Self-organizing data driven learning hardware with local interconnections
First Claim
1. A data-driven self-organizing hardware system, comprising:
- a plurality of self-organizing processors arranged in a learning array; and
a plurality of local interconnections that connect subsets of the plurality of self-organizing processors, where the local interconnections are chosen, substantially in parallel, by the self-organizing processors.
1 Assignment
0 Petitions
Accused Products
Abstract
A method for organizing processors to perform artificial neural network tasks is provided. The method provides a computer executable methodology for organizing processors in a self-organizing, data driven, learning hardware with local interconnections. A training data is processed substantially in parallel by the locally interconnected processors. The local processors determine local interconnections between the processors based on the training data. The local processors then determine, substantially in parallel, transformation functions and/or entropy based thresholds for the processors based on the training data.
-
Citations
27 Claims
-
1. A data-driven self-organizing hardware system, comprising:
-
a plurality of self-organizing processors arranged in a learning array; and
a plurality of local interconnections that connect subsets of the plurality of self-organizing processors, where the local interconnections are chosen, substantially in parallel, by the self-organizing processors. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. A method for organizing artificial neurons to facilitate performing artificial neural network tasks, comprising:
-
pseudo randomly connecting a plurality of artificial neurons;
receiving a training data; and
locally determining, substantially in parallel, for an artificial neuron, one or more of, local interconnections between two or more artificial neurons, a transformation function, and an entropy based threshold, based, at least in part, on the training data. - View Dependent Claims (19, 20, 21, 22, 23)
-
-
24. A method for classifying a data point in an input data space using a self organized, locally interconnected learning array, comprising:
-
receiving an input vector;
identifying one or more artificial neurons that code for the input vector; and
providing an indicator of the one or more associations.
-
-
25. A system for organizing a data-driven learning array, comprising:
-
means for establishing an initial pseudo random array of locally connected self-organizing processors; and
means for training the array of self-organizing processors to facilitate performing artificial neural network functions. - View Dependent Claims (26)
-
-
27. A method for organizing artificial neurons to facilitate performing artificial neural network tasks, comprising:
-
pseudo randomly connecting a plurality of artificial neurons in a first array;
training up the artificial neurons to perform an artificial neural network task;
identifying a plurality of artificial neurons that code for the artificial neural network task;
identifying a plurality of connections between the identified artificial neurons; and
designing a second array that comprises the identified artificial neurons, the identified connections, and a pseudo random connection of a plurality of artificial neurons.
-
Specification