Method and apparatus for constructing a dynamic adaptive neural network array (DANNA)

A circuit element for use in a multi-dimensional artificial neural network array, the circuit element comprising a neuron/synapse select input for selecting one of a neuron function and a synapse function of the circuit element of the multi-dimensional array, the circuit element being addressably configurable and comprising:

• a special purpose computer processor comprising an artificial neural network configuration structure for addressably configuring the multi-dimensional array of addressably configurable circuit elements and an interface and control structure for connecting the addressably configured multi-dimensional array to an external process;
• the special purpose computer being responsive to input data from the external process for addressably configuring the multi-dimensional array;
• an accumulator electronic component shared between the selected neuron function and the selected synapse function of the addressably configurable circuit element,
• a comparator electronic component of the addressably configurable circuit element,
• a refractory period input to a counter electronic component of the addressably configurable multi-dimensional array, and
• a threshold input to the comparator electronic component,
• the accumulator electronic component accumulating charge from a plurality of selectable inputs for comparison at the comparator electronic component with the threshold input when the circuit element is addressably configured as a neuron, and,
• if a refractory period has lapsed and the accumulated charge is greater than the threshold input, the comparator electronic component causing a fire output of the circuit element addressably configured as a neuron of the addressably configured multi-dimensional artificial neural network array when the circuit element selectively performs a neuron function.

A method for constructing a neuromorphic network comprising at least one artificial neural network, the artificial neural network comprising an addressably reconfigurable multi-dimensional array of interconnected circuit elements responsive to an external process, each circuit element having similar electronic components to one another, the interconnected circuit elements addressably configured, under special purpose program control of a special purpose computer processor, as one of a neuron and a synapse,

• the special purpose computer processor comprising an artificial neural network configuration structure for addressably configuring the multi-dimensional array of addressably configurable circuit elements and an interface and control structure for connecting the addressably configured and connected multi-dimensional array of circuit elements to the external process;
• the special purpose computer being responsive to input data from the external process for addressably configuring the multi-dimensional array;
• at least one circuit element of the addressably configured multi-dimensional array addressably configured as an input neuron,
• at least one circuit element of the addressably configured multi-dimensional array addressably configured as an output neuron and
• under special purpose program control, one to multiple circuit elements of the addressably configurable multi-dimensional array addressably configured as a neuron or a synapse forming a pathway between the input neuron receiving input data from the external process and the output neuron outputting data to the external process.

A method for use in an addressably configurable circuit element of an artificial neural network, the circuit element having a neuron/synapse select input for selecting one of a neuron and synapse function of the addressably configurable circuit element,

• an accumulator electronic component of the addressably configurable circuit element capable of selectively functioning as one of a neuron and a synapse, and the neuron function comprising the same electronic components as the synapse function, the neuron function comprising:
• a comparator electronic component,
• a refractory period input to a counter electronic component and a threshold input to the comparator electronic component,
• the accumulator electronic component accumulating charge from a plurality of selectable inputs for comparison at the comparator electronic component with the threshold input, and,
• if a refractory period has lapsed and the accumulated charge is greater than the threshold input, the comparator electronic component causing a fire output of the circuit element when the circuit element selectively performs the neuron function.