Types of Machine Learning

• The lecture discusses different types of machine learning, specifically supervised, unsupervised, semi-supervised, and reinforcement learning.
• It gives examples of algorithms for each type and real-world applications.

Supervised Learning

• In supervised learning, the algorithm learns from labeled data, meaning each data point is tagged with the correct answer.
• The goal is to learn a mapping function that can predict the output for new, unseen inputs.
• Algorithms include linear regression, support vector machines (SVM), decision trees, and neural networks.
• Linear regression models the relationship between variables with a linear equation.
• Support vector machines find the optimal hyperplane to separate data into classes.
• Decision trees split data based on features to create a tree-like structure for classification or regression.
• Neural networks are complex models inspired by the human brain with interconnected nodes arranged in layers.
• Uses include image classification, where an algorithm is trained to recognize objects in images.
• Another application is spam detection, where emails are classified as spam or not spam.
• Example: predicting house prices based on features like size and location.

Unsupervised Learning

• In unsupervised learning, the algorithm learns from unlabeled data, discovering patterns and structures without prior knowledge.
• The goal is to find hidden relationships, cluster data, or reduce dimensionality.
• Algorithms include k-means clustering, hierarchical clustering, and principal component analysis (PCA).
• K-means clustering partitions data into k clusters, where each data point belongs to the cluster with the nearest mean.
• Hierarchical clustering builds a hierarchy of clusters, either agglomerative (bottom-up) or divisive (top-down).
• Principal component analysis reduces the dimensionality of data by finding the principal components that explain the most variance.
• Uses include customer segmentation, where customers are grouped based on purchasing behavior.
• Another application is anomaly detection, identifying data points that deviate significantly from the norm.
• Example: grouping news articles by topic without predefined categories.

Semi-Supervised Learning

• Semi-supervised learning combines labeled and unlabeled data to improve learning performance.
• Falls between supervised and unsupervised learning.
• Useful when labeling data is expensive or time-consuming.
• Algorithms include self-training and label propagation.
• Self-training trains a model on labeled data and then uses it to predict labels for unlabeled data, adding high-confidence predictions to the training set.
• Label propagation propagates labels from labeled data points to nearby unlabeled data points based on similarity.
• Uses include speech recognition, where a small amount of labeled speech data is combined with a large amount of unlabeled speech data.
• Another application is document classification, where a few labeled documents are used to classify a larger set of unlabeled documents.

Reinforcement Learning

• In reinforcement learning, an agent learns to make decisions in an environment to maximize a reward signal.
• The agent interacts with the environment, takes actions, and receives feedback in the form of rewards or penalties.
• The goal is to learn an optimal policy that maps states to actions.
• Algorithms include Q-learning and policy gradients.
• Q-learning learns a Q-value function that estimates the expected cumulative reward for taking a specific action in a specific state.
• Policy gradients directly optimize the policy by adjusting the parameters to increase the probability of actions that lead to higher rewards.
• Uses include game playing, where an agent learns to play games like chess or Go.
• Another application is robotics, where a robot learns to perform tasks through trial and error.
• Example: training a robot to navigate a maze by rewarding it for reaching the goal.