Neural network analysis

A neural network computing system comprising:

• a first neural network layer including:
• a first storage device storing a first dictionary entry correlating a first data structure with a first electronic message;
• and
• a first transmission component configured to transmit the first electronic message, the first electronic message including fewer bits than the first data structure;
• a second neural network layer including:
• a second storage device configured to store the first electronic message received from the first neural network layer;
• a second processing element configured to:
• read the first electronic message and output the first data structure based on the first dictionary entry;
• perform a data structure analysis on the first data structure to yield a second data structure;
• match the second data structure to a second dictionary entry that correlates the second data structure to a second electronic message, the second electronic message including fewer bits than the second data structure;
• test performance of a second dictionary that includes the second dictionary entry;
• and
• based on the test of the second dictionary, adjust the size of the second dictionary;
• and
• a second transmission component configured to transmit the second electronic message.

A method of preparing a neural network, the method comprising:

• determining potential results of a data structure analysis associated with a layer of a neural network;
• setting a size of a dictionary to a number of dictionary entries for the layer;
• clustering the potential results of the data structure analysis to determine data structures and an associated electronic message for each of the number of dictionary entries, the number of dictionary entries corresponding to the size of the dictionary;
• testing the performance of the dictionary associated with the layer of the neural network;
• and
• based on the testing, storing the dictionary in adjacent layers in the neural network such that each layer in the neural network has a receiving dictionary that is correlated to a previous neural network layer transmitting dictionary, and a transmitting dictionary that is correlated to a following neural network layer receiving dictionary.