From the course: Deep Learning: Model Optimization and Tuning
Join today to access over 24,700 courses taught by industry experts.
Avoiding overfitting
From the course: Deep Learning: Model Optimization and Tuning
Avoiding overfitting
- [Instructor] Let's jump to choosing regularizers and dropouts in this video. We will experiment with four values for the regularizer namely none, L1, L2 and L1 _ L2. Let's run the experiment and review the results. Both none and L2 seems to provide equivalent performance and if we run the experiment multiple times, we will see that they switch lead positions. We will go with L2 in this case. Next we move to dropouts. We will experiment with four values, namely 0%, 10%, 20% and 50%. Running these experiments, we see a dropout of 20% provide slightly better results. So we will choose L2 for regularization with a dropout rate of 0.2.
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.
