Supervised learning involves training a model on labeled data, where the input-output pairs are known.
The model learns to map inputs to outputs, and it is evaluated based on its performance on unseen data.
Common Models:
Linear Regression: Predicts continuous values.
Logistic Regression: Used for binary classification.
Support Vector Machines (SVM): Classification and regression.
Decision Trees & Random Forests: Classification and regression using tree structures.
k-Nearest Neighbors (k-NN): Instance-based learning for classification and regression.
Neural Networks: Deep learning models capable of complex mappings.
2. Unsupervised Learning Models
Overview:
Unsupervised learning models are trained on data without labeled outputs.
The goal is to identify patterns or groupings within the data.
Common Models:
k-Means Clustering: Groups data into clusters based on similarity.
Hierarchical Clustering: Creates a tree of clusters.
Principal Component Analysis (PCA): Dimensionality reduction technique.
Autoencoders: Neural networks used for encoding data into lower dimensions and reconstructing it.
t-SNE: Visualization technique for high-dimensional data.
3. Reinforcement Learning Models
Overview:
Reinforcement learning (RL) models learn by interacting with an environment and receiving feedback in the form of rewards or penalties.
The model aims to maximize cumulative rewards over time.
Key Concepts:
Agents: Entities that make decisions.
Environment: The space where agents operate.
Actions, States, Rewards: Fundamental components of RL.
Common Algorithms:
Q-Learning: A value-based approach to learning the quality of actions.
Deep Q-Networks (DQN): Combines Q-Learning with deep learning.
Policy Gradient Methods: Directly optimize the policy (the decision-making strategy).
Proximal Policy Optimization (PPO): A robust policy optimization algorithm.
4. Generative Models
Overview:
Generative models aim to generate new data that resembles a given dataset.
These models can create realistic data, such as images, text, or audio.
Common Models:
Generative Adversarial Networks (GANs): Consist of two neural networks (generator and discriminator) that train together to generate realistic data.
Variational Autoencoders (VAEs): Encode data into a latent space and generate new data by sampling from this space.
Diffusion Models: Generate data by reversing a diffusion process, where data is gradually corrupted and then restored.
5. Sequence Models
Overview:
Sequence models are designed to work with sequential data, such as time series, language, or audio.
These models capture temporal dependencies and can generate or predict sequences.
Common Models:
Recurrent Neural Networks (RNNs): Handle sequences by maintaining a hidden state across time steps.
Long Short-Term Memory (LSTM): An advanced RNN that mitigates the vanishing gradient problem.
Gated Recurrent Units (GRU): A simpler alternative to LSTMs.
Transformers: Use attention mechanisms to process sequences without the need for recurrence, leading to faster training and better performance on tasks like language translation.
