Overview
Explore neural network training fundamentals in this 77-minute lecture covering blood-brain barrier (BBB) transmission prediction modeling. Learn essential concepts starting with model creation for BBB transmission, followed by comprehensive coverage of neural network training principles including momentum methods and simple linear model fitting. Dive into crucial concepts like bias/variance tradeoff and various regularization techniques. Master advanced optimization approaches with Adam optimizer and L2 penalty implementation, while understanding the importance of batch normalization in deep learning. Conclude by applying these concepts to practical model development, with access to detailed lecture notes, slides, and supplementary materials for deeper understanding.
Syllabus
Let’s make a model to predict BBB transmission!
Training a neural network
Momentum methods
Fitting a simple linear model
Bias/variance tradeoff
Regularization
Adam and the L2 penalty
Batch normalization
Back to our model
Taught by
Manolis Kellis