Device and method for random walk simulation

Patent metadata

Field	Value
Publication number	US-12572818-B2
Application number	US-202117200003-A
Country	US
Kind code	B2
Filing date	Mar 12, 2021
Priority date	Mar 12, 2021
Publication date	Mar 10, 2026
Grant date	Mar 10, 2026

Abstract

A method for simulating a random walk using spiking neuromorphic hardware is provided. The method comprises receiving, by a buffer count neuron, spiking inputs from upstream mesh nodes, wherein the inputs include information packets comprising information associated with a simulation of a random walk process. A buffer generator neuron generates spikes until the buffer count reaches a first predefined threshold, after which it sends buffer spiking outputs to a spike count neuron. The spike count neuron counts the buffer spiking outputs, and a spike generator neuron generates spikes until the spike count neuron reaches a second specified threshold. The spike generator neuron then sends counter spiking outputs to a probability neuron, which selects downstream mesh nodes to receive the counter spiking outputs, wherein the spiking outputs include updated information packets. The probability neuron then sends the spiking outputs to the selected downstream nodes.

First claim

What is claimed is:

1. A computer-implemented method for simulating a random walk in spiking neuromorphic hardware, the method comprising:
using a number of processors to perform the steps of:
receiving, by a buffer count neuron in a mesh node, a number of spiking inputs from a number of upstream mesh nodes, wherein each spiking input comprises a tag that provides a full history of a trajectory from an origin of the spiking input and information associated with synchronizing spikes in a simulation of a specific random walk process;
generating, by a buffer generator neuron in the mesh node in response to the number of spiking inputs, a first number of the spikes until the buffer count neuron reaches a first predefined threshold;
upon reaching the first predefined threshold, sending, by the buffer generator neuron, a number of buffer spiking outputs to a spike count neuron in the mesh node;
counting, by the spike count neuron, the buffer spiking outputs from the buffer generator neuron;
generating, by a spike generator neuron in the mesh node in response to the buffer spiking outputs, a second number of the spikes until the spike count neuron reaches a second predefined threshold;
upon reaching the second predefined threshold, sending, by the spike generator neuron, a number of counter spiking outputs comprising the tag to a probability neuron in the mesh node, wherein the number of counter spiking outputs each include an information packet comprising updated information associated with a simulation of the specific random walk process;
compressing communication events and computations by selecting, by the probability neuron, a random number of downstream mesh nodes to receive the counter spiking outputs generated by the spike generator neuron; and
sending, by the probability neuron, the counter spiking outputs to the random number of downstream mesh nodes selected.

2. The method of claim 1 , further comprising:
activating the neuromorphic hardware by a supervisor spiking a set of spiking inputs to a number of start location mesh nodes; and
performing, by a circuit at each mesh node respectively, a specific computation that is local to the mesh node.

3. The method of claim 1 , wherein the buffer count neuron and the spike count neuron are further configured to produce an inhibitory output signal to stop activities of the buffer generator neuron and the spike generator neuron, respectively, when thresholds for the buffer generator neuron and the spike generator neuron are reached.

4. The method of claim 1 , further comprising;
the buffer count neuron receiving and synchronizing spiking inputs from multiple upstream mesh nodes; and
running the simulation asynchronously.

5. The method of claim 1 , wherein the downstream mesh nodes are selected using a probability tree.

6. The method of claim 1 , wherein the probability neuron is configured to select multiple downstream mesh nodes to send the number of counter spiking output generated by the spike generator neuron.

7. The method of claim 1 , wherein each information packet further comprises programmed instructions configured to track and regulate spiking activity in the mesh node.

8. A computer-implemented method for simulating a random walk in spiking neuromorphic hardware, the method comprising:
using a number of processors to perform the steps of:
receiving, by a tag circuit in a mesh node, a number of random walk information packets comprising a full history, respectively, of a trajectory of a spiking input from its origin and through a number of upstream mesh nodes, wherein the random walk information packets each comprise information synchronizing spiking activity of a random walk simulation;
receiving, by a buffer count neuron in the mesh node, a number of spiking inputs from the upstream mesh nodes, wherein each of the random walk information packets is associated with a spiking input;
generating, by a buffer generator neuron in the mesh node in response to the number of spiking inputs, a first number of spikes until the buffer count neuron reaches a first predefined threshold;
upon reaching the first predefined threshold, sending, by the buffer generator neuron, a number of buffer spiking outputs to a spike count neuron;
counting, by a spike count neuron in the mesh node, the buffer spiking outputs from the buffer generator neuron;
generating, by a spike generator neuron in the mesh node in response to the buffer spiking outputs, a second number of spikes until the spike count neuron reaches a second predefined threshold;
upon reaching the second predefined threshold, sending, by the spike generator neuron, a number of counter spiking outputs to a probability neuron in the mesh node;
compressing communication events and computations by selecting, by the probability neuron, a random number of downstream mesh nodes to receive the counter spiking outputs generated by the spike generator neuron;
sending, by the probability neuron, the counter spiking outputs to the random number of downstream mesh nodes selected; and
sending, by the tag circuit, the random walk information packets to the random number of downstream mesh nodes selected by the probability neuron, wherein the random walk information packets are updated with spiking activity of the mesh node.

9. The method of claim 8 , further comprising:
activating the spiking neuromorphic hardware by a supervisor spiking a set of spiking inputs to a number of start location mesh nodes; and
performing, by a circuit at each mesh node respectively, a specific computation that is local to the mesh node.

10. The method of claim 8 , wherein the buffer count neuron and the spike count neuron are further configured to produce an inhibitory output signal to stop the spiking activities of the buffer generator neuron and the spike generator neuron, respectively, when thresholds for the buffer generator neuron and the spike generator neuron are reached.

11. The method of claim 8 , further comprising running a random walk simulation asynchronously.

12. The method of claim 8 , further comprising selecting the random number of downstream mesh nodes using a probability tree.

13. The method of claim 8 , wherein the probability neuron is configured to select multiple downstream mesh nodes to send the number of counter spiking output outputs generated by the spike generator neuron.

14. The method of claim 8 , wherein the tag circuit is configured to allow the neuromorphic hardware to run multiple variants of a random walk simulation simultaneously.

15. A spiking neuromorphic mesh node configured to execute a random walk simulation, wherein the spiking neuromorphic mesh node comprises:
a tag circuit configured to transmit information packets between mesh nodes, wherein each information packet comprises a full history of a trajectory of a spiking input from its origin and information associated with synchronizing spiking activity of the random walk simulation;
a buffer circuit, that comprises:
a buffer count neuron configured to receive a number of spiking inputs from a number of upstream mesh nodes;
a buffer generator neuron configured to generate a first number of spikes until the buffer count neuron reaches a first predefined threshold and, upon reaching the first predefined threshold, generate and synchronize a number of buffer spiking outputs;

a spike circuit, that comprises:
a spike count neuron configured to receive and count the buffer spiking outputs from the buffer generator neuron;
a spike generator neuron configured to generate a second number of spikes until the spike count neuron reaches

Assignees

• Nat Tech & Eng Solutions Sandia Llc

Inventors

• Aimone James Bradley
• Severa William Mark
• Smith John Darby

Classifications

• G06N3/049 Primary

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

• G06N3/063

  using electronic means · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N3/047

  Probabilistic or stochastic networks · CPC title

• Y02D10/00

  Energy efficient computing, e.g. low power processors, power management or thermal management · CPC title