Dropout and Early Stopping: Two Simple Tricks That Stop Overfitting Without Touching the Math
中文摘要
Dropout与早停法是通过简单技巧防止神经网络过拟合的有效方法，无需涉及复杂数学。
English Summary
Dropout and early stopping are simple techniques to prevent neural network overfitting without requiring complex mathematics.
原文节选
The smartest thing you can do while training a neural network is often to make it worse at its job on purpose. Continue reading on Medium »