Loss Function
A way to describe your objective in the form of a function
What Is a Loss Function and Loss?
A loss / cost / error function, also known as the objective function describes what we want to work towards: more money, getting this damn car to drive etc. We may want to minimize or maximize this loss function meaning that we're trying to find the weights that'll get us closest to our objective.
The loss function plays a crucial role in machine learning namely that it tries to distill the entirety of your model into one number representing that it's doing better or worse. Therefore it's very important to choose the correct loss function for your project since that function needs to capture the essence of your project. There are a wide variety of loss functions to choose from and you can even write one yourself if you'd like. Some of the more common ones include:
Mean Squared Error
Mean Absolute Error
Mean Bias Error
Cross-Entropy Loss
Last updated
Was this helpful?