nonlinear models outperform linear baselines.

Lesson 1 • Classification Evaluation Metrics

Covers accuracy, precision, recall, F1, ROC-AUC, and confusion matrices. Teaches metric selection for imbalanced classes and asymmetric error costs.

Lesson 2 • Tree-Based Classification Models

Covers decision trees, random forests, and gradient boosting for classification tasks. Emphasizes ensemble methods that consistently outperform single-tree baselines.

Lesson 3 • Naive Bayes and k-NN Classifiers

Introduces probabilistic and instance-based classifiers as fast, interpretable alternatives. Highlights use cases where these simpler models outperform complex ones.

Lesson 4 • Support Vector Machines

Explains the maximum-margin hyperplane and kernel trick for nonlinear boundaries. Positions SVMs as powerful classifiers for high-dimensional, small-sample problems.

Lesson 1 • Principal Component Analysis

Derives PCA through variance maximization and explains principal components geometrically. Demonstrates PCA as a preprocessing step that improves downstream model performance.

Lesson 2 • Anomaly Detection Methods

Covers isolation forests, autoencoders, and statistical methods for detecting outliers. Applies anomaly detection to fraud, quality control, and system monitoring scenarios.

Lesson 3 • Nonlinear Dimensionality Reduction

Introduces t-SNE and UMAP for visualizing high-dimensional data in two dimensions. Contrasts their strengths and limitations against linear PCA.

Lesson 4 • Association Rule Learning

Explains the Apriori algorithm and key metrics: support, confidence, and lift. Applies market basket analysis to recommendation and cross-sell use cases.

Lesson 1 • Recurrent Networks and Sequence Models

Introduces RNNs, LSTMs, and GRUs for sequential and time-series data. Prepares students for the natural language processing chapter that follows.

Lesson 2 • Perceptrons and Feedforward Networks

Builds a neural network from a single perceptron up to multilayer architectures. Establishes the forward-pass computation graph used in all subsequent deep learning sections.

Lesson 3 • Backpropagation and Optimization

Derives backpropagation using the chain rule and connects it to gradient descent variants. Equips students to tune learning rates and optimizers for stable convergence.

Lesson 4 • Regularization in Deep Networks

Covers dropout, batch normalization, and early stopping to prevent overfitting. Connects these techniques to the bias-variance concepts established in earlier chapters.

Lesson 1 • Transformer Architecture and BERT

Explains self-attention, positional encoding, and the encoder-decoder structure. Demonstrates fine-tuning BERT for classification and question-answering tasks.

Lesson 2 • Word Embeddings

Explains Word2Vec, GloVe, and FastText as dense semantic representations of words. Shows how pretrained embeddings accelerate downstream NLP model training.

Lesson 3 • Text Preprocessing and Representation

Covers tokenization, stemming, stop-word removal, and bag-of-words encoding. Establishes the text-to-vector pipeline required by all NLP models in this chapter.

Lesson 4 • Text Classification and Sentiment Analysis

Builds classifiers for topic labeling and sentiment scoring using classical and deep models. Applies evaluation metrics from the classification chapter to text-specific benchmarks.

Lesson 1 • Scalability and Infrastructure Choices

Compares cloud, on-premise, and edge deployment options for different latency and cost profiles. Guides students in matching infrastructure to model size and request volume.

Lesson 2 • A/B Testing and Model Governance

Covers shadow deployment, A/B testing, and canary releases for safe model rollouts. Introduces governance practices for audit trails and rollback procedures.

Lesson 3 • Model Monitoring and Drift Detection

Teaches data drift, concept drift, and performance degradation detection in production. Connects monitoring alerts to the retraining pipelines introduced in the previous section.

Lesson 4 • ML Pipelines and Orchestration

Introduces pipeline frameworks for automating data ingestion, training, and deployment steps. Ensures reproducibility and reduces manual intervention in recurring ML workflows.