Machine Learning BSCS (CS-4090)

Course Description

This basic course explores the concepts, techniques, and applications of Machine Learning (ML), providing students with the theoretical foundations and hands-on skills required to design, train, and evaluate predictive models. Students will work with Python-based ML tools and libraries such as Scikit-learn, NumPy, and Pandas to solve real-world problems.

Key Topics Covered

• Introduction to Machine Learning
  • History and evolution of ML.
  • Overview of ML categories: supervised, unsupervised, and reinforcement learning.
• Python for Machine Learning
  • Data manipulation with Pandas and NumPy.
  • Data visualization with Matplotlib and Seaborn.
• Regression Models
  • Simple and multiple linear regression.
  • Model evaluation metrics for regression tasks.
• Data Preprocessing and Pipelines
  • Feature scaling, encoding, and transformation.
  • Building automated ML pipelines.
• Classification Models
  • Logistic regression, k-NN, Naïve Bayes.
  • Evaluation metrics: accuracy, precision, recall, F1-score, ROC-AUC.
• Neural Networks and Deep Learning
  • Perceptron and multilayer perceptron architectures.
  • Backpropagation and activation functions.
• Optimization and Regularization
  • Gradient descent variants (SGD, Adam, RMSProp).
  • L1/L2 regularization, dropout.
• Advanced Models
  • Support Vector Machines (SVM) and kernel tricks.
  • Decision trees, random forests, gradient boosting.
• Unsupervised Learning
  • Clustering (K-means, hierarchical, DBSCAN).
  • Dimensionality reduction (PCA, t-SNE).
• Project Work
  • Applying ML models to real datasets.
  • Presentations and discussions on project outcomes.

Learning Outcomes:

By the end of the course, students will:

1. Understand the foundational principles of machine learning and its applications in robotics.
2. Develop and evaluate machine learning models for real-world problems using Python and Scikit-learn.
3. Analyze the impact of various machine learning techniques on learning applications and optimize models for performance.
4. Apply neural network architectures for predictive modeling tasks.

Hands-On Labs and Projects

• Implement regression and classification models using Scikit-learn.
• Create automated ML pipelines for end-to-end workflows.
• Build a neural network for a classification or regression problem.
• Compare performance of various ML algorithms on benchmark datasets.
• Present a complete ML project from data preprocessing to model deployment.