Programming model for a bayesian neuromorphic compiler

1. A system for performing probabilistic computations on mobile platform sensor data, the system comprising:

one or more processors and a non-transitory computer-readable medium having executable instructions encoded thereon such that when executed, the one or more processors perform operations of:

translating a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model;

using the spiking neuronal network unit, computing conditional probabilities for the input mobile platform sensor data, wherein the input mobile platform sensor data is a time series of mobile platform error codes encoded as neuronal spikes,

wherein the spiking neuronal network unit comprises a plurality of neurons A, B, C, D, and E having the following dynamics:

a tonic input causes neuron A to spike, and neuron A causes neuron B to spike, resulting in an increase to a synaptic weight w,

neuron B causes neuron C to spike twice, once through neuron D and once through neuron E, and

when neuron C spikes, then a delay τ 1 will cause neuron B to spike before neuron A, resulting in a decrease to the synaptic weight w, wherein the synaptic weight w decreases in proportion to itself;

decoding the neuronal spikes, wherein the decoded neuronal spikes represent at least one mobile platform failure mode; and

causing the mobile platform to initiate a mitigation action based on the at least one mobile platform failure mode.

2. The system as set forth in claim 1 , wherein the Bayesian model represents mobile platform error codes, mobile platform failure modes, and connections between the mobile platform error codes and the mobile platform.

3. The system as set forth in claim 1 , wherein in order to compute a conditional probability of two input processes X and Y, inputs to a phasic input and the tonic input of the spiking neuronal network unit are defined, wherein the phasic input comes from a logical conjunction of the two input processes X and Y, and the tonic input comes from the input process X, wherein the tonic input corresponds to P(X), and the phasic input corresponds to P(X,Y), resulting in the synaptic weight w converging to P(X,Y)/P(X)=P(Y|X), where P denotes probability, / denotes division, and | denotes a conditional relationship.

4. The system as set forth in claim 1 , wherein the spiking neuronal network unit is decomposable into a convergence unit CON, which causes the synaptic weight w to exponentially decay to a value, and a multiplier unit M, which acts as a logical conjunction, wherein the convergence unit CON and the multiplier unit M correspond to functions CON: X×Y→Z and M:X×Y→Z, respectively, where Z represents the function output, × denotes paired input arguments, and → denotes mapping from input to output.

5. A computer implemented method for performing probabilistic computations on mobile platform sensor data, the method comprising acts of:

causing one or more processers to execute instructions encoded on a non-transitory computer-readable medium, such that upon execution, the one or more processors perform operations of:

translating a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model;

using the spiking neuronal network unit, computing conditional probabilities for the input mobile platform sensor data, wherein the input mobile platform sensor data is a time series of vehicle error codes encoded as neuronal spikes,

wherein the spiking neuronal network unit comprises a plurality of neurons A, B, C, D, and E having the following dynamics:

a tonic input causes neuron A to spike, and neuron A causes neuron B to spike, resulting in an increase to a synaptic weight w,

neuron B causes neuron C to spike twice, once through neuron D and once through neuron E, and

when neuron C spikes, then a delay τ 1 will cause neuron B to spike before neuron A, resulting in a decrease to the synaptic weight w, wherein the synaptic weight w decreases in proportion to itself;

decoding the neuronal spikes, wherein the decoded neuronal spikes represent at least one mobile platform failure mode; and

causing the mobile platform to initiate a mitigation action based on the at least one mobile platform failure mode.

6. The method as set forth in claim 5 , wherein the Bayesian model represents mobile platform error codes, mobile platform failure modes, and connections between the mobile platform error codes and the mobile platform.

7. The method as set forth in claim 5 , wherein in order to compute a conditional probability of two input processes X and Y, inputs to a phasic input and the tonic input of the spiking neuronal network unit are defined, wherein the phasic input comes from a logical conjunction of the two input processes X and Y, and the tonic input comes from the input process X, wherein the tonic input corresponds to P(X), and the phasic input corresponds to P(X,Y), resulting in the synaptic weight w converging to P(X,Y)/P(X)=P(Y|X), where P denotes probability, / denotes division, and | denotes a conditional relationship.

8. The method as set forth in claim 5 , wherein the spiking neuronal network unit is decomposable into a convergence unit CON, which causes the synaptic weight w to exponentially decay to a value, and a multiplier unit M, which acts as a logical conjunction, wherein the convergence unit CON and the multiplier unit M correspond to functions CON: X×Y→Z and M:X×Y→Z, respectively, where Z represents the function output, × denotes paired input arguments, and → denotes mapping from input to output.

9. A computer program product for performing probabilistic computations on mobile platform sensor data, the computer program product comprising:

computer-readable instructions stored on a non-transitory computer-readable medium that are executable by a computer having one or more processors for causing the processor to perform operations of:

translating a Bayesian model representing input mobile platform sensor data to a spiking neuronal network unit that implements the Bayesian model;

using the spiking neuronal network unit, computing conditional probabilities for the input mobile platform sensor data, wherein the input mobile platform sensor data is a time series of mobile platform error codes encoded as neuronal spikes,

wherein the spiking neuronal network unit comprises a plurality of neurons A, B, C, D, and E having the following dynamics:

a tonic input causes neuron A to spike, and neuron A causes neuron B to spike, resulting in an increase to a synaptic weight w,

neuron B causes neuron C to spike twice, once through neuron D and once through neuron E, and

when neuron C spikes, then a delay τ 1 will cause neuron B to spike before neuron A, resulting in a decrease to the synaptic weight w, wherein the synaptic weight w decreases in proportion to itself;

decoding the neuronal spikes, wherein the decoded neuronal spikes represent at least one mobile platform failure mode; and

causing the mobile platform to initiate a mitigation action based on the at least one mobile platform failure mode.

10. The computer program product as set forth in claim 9 , wherein the Bayesian model represents mobile platform error codes, mobile platform failure modes, and connections between the mobile platform error codes and the mobile platform.

11. The computer program product as set forth in claim 9 , wherein in order to compute a conditional probability of two input processes X and Y, inputs to a phasic input and the tonic input of the spiking neuronal network unit are defined, wherein the phasic input comes from a logical conjunction of the two input processes X and Y, and the tonic input comes from the input process X, wherein the tonic input corresponds to P(X), and the phasic input corresponds to P(X,Y), resulting in the synaptic weight w converging to P(X,Y)/P(X)=P(Y|X), where P denotes probability, / denotes division, and | denotes a conditional relationship.

12. The computer program product as set forth in claim 9 , wherein the spiking neuronal network unit is decomposable into a convergence unit CON, which causes the synaptic weight w to exponentially decay to a value, and a multiplier unit M, which acts as a logical conjunction, wherein the convergence unit CON and the multiplier unit M correspond to functions CON: X×Y→Z and M:X×Y→Z, respectively, where Z represents the function output, × denotes paired input arguments, and → denotes mapping from input to output.