Anomaly detection with spiking neural networks

Patent metadata

Field	Value
Publication number	US-11436475-B2
Application number	US-201916436744-A
Country	US
Kind code	B2
Filing date	Jun 10, 2019
Priority date	Jun 10, 2019
Publication date	Sep 6, 2022
Grant date	Sep 6, 2022

Abstract

Detecting anomalies with a spiking neural network is provided. An input layer receives a number of inputs and converts them into phase-coded spikes, wherein each input is contained within a number of progressively larger neighborhoods of surrounding inputs. From the phase-coded spikes, a median value of each input is computed for each size neighborhood. An absolute difference of each input from its median value is computed for each size neighborhood. A median absolute difference (MAD) of each input is computed for each size neighborhood. For each input, an adaptive median filter (AMF) determines if a MAD for any size neighborhood exceeds a respective threshold. If one or more neighborhoods exceeds its threshold, the AMF outputs the median value of the input for the smallest neighborhood. If none of the neighborhoods exceeds the threshold, the AMF outputs the original value of the input.

First claim

What is claimed is:

1. A computer-implemented method of anomaly detection, comprising:
receiving, by an input layer in a spiking neural network, a number of inputs, wherein each input is contained within a number of progressively larger neighborhoods of surrounding inputs;
converting, by the input layer, the inputs into phase-coded spikes;
computing from the phase-coded spikes, by a first median layer, a median value of each input for each size neighborhood;
computing, by an absolute difference layer, an absolute difference of each input from its median value for each size neighborhood;
computing from absolute differences, by a second median layer, a median absolute difference (MAD) value of each input for each size neighborhood;
determining for each input, by an adaptive median filter (AMF) layer, if a MAD value for any size neighborhood exceeds a respective threshold, wherein:
if a MAD value of one or more neighborhoods exceeds its threshold, outputting, by the AMF layer, the median value of the input for the smallest size neighborhood; or
if none of the MAD values for the neighborhoods exceeds the threshold, outputting, by the AMF layer, the original value of the input received by the input layer.

2. The method of claim 1 , wherein if the MAD value for more than one size neighborhood exceeds the threshold, the AMF layer outputs the median value of the input for the smallest sized neighborhood exceeding the threshold.

3. The method of claim 1 , further comprising routing output values from the AMF layer back to the first median layer.

4. The method of claim 1 , wherein phase-coding represents a neuron spiking delay for a specified length of time to allow the AMF layer to process the inputs in ascending order.

5. The method of claim 1 , wherein the inputs represent image pixels, and wherein the neighborhoods of inputs represent symmetric matrices of adjacent pixels.

6. The method of claim 1 , wherein the inputs comprise a steam of data.

7. The method of claim 1 , wherein the spiking neural network comprises leaky integrate-and-fire neurons.

8. A computer program product for anomaly detection, the computer program product comprising:
a non-volatile computer readable storage medium having program instructions embodied therewith, the program instructions executable by a number of processors to implement a spiking neural network to perform the steps of:
receiving, by an input layer in a spiking neural network, a number of inputs, wherein each input is contained within a number of progressively larger neighborhoods of surrounding inputs;
converting, by the input layer, the inputs into phase-coded spikes;
computing from the phase-coded spikes, by a first median layer, a median value of each input for each size neighborhood;
computing, by an absolute difference layer, an absolute difference of each input from its median value for each size neighborhood;
computing from absolute differences, by a second median layer, a median absolute difference (MAD) value of each input for each size neighborhood;
determining for each input, by an adaptive median filter (AMF) layer, if a MAD value for any size neighborhood exceeds a respective threshold, wherein:
if a MAD value of one or more neighborhoods exceeds its threshold, outputting, by the AMF layer, the median value of the input for the smallest size neighborhood; or
if none of the MAD values for the neighborhoods exceeds the threshold, outputting, by the AMF layer, the original value of the input received by the input layer.

9. The computer program product of claim 8 , wherein if the MAD value for more than one size neighborhood exceeds the threshold, the AMF layer outputs the median value of the input for the smallest sized neighborhood exceeding the threshold.

10. The computer program product of claim 8 , further comprising routing output values from the AMF layer back to the first median layer.

11. The computer program product of claim 8 , wherein phase-coding represents a neuron spiking delay for a specified length of time to allow the AMF layer to process the inputs in ascending order.

12. The computer program product of claim 8 , wherein the inputs represent image pixels, and wherein the neighborhoods of inputs represent symmetric matrices of adjacent pixels.

13. The computer program product of claim 8 , wherein the inputs comprise a steam of data.

14. The computer program product of claim 8 , wherein the spiking neural network comprises leaky integrate-and-fire neurons.

15. A spiking neural network for anomaly detection, the spiking neural network comprising:
an input layer configured to receive a number of inputs and convert the inputs into phase-coded spikes, wherein each input is contained within a number of progressively larger neighborhoods of surrounding inputs;
a first median layer configured to compute, from the phase-coded spikes, a median value of each input for each size neighborhood;
an absolute difference layer configured to compute an absolute difference of each input from its median value for each size neighborhood;
a second median layer configured to compute, from absolute differences, a median absolute difference (MAD) value of each input for each size neighborhood; and
an adaptive median filter (AMF) layer configured to determine for each input if a MAD value for any size neighborhood exceeds a respective threshold, wherein:
if a MAD value of one or more neighborhoods exceeds its threshold, the AMF layer outputs the median value of the input for the smallest size neighborhood; or
if none of the MAD values for the neighborhoods exceeds the threshold, the AMF layer outputs the original value of the input received by the input layer.

16. The spiking neural network of claim 15 , wherein if the MAD value for more than one size neighborhood exceeds the threshold, the AMF layer outputs the median value of the input for the smallest sized neighborhood exceeding the threshold.

17. The spiking neural network of claim 15 , wherein output values from the AMF layer are routed by back to the first median layer.

18. The spiking neural network of claim 15 , wherein phase-coding represents a neuron spiking delay for a specified length of time to allow the AMF layer to process the inputs in ascending order.

19. The spiking neural network of claim 15 , wherein the inputs represent image pixels, and wherein the neighborhoods of inputs represent symmetric matrices of adjacent pixels.

20. The spiking neural network of claim 15 , wherein the inputs comprise a steam of data.

21. The spiking neural network of claim 15 , wherein the spiking neural network comprises leaky integrate-and-fire neurons.

Assignees

• Nat Tech & Eng Solutions Sandia Llc

Inventors

• Verzi Stephen Joseph
• Vineyard Craig Michael
• Aimone James Bradley

Classifications

• G06N3/048

  Activation functions · CPC title

• G06N3/047

  Probabilistic or stochastic networks · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/049 Primary

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

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title