Random Seed

Domain: Data & Datasets

🎯 Motivation

A number that initializes random processes. Same seed = same random sequence = same results. Essential for reproducibility.

📋 Concrete Example

Seed = 42: model gets 85.2% accuracy. Seed = 123: model gets 84.8%. Same seed always gives the same result. Fix the seed for reproducibility.

🔗 Analogy

The starting position in a board game - same starting position = same game possibilities. A random seed sets the starting position for AI training.

Related Concepts

• Reproducibility
• Training Data
• Determinism