Mar 20, 2024 • machine-learning

Machine Learning Fundamentals

Core concepts, algorithm categories and basic terminology of ML

#Machine Learning #AI #Fundamentals

Machine Learning Fundamentals

What is Machine Learning

Machine learning is a branch of artificial intelligence that enables computers to learn from data without being explicitly programmed.

Types of Learning

Supervised Learning

Given labeled training data, learn the mapping relationship from input to output.

Common Algorithms:

Algorithm	Use Case	Characteristics
Linear Regression	Predicting continuous values	Simple, interpretable
Logistic Regression	Binary classification	Outputs probability
Decision Tree	Classification/Regression	Easy to understand
Support Vector Machine	Classification	Effective in high dimensions
Neural Network	Complex pattern recognition	Strong expressive power

Unsupervised Learning

Process unlabeled data to discover intrinsic structures within the data.

Common Algorithms:

Clustering: K-Means, DBSCAN, Hierarchical Clustering
Dimensionality Reduction: PCA, t-SNE, UMAP
Association Rules: Apriori, FP-Growth

Reinforcement Learning

Learn optimal policies through interaction with the environment to maximize cumulative rewards.

Core Concepts:

Agent → Action → Environment
                     ↓
              Reward
                     ↓
              State (new)

Key Terminology

Feature: Attributes or variables used to describe data.

Label: The target value to predict in supervised learning.

Overfitting: The model performs well on training data but poorly on new data.

Underfitting: The model fails to capture the basic patterns in the data.

Cross Validation: A statistical method for evaluating model performance.

Model Evaluation Metrics

Classification Problems

Accuracy = (TP + TN) / (TP + TN + FP + FN)
Precision = TP / (TP + FP)
Recall = TP / (TP + FN)
F1 Score = 2 × (Precision × Recall) / (Precision + Recall)

Regression Problems

MSE (Mean Squared Error) = Σ(y - ŷ)² / n
RMSE (Root Mean Squared Error) = √MSE
MAE (Mean Absolute Error) = Σ|y - ŷ| / n
R² (Coefficient of Determination) = 1 - SS_res / SS_tot

Learning Recommendations

Master mathematical foundations first (linear algebra, probability theory, calculus)
Understand algorithm principles, don’t just use libraries
Do more practical projects to accumulate hands-on experience
Read classic papers and track cutting-edge developments