Quantum Noise Accelerated Learning

This presentation addresses the following topic(s):

• The next innovation: What will be the big news at ISC 2032? Be as specific as you can.

Randomness in machine learning has been an effective method to create greater generalisation in models and improve accuracy. The choice of noise has often been limited to pseudo-random sources that are deterministic in nature causing issues from biasing learning models to the deterministic entropy source. High-throughput Quantum Random Number Generators offer an alternative noise source that are not only truly random, but provide true uniform distributions. Our initial results show promising capabilities for improving the training time of Deep Convolutional Neural Networks.