Overview

Supervised Learning

• DATA: Data with labels
• TASKS: Classification, Regression, Object Detection, Semantic Segmentation, Image Captioning
• GOAL: To learn a function to map inputs \(x\) to outputs \(y\)
• ALGORITHMS: Support Vector Machines (SVM), Decision Trees, Random Forests, Gradient Boosting Machines (GBM), Neural Networks
• ARCHITECTURES: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), Transformers

Unsupervised Learning

• DATA: Data without labels
• TASKS: Clustering, Dimensionality Reduction, Feature Learning, Density Estimation
• GOAL: To learn some underlying hidden structure of the data
• ALGORITHMS: k-means, Hierarchical Clustering, Principal Component Analysis (PCA), Gaussian Mixture Models, Variational Autoencoders (VAE), Diffusion Models
• ARCHITECTURES: Autoencoders, Generative Adversarial Networks (GANs)

Self-Supervised Learning

• DATA: Data without labels; uses inherent structure of the data itself for supervision
• TASKS: Representation Learning, Anomaly Detection, Contrastive Learning
• GOAL: To learn useful representations or features from the data itself without explicit external labels
• ALGORITHMS: Contrastive Predictive Coding, SimCLR, BYOL, Masked Auto Encoders
• ARCHITECTURES: Transformers, CNNs, Variational Auto Encoders, Siamese Networks

Ponderers