20.03.2021 Views

Deep-Learning-with-PyTorch

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

442 CHAPTER 14 End-to-end nodule analysis, and where to go next

• TensorBoard allows us to write out many different types of diagrams that help

us determine what’s going on. But this is not a replacement for looking at data

on which our model works particularly badly.

• Successful training seems to involve an overfitting network at some stage, and

which we then regularize. We might as well take that as a recipe; and we should

probably learn more about regularization.

• Training neural networks is about trying things, seeing what goes wrong, and

improving on it. There usually isn’t a magic bullet.

• Kaggle is an excellent source of project ideas for deep learning. Many new datasets

have cash prizes for the top performers, and older contests have examples

that can be used as starting points for further experimentation.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!