Machine Learning: Clustering & Retrieval Course Syllabus

Overview (80-120 words) describing structure and time commitment. This course provides a comprehensive introduction to clustering and retrieval methods in machine learning, with a focus on document retrieval and topic modeling. Learners will gain hands-on experience implementing algorithms such as k-nearest neighbors, k-means, and latent Dirichlet allocation. The course spans approximately 30 hours of learning, divided into six modules, each combining theory and practical programming assignments. Designed for self-paced learning, it includes real-world applications in text analysis and information retrieval, culminating in a final project that integrates all techniques covered.

Module 1: Introduction to Clustering and Retrieval

Estimated time: 4 hours

• Overview of clustering tasks in machine learning
• Introduction to information retrieval systems
• Course structure and learning objectives
• Prerequisites and technical background review

Module 2: Nearest Neighbor Search

Estimated time: 6 hours

• Implementing k-NN for document retrieval
• Measuring similarity in text data using various metrics
• Optimizing k-NN search with KD-trees
• Scaling search using locality-sensitive hashing (LSH)

Module 3: Clustering

Estimated time: 7 hours

• Applying k-means to cluster documents by topic
• Understanding convergence and initialization in k-means
• Parallelizing k-means using MapReduce for large datasets

Module 4: Mixture Models and EM

Estimated time: 6 hours

• Introduction to probabilistic clustering
• Fitting mixture of Gaussian models
• Understanding and implementing the expectation maximization (EM) algorithm

Module 5: Topic Modeling with LDA

Estimated time: 7 hours

• Performing mixed membership modeling with LDA
• Understanding the structure of latent Dirichlet allocation
• Implementing Gibbs sampling for inference in topic models

Module 6: Case Study and Applications

Estimated time: 6 hours

• Applying retrieval and clustering techniques to real-world datasets
• Building a document retrieval system end-to-end
• Comparing supervised and unsupervised learning in retrieval contexts

Prerequisites

• Familiarity with basic machine learning concepts
• Intermediate programming skills in Python
• Basic understanding of probability and statistics

What You'll Be Able to Do After

• Implement document retrieval systems using k-NN
• Apply k-means and LDA for document clustering and topic modeling
• Optimize similarity search with KD-trees and LSH
• Use Gibbs sampling for inference in probabilistic models
• Design scalable clustering solutions using MapReduce