Introduction to Machine Learning

• Machine Learning (ML) involves programming computers so they can learn from data.
• ML is a field of study providing computers with the ability to learn without explicit programming.
• A computer program learns from experience E with respect to task T and performance measure P if its performance on T, measured by P, improves with experience E.
• If spam filters were created traditionally, it would involve noticing spam-like words, creating a detection algorithm, and repeating until good enough.

Why to Use Machine Learning

• Useful for the problems with hand-tuning or long lists of rules, one Machine Learning algorithm can often simplify code and perform better.
• Useful for complex problems with no traditional solution, Machine Learning identifies one.
• Useful for fluctuating environments where a Machine Learning system can adapt to data.
• Useful for insights about complex problems and large amounts of data.

Types of Machine Learning Systems

• Can be classified based on whether they are trained with human supervision.
• Can be classified based on whether they can learn incrementally on the fly.
• Can be classified based on how new data points are compared to existing ones.

Supervised Learning

• Algorithms use training data that includes desired solutions, as seen in classification tasks such as spam filters learning to classify new emails.
• Another typical task is predicting a target numeric value, called regression.
• The most important supervised learning algorithms are k-Nearest Neighbors, Linear Regression, Logistic Regression, Support Vector Machines (SVMs), Decision Trees and Random Forests, and Neural networks2.

Unsupervised Learning

• Algorithms utilize unlabeled training data, focusing on learning without a teacher.
• The unsupervised learning algorithms include Clustering, K-Means, DBSCAN, Hierarchical Cluster Analysis (HCA), Anomaly detection and novelty detection, One-class SVM, and Isolation Forest.

Semi-Supervised Learning

• Algorithms deal with partially labeled training data, typically a lot of unlabeled and a little labeled data.

Reinforcement Learning

• Involves a learning system (agent) that observes, selects, and performs actions in an environment.
• The agent receives rewards (or penalties) and learns a policy to maximize rewards over time.

Batch Learning

• Systems cannot learn incrementally and must be trained using all available data.
• This is often done offline due to time and computing resource demands, also called offline learning.

Online Learning

• Systems train incrementally by feeding data instances sequentially in mini-batches.
• Each learning step is fast and cheap.

Instance-Based Learning

• System learns the examples by heart.
• System generalizes to new cases by comparing them to the learned examples.

Model-Based Learning

• Generalization proceeds by building a model from a set of examples.
• The model is then used to make predictions.

Main Challenges of Machine Learning

• Insufficient data.
• Nonrepresentative training data.
• Poor quality data.
• Irrelevant features.
• Overfitting the training data.
• Underfitting the training Data.
• Testing and Validating.