UNSUPERVISED, SUPERVISED AND REINFORCED LEARNING VIA SPIKING COMPUTATION

• US 20180268294A1
• Filed: 05/18/2018
• Published: 09/20/2018
• Est. Priority Date: 09/16/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Modha, Dharmendra S.
• Granted Patent: US 11,164,080 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G06N 3/063 using electronic means
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/088 Non-supervised learning, e.g. competitive learning

Abstract

The present invention relates to unsupervised, supervised and reinforced learning via spiking computation. The neural network comprises a plurality of neural modules. Each neural module comprises multiple digital neurons such that each neuron in a neural module has a corresponding neuron in another neural module. An interconnection network comprising a plurality of edges interconnects the plurality of neural modules. Each edge interconnects a first neural module to a second neural module, and each edge comprises a weighted synaptic connection between every neuron in the first neural module and a corresponding neuron in the second neural module.

Claims

1. A method comprising:

receiving output generated by a neuron population of a neural network;

determining whether the output is a false negative or a false positive; and

self-tuning the neural network by;

providing the output to a first set of neuron populations of the neural network in response to determining the output is a false negative, wherein the output is learned as the output propagates through the first set of neuron populations; and

providing the output to a second set of neuron populations of the neural network in response to determining the output is a false positive, wherein the output is unlearned as the output propagates through the second set of neuron populations.

2. The method of claim 1, wherein self-tuning the neural network further comprises:

applying a first learning rule based on firing activity of the first set of neuron populations to adjust synaptic weights of a first set of pathways interconnecting the first set of neuron populations in response to determining the output is a false negative; and

applying a second learning rule based on firing activity of the second set of neuron populations to adjust synaptic weights of a second set of pathways interconnecting the second set of neuron populations in response to determining the output is a false positive.

3. The method of claim 2, wherein the first learning rule is a Hebb learning rule and the second learning rule is an anti-Hebb learning rule.

4. The method of claim 2, wherein the first learning rule is a spike-timing dependent plasticity (STDP) learning rule and the second learning rule is an anti-STDP learning rule.

5. A system comprising a computer processor, a computer-readable hardware storage medium, and program code embodied with the computer-readable hardware storage medium for execution by the computer processor to implement a method comprising:

receiving output generated by a neuron population of a neural network;

determining whether the output is a false negative or a false positive; and

self-tuning the neural network by;

providing the output to a first set of neuron populations of the neural network in response to determining the output is a false negative, wherein the output is learned as the output propagates through the first set of neuron populations; and

providing the output to a second set of neuron populations of the neural network in response to determining the output is a false positive, wherein the output is unlearned as the output propagates through the second set of neuron populations.

6. The system of claim 5, wherein self-tuning the neural network further comprises:

applying a first learning rule based on firing activity of the first set of neuron populations to adjust synaptic weights of a first set of pathways interconnecting the first set of neuron populations in response to determining the output is a false negative; and

applying a second learning rule based on firing activity of the second set of neuron populations to adjust synaptic weights of a second set of pathways interconnecting the second set of neuron populations in response to determining the output is a false positive.

7. The system of claim 6, wherein the first learning rule is a Hebb learning rule and the second learning rule is an anti-Hebb learning rule.

8. The system of claim 6, wherein the first learning rule is a spike-timing dependent plasticity (STDP) learning rule and the second learning rule is an anti-STDP learning rule.

9. A computer program product comprising a computer-readable hardware storage device having program code embodied therewith, the program code being executable by a computer to implement a method comprising:

receiving output generated by a neuron population of a neural network;

determining whether the output is a false negative or a false positive; and

self-tuning the neural network by;

providing the output to a first set of neuron populations of the neural network in response to determining the output is a false negative, wherein the output is learned as the output propagates through the first set of neuron populations; and

providing the output to a second set of neuron populations of the neural network in response to determining the output is a false positive, wherein the output is unlearned as the output propagates through the second set of neuron populations.

10. The computer program product of claim 9, wherein self-tuning the neural network further comprises:

applying a first learning rule based on firing activity of the first set of neuron populations to adjust synaptic weights of a first set of pathways interconnecting the first set of neuron populations in response to determining the output is a false negative; and

applying a second learning rule based on firing activity of the second set of neuron populations to adjust synaptic weights of a second set of pathways interconnecting the second set of neuron populations in response to determining the output is a false positive.

11. The computer program product of claim 10, wherein the first learning rule is a Hebb learning rule and the second learning rule is an anti-Hebb learning rule.

12. The computer program product of claim 10, wherein the first learning rule is a spike-timing dependent plasticity (STDP) learning rule and the second learning rule is an anti-STDP learning rule.