Method for Designing Optimal Convolutional Neural Networks using Parallel Computers

This invention provides a method by which the design of a convolutional neural network can be highly optimized to a specific, labeled data set. The method allows for different mathematical functions to be used that can scope and define the criteria by which the network can be optimized. For example, the network can be optimized for highest accuracy in classification tasks, and/or to meet certain hardware requirements such as minimal use of computer memory. Given a data set, the method can be applied to the data with the end result being an optimized network for that data set.

---

---