Spiking neural network for probabilistic computation

Patent metadata

Field	Value
Publication number	US-11449735-B2
Application number	US-201916577908-A
Country	US
Kind code	B2
Filing date	Sep 20, 2019
Priority date	Apr 17, 2018
Publication date	Sep 20, 2022
Grant date	Sep 20, 2022

Abstract

Described is a system for computing conditional probabilities of random variables for Bayesian inference. The system implements a spiking neural network of neurons to compute the conditional probability of two random variables X and Y. The spiking neural network includes an increment path for a synaptic weight that is proportional to a product of the synaptic weight and a probability of X, a decrement path for the synaptic weight that is proportional to a probability of X, Y, and delay and spike timing dependent plasticity (STDP) parameters such that the synaptic weight increases and decreases with the same magnitude for a single firing event.

First claim

What is claimed is:

1. A system for computing conditional probabilities of random variables for Bayesian inference, the system comprising:
neuromorphic hardware configured to implement a spiking neural network, the spiking neural network comprising a plurality of neurons to compute the conditional probability of two random variables X and Y according to the following:
w*P ( X )= P ( X,Y ),

where P denotes probability, and w denotes a synaptic weight between an A neuron and a connected B neuron;
wherein an X neuron and a Y neuron are configured to spike along with the random variables X and Y;
wherein the spiking neural network comprises an increment path for w that is proportional to a product of w*P(X), a decrement path for w that is proportional to P(X, Y), and delay and spike timing dependent plasticity (STDP) parameters such that w increases and decreases with the same magnitude for a single firing event; and
wherein the neuromorphic hardware controls one or more motor vehicle components based on the computed conditional probability of the random variables X and Y.

2. The system as set forth in claim 1 , wherein the spiking neural network implemented by the neuromorphic hardware comprises a plurality of synapses, wherein all neurons, except for the B neuron, have the same threshold voltage, and wherein the synaptic weight w between the A neuron and the B neuron is the only synapse that has STDP, wherein all other synapses have a fixed weight that is designed to trigger post-synaptic neurons when pre-synaptic neurons fire.

3. The system as set forth in claim 2 , wherein a sign of the STDP is inverted such that if the A neuron spikes before the B neuron, the synaptic weight w is decreased.

4. The system as set forth in claim 3 , wherein the spiking neural network implemented by the neuromorphic hardware further comprises an XY neuron connected with both the A neuron and the B neuron, and wherein a delay is imposed between the XY neuron and the A neuron, which causes an increase in the synaptic weight w.

5. The system as set forth in claim 4 , wherein the X neuron is connected with the A neuron, wherein when the X neuron fires, the B neuron spikes after the A neuron in proportion to the synaptic weight w, such that a spiking rate for the B neuron depends on a product between a spiking rate of the X neuron and the synaptic weight w.

6. A neuromorphic hardware implemented method for computing conditional probabilities of random variables for Bayesian inference, the method comprising an act of:
operating a spiking neural network comprising a plurality of neurons to compute the conditional probability of two random variables X and Y according to the following:
w*P ( X )= P ( X,Y ),

where P denotes probability, and w denotes a synaptic weight between an A neuron and a connected B neuron;
wherein an X neuron and a Y neuron are configured to spike along with the random variables X and Y;
wherein the spiking neural network comprises an increment path for w that is proportional to a product of w*P(X), a decrement path for w that is proportional to P(X, Y), and delay and spike timing dependent plasticity (STDP) parameters such that w increases and decreases with the same magnitude for a single firing event; and
wherein the neuromorphic hardware controls one or more motor vehicle components based on the computed conditional probability of the random variables X and Y.

7. The method as set forth in claim 6 , wherein the spiking neural network comprises a plurality of synapses, wherein all neurons, except for the B neuron, have the same threshold voltage, and wherein the synaptic weight w between the A neuron and the B neuron is the only synapse that has STDP, wherein all other synapses have a fixed weight that is designed to trigger post-synaptic neurons when pre-synaptic neurons fire.

8. The method as set forth in claim 7 , wherein a sign of the STDP is inverted such that if the A neuron spikes before the B neuron, the synaptic weight w is decreased.

9. The method as set forth in claim 8 , wherein the spiking neural network further comprises an XY neuron connected with both the A neuron and the B neuron, and wherein the method further comprises an act of imposing a delay between the XY neuron and the A neuron, which causes an increase in the synaptic weight w.

10. The method as set forth in claim 9 , wherein the X neuron is connected with the A neuron, wherein when the X neuron fires, the B neuron spikes after the A neuron in proportion to the synaptic weight w, such that a spiking rate for the B neuron depends on a product between a spiking rate of the X neuron and the synaptic weight w.

11. The system as set forth in claim 1 , wherein the spiking neural network implemented by the neuromorphic hardware further comprises a subtractor circuit, and wherein the subtractor circuit is used to compare the random variables X and Y.

12. The method as set forth in claim 6 , wherein the spiking neural network further comprises a subtractor circuit, and wherein the method further comprises an act of using the subtractor circuit to compare the random variables X and Y.

Assignees

• Hrl Lab Llc

Inventors

• Chang Hao-Yuan
• Jammalamadaka Aruna
• Stepp Nigel D

Classifications

• G06N3/047

  Probabilistic or stochastic networks · CPC title

• G06N3/063

  using electronic means · CPC title

• G06N3/049 Primary

  Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs · CPC title

• G06N3/08

  Learning methods · CPC title