Randomness and Monads

Functional Programming and Intelligent Algorithms

(Sist oppdatert: 1 January 2015)

Menu

Overview

Randomness

Step 1: Random weights

Amend your neural network so that weights can be initialised randomly.

Keep a copy of your original program, so that you can compare performance.

Step 2: Testing

Test both your versions, with fixed and with random weights, three times each and compare the results. What observations can you make?



Hans Georg Schaathun / hasc@hials.no