Table of Contents

Lecture Slides

IoT ML SEMR1 2024-2025.pdf

Machine Learning

ML Tasks

Aspect	Regression	Classification
Output	Continuous numeric value	Discrete class or category
Goal	Predict a specific numeric quantity	Assign data to predefined classes
Examples	House prices, stock prices, temperatures	spam detection, image recognition
Evaluation Metrics	Mean squared error (MSE), R-squared	Accuracy, F1 Score, precision, recall
Common Algorithm	Linear Regression, Decision Trees (Reg.)	logistic regression, decision trees (Class.)

Categories of ML

Supervised Learning
- Algorithm learns from labelled data
- Common tasks: Classification / Regression
Unsupervised Learning
- The algorithm find patterns and similarities
- Common Tasks: Clustering, Dimensionality Reduction
Reinforcement Learning
- The algorithm learns by interacting with an environment
- Used for decision making, game-playing, robotics, automation

ML Workflow

Why Lightweight ML?

Definition: Lightweight Machine Learning models are computationally efficient, with low memory and power usage.

These models are particularly suitable for resource-constrained IoT environments due to:

Low computational power requirements
Limited memory and storage needs
Finite power supply considerations