System, method, and computer device for transistor-based neural networks

A system for operating an artificial neural network (ANN), the system comprising neurons, wherein each neuron comprises:

• a plurality of synapses comprising charge-trapped transistors (CTTs) for processing input signals transformed into time pulses, the CTTs supplying synaptic weights;
• an accumulation block for receiving drain currents from the plurality of synapses, each drain current produced as an output of multiplication from a synapse and each drain current producing a quantity of charge equal to a product of a time pulse received at the synapse and a synaptic weight;
• a capacitor for accumulating charge from the drain currents to act as short-term memory for accumulated signals;
• a discharge pulse generator for generating an output signal by discharging the accumulated charge during a discharging cycle;
• and
• a comparator for comparing the output signal as an input voltage with a reference voltage, wherein the comparator produces a first output if the input voltage is above the reference voltage, and wherein the comparator produces a second output if the input voltage is below the reference voltage, and wherein the comparator is a threshold inverter quantization (TIQ) comparator comprising a cascade of at least one complementary metal oxide semiconductor (CMOS) inverter, each CMOS inverter comprising a p-channel metal oxide semiconductor (PMOS) transistor and an n-channel metal oxide semiconductor (NMOS) transistor, wherein the reference voltage corresponds to a threshold of the TIQ comparator self-generated by the TIQ comparator, wherein the threshold of the TIQ comparator is adjustable via the PMOS transistor or the NMOS transistor, and wherein the threshold of the TIQ comparator corresponds to a ratio of a strength of the PMOS transistor divided by a strength of the NMOS transistor.