Module 1, L4: remarks

In lecture 4, we will learn about techniques to help your network perform well in practice: 

• How to use train/dev/test data to select hyperparameters. 
• How to help your network generalize well to unseen data, using regularization, batch norm, etc.  

Most videos are from Andrew Ng's course in deep learning at Coursera, https://www.coursera.org/specializations/deep-learning, (Länkar till en externa sida.). See also the page with reading directions.

 

General advices: 

• The videos have intentional been trimmed (the start and end are sometimes omitted). 
• Sometimes the rest of the video appears later (say, after a small quiz) and we therefore recommend that you watch precisely the included part of each video. 
• Use the button "Next" or "Nästa" to move to the next video, quiz or page in the same lecture module. 
• Keep pen and paper ready to take notes, since this will help you stay alert.
• Pause the videos occasionally and try to summarise what you have learned. It is often with brief pause and you don't even need to write down the summary for this to affective. 
• Adjust the playback speed of the videos. Andrew Ng talks very slowly and 1.25 may be a reasonable alternative.

We apologize in advance for all the clicking that you will have to do to watch the videos and hope that you will still appreciate them.