AI Terms Dictionary

A comprehensive multilingual AI terminology dictionary

Definition

The bias-variance tradeoff describes the tension between underfitting (high bias) and overfitting (high variance). High bias models make strong assumptions about data, potentially ignoring relevant relationships, while high variance models capture noise as if it were signal. In ethical AI, managing this tradeoff is crucial to ensure models generalize fairly across diverse demographic groups without perpetuating historical biases or failing in real-world deployment scenarios.

Summary

A fundamental problem in supervised learning where minimizing error requires balancing model complexity against generalization ability.

Key Concepts

Use Cases