Time Series with PyTorch

6

Optimization

“If he [Thomas Edison] had a needle to find in a haystack, he would not stop to reason where it was most likely to be, but would proceed at once with the feverish diligence of a bee, to examine straw after straw until he found the object of his search… Just a little theory and calculation would have saved him ninety percent of his labor.” ― Nikola Tesla

As we have seen, neural network architectures are highly flexible structures with vast parameter spaces that allow them to capture intricate patterns within datasets. However, realizing each architecture’s full potential requires thoughtful consideration of hyperparameter optimization — the process of finding the optimal design of a model relative to our data, including selecting ...