System and method for synthetic aperture radar target recognition utilizing spiking neuromorphic networks

1. A method for identifying a target in a synthetic aperture radar signal, the method comprising:

receiving phase history data of the synthetic aperture radar signal;

extracting a plurality of features from the phase history data;

encoding, by a spiking neural network, the features as a plurality of spiking signals;

computing, by a readout neural layer, a signal identifier based on the spiking signals in accordance with a mapping weight matrix, the readout neural layer being configured using a supervised learning rule; and

outputting the signal identifier from the readout neural layer.

2. The method of claim 1 , wherein the plurality of features comprise an amplitude of the synthetic aperture radar signal.

3. The method of claim 1 , wherein the plurality of features comprise an off-amplitude of the synthetic aperture radar signal.

4. The method of claim 1 , wherein the plurality of features comprise a positive phase of the synthetic aperture radar signal.

5. The method of claim 1 , wherein the plurality of features comprise a negative phase of the synthetic aperture radar signal.

6. The method of claim 1 , wherein the readout neural layer comprises a linear classifier.

7. The method of claim 1 , wherein the spiking neural network and the readout neural layer are implemented by a neuromorphic chip.

8. The method of claim 1 , further comprising computing average spiking rates from the spiking signals,

wherein the computing the signal identifier is computed based on the average spiking rates.

9. The method of claim 1 , wherein the spiking neural network comprises:

an input layer configured to receive the features;

an excitatory neuron layer; and

an inhibitory neuron layer,

the input layer being configured to supply spiking signals to the excitatory neuron layer and the inhibitory neuron layer,

the inhibitory neuron layer being configured to supply spiking signals to the inhibitory neuron layer and the excitatory neuron layer,

the excitatory neuron layer being configured to supply spiking signals to the inhibitory neuron layer and the excitatory neuron layer, and

the excitatory neuron layer comprising output neurons configured to output the spiking signals to the readout neural layer.

10. A system configured to identify a target in a synthetic aperture radar signal, the system comprising:

a feature extractor configured to receive phase history data of the synthetic aperture radar signal and to extract a plurality of features from the phase history data;

a spiking neural network configured to encode the features as a plurality of spiking signals;

a readout neural layer configured, using a supervised learning rule, to compute a signal identifier based on the spiking signals in accordance with a mapping weight matrix; and

an output configured to output the signal identifier.

11. The system of claim 10 , wherein the plurality of features comprise an amplitude of the synthetic aperture radar signal.

12. The system of claim 10 , wherein the plurality of features comprise an off-amplitude of the synthetic aperture radar signal.

13. The system of claim 10 , wherein the plurality of features comprise a positive phase of the synthetic aperture radar signal.

14. The system of claim 10 , wherein the plurality of features comprise a negative phase of the synthetic aperture radar signal.

15. The system of claim 10 , wherein the readout neural layer comprises a linear classifier.

16. The system of claim 10 , wherein the spiking neural network and the readout neural layer are implemented by a neuromorphic chip.

17. The system of claim 10 , wherein the readout neural layer is configured to compute average spiking rates from the spiking signals, and

wherein the readout neural layer is configured to compute the signal identifier based on the average spiking rates.

18. The system of claim 10 , wherein the spiking neural network comprises:

an input layer configured to receive the features;

an excitatory neuron layer; and

an inhibitory neuron layer,

the input layer being configured to supply spiking signals to the excitatory neuron layer and the inhibitory neuron layer,

the inhibitory neuron layer being configured to supply spiking signals to the inhibitory neuron layer and the excitatory neuron layer,

the excitatory neuron layer being configured to supply spiking signals to the inhibitory neuron layer and the excitatory neuron layer, and

the excitatory neuron layer comprising output neurons configured to output the spiking signals to the readout neural layer.

19. A method for training a system for identifying a target in a synthetic aperture radar signal, the system comprising a feature extractor, a spiking neural network, and a readout neural layer, the method comprising:

extracting, by a feature extractor, a plurality of features from phase history data from a plurality of labeled synthetic aperture radar training signals;

encoding, by a spiking neural network, the features as a plurality of spiking signals; and

training a readout neural layer and a mapping weight matrix to compute a signal identifier based on the plurality of spiking signals and the labeled synthetic aperture radar training signals using a supervised learning rule.

20. The method of claim 19 , wherein the spiking neural network and the readout neural layer are implemented by a neuromorphic chip.

21. The method of claim 19 , wherein the plurality of features comprise an amplitude and an off-amplitude of the synthetic aperture radar signal.

22. The method of claim 19 , wherein the plurality of features comprise a positive phase and a negative phase of the synthetic aperture radar signal.

23. The method of claim 19 , wherein the spiking neural network is configured in accordance with a plurality of parameters, the parameters being computed in accordance with a parameter search process and based on the labeled synthetic aperture radar training signals.

24. The method of claim 23 , wherein the parameter search process comprises:

initializing a set of parameters;

computing a set of features based on the labeled synthetic aperture radar training signals;

configuring the spiking neural network based on the set of parameters;

generating, by the spiking neural network, a plurality of spiking sequences from the set of features;

training a readout layer based on the spiking sequences and a plurality of labels of the labeled synthetic aperture radar training signals;

computing a classification rate;

determining that the classification rate fails to exceed a threshold; and

updating the parameters based on the classification rate.

25. A method for training a system for identifying a target in a synthetic aperture radar signal, the system comprising a feature extractor, a spiking neural network, and a readout neural layer, the method comprising:

extracting, by a feature extractor, a plurality of features from a plurality of labeled synthetic aperture radar training signals without synthesizing an image from the synthetic aperture radar signal;

encoding, by a spiking neural network, the features as a plurality of spiking signals, the spiking neural network being configured using parameter search; and

training a readout neural layer and a mapping weight matrix to compute a signal identifier based on the plurality of spiking signals and the labeled synthetic aperture radar training signals using a supervised learning rule,

wherein the readout neural layer comprises a linear classifier, the linear classifier comprising a plurality of classifier weights, and

wherein the training the readout neural layer comprises applying a rank-1 learning rule to compute the classifier weights.