US 10,891,536 B1 · Artificial neural network for reservoir computing using stochastic logic

Abstract

An artificial neuron includes a signal mixer that combines input signals to provide a first stochastic bit-stream as output and a stochastic activation function circuit configured to receive the first stochastic bit-stream from the signal mixer and to generate therefrom a second stochastic bit-stream. The first stochastic bit-stream is representative of a first output value. In the stochastic activation function circuit, n independent stochastic bit-streams, each representative of the first output value, are summed to provide a selection signal that is provided to a multiplexer to select between n+1 coefficient bit-streams and provide the second stochastic bit-stream. The activation function has a characteristic determined by the proportion of ones in each of the n+1 coefficients bit-streams. One or more artificial neurons may be used in an Artificial Neural Network, such as a Time Delay Reservoir network.

Claims16

Dependent claims

An artificial neuron comprising:

a signal mixer configured to mix together two or more input signals and provide a first stochastic bit-stream as output, where the first stochastic bit-stream is representative of a first output value;
and
a stochastic activation function circuit configured to receive the first stochastic bit-stream and to generate therefrom a second stochastic bit-stream, the stochastic activation function circuit comprising:
a circuit for generating n independent stochastic bit-streams each representative of the first output value;
an adder circuit configured to sum the n independent stochastic bit-streams and provide a selection signal as output;
and
a multiplexer responsive to the selection signal from the adder circuit and configured to select between n+1 coefficient bit-streams dependent upon the selection signal to provide the second stochastic bit-stream,
wherein the stochastic activation function has a characteristic determined by proportion of ones in each of the n+1 coefficient bit-streams.

5 more elements

An artificial neural network circuit responsive to one or more input signals to produce an output signal, the artificial neural network circuit comprising:

an input layer responsive to the one or more input signals to produce a first signal;
a reservoir layer configured to receive the first signal from the input layer and comprising:
a delay line that receives a second signal as input and provides a third signal as output;
a mixer for combining the first signal and the third signal to provide a first stochastic bit-stream as output;
a stochastic logic circuit responsive to the first stochastic bit-stream to produce a second stochastic bit-stream as output, where the stochastic logic circuit implements an activation function;
and
a stochastic-to-binary converter that converts the second stochastic bit-stream to the second signal;
and
an output layer configured to combine elements of the delay line with one or more output weights to produce the output signal.

6 more elements

Dependencies are inferred from the claim text.

Artificial neural network for reservoir computing using stochastic logic

Life of this patent

Artificial neural network for reservoir computing using stochastic logic

Life of this patent

Abstract

Claims16