Tutorial 107 - 7 steps to training your deep learning model
Code associated with these tutorials can be downloaded from here: https://github.com/bnsreenu/python_fo...
1. Load data and understand the problem to be solved: Plot data, is it balanced, do you have enough data, do you need image augmentation, is it large data that doesn't fit memory, etc.
2. Cleanup and preprocess data: Remove null for structured data, extract features if not provided, select/drop appropriate features, balance classes, scale/normalize data, etc.
3. Get data ready for training: Reshape data and define X_train, X_test, X_val, y_train, y_test, y_val, define augmentation
4. Define the model: How many layers, appropriate activation functions, do you need to add dropout, appropriate optimizer, loss, metrics, use transfer learning?
5. Train the model: Ensure enough validation data for good accuracy checks during training.
6. Re-train, if necessary: Check accuracy on test data, understand learning curves and continue training the model, if needed.
7. Save the model for future use