Data Augmentation

Category: Technical Terms

Definition

Data augmentation creates additional training examples by modifying existing data in realistic ways.

How It Works

Instead of collecting more data, you create variations of what you have. For images, this means rotating, cropping, or changing brightness. For text, it means paraphrasing or translating and back-translating.

The AI learns from these variations, making it better at handling new situations.

Why It Matters

Good data is expensive and time-consuming to collect. Data augmentation multiplies your training data without the cost of gathering new examples.

It makes AI models more robust by exposing them to more variety during training.

Related Terms

• Training Data
• Synthetic Data
• Machine Learning
• Model Performance

← Back to Technical Terms | All Terms