Hyperparameter (machine learning)

Ever wondered how to tune a car for maximum performance?

You're trying to get the best mileage out of your car, but you're not sure which gear to shift into at what speed.

Imagine adjusting the gear shift based on speed to get the most miles per gallon. Hyperparameter tuning is like finding the perfect gear for your car's engine.

Example

If you shift gears too early or too late, you'll waste fuel. Finding the right gear shifts (hyperparameters) optimizes fuel efficiency (model performance).

Remember this

Choosing the right gear shift (hyperparameter) optimizes fuel efficiency (model performance).

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

Adam combines momentum and RMSprop: adapts per-parameter learning rates

Ever wondered how a car adjusts its speed for smooth turns?

Tesla Model Y

Tesla Model Y is the world's best-selling electric vehicle in 2023

AdaGrad's learning rate decays to zero

Why does a car's speed drop when it goes uphill?

Neural network (machine learning)

Ever tried adjusting the learning rate like tuning a musical instrument?

Fine-tuning (deep learning)

Why can we make a big change with just a tiny tweak?

to standardize: when you need zero mean and unit variance for gradient-based optimization

Why do we need to make data uniform before training a model?

Swipe through more Machine Learning concepts

Open Pocket Polymath