This repository holds my thesis 'Block BFGS: A class of quasi-Newton methods forunconstrained continuous optimisation'. Mathematical properties of this class of optimisers are explored, followed by experiments testing their performance against the classical BFGS method, and steepest descent. The code for the optimisers is provided, along with the code used to create a neural network for regression.
Vonny0123/Thesis-BlockBFGSMethods
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|