Neuromorphic Post-Synaptic Neuron Comparator Reset Mechanism

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Neuromorphic devices face malfunctions when quasi-trained synapses are not properly reset, leading to incorrect data pattern output during learning mode.

Innovation Solution

Incorporating post-synaptic neurons with comparators and integrators that utilize reference voltages to decide and reset quasi-learned synapses, allowing for precise classification of learning states and efficient reset processes through primary and secondary reset signals.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If quasi-trained synapses are used in learning mode, then learning efficiency is improved, but device reliability deteriorates due to malfunctions and incorrect data pattern output

Engineering Contradiction:

Improvelearning efficiencyVSAvoiddevice reliability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent applies preliminary action by introducing a reset mechanism that proactively identifies and resets quasi-trained synapses before they cause malfunctions. The comparator circuit continuously monitors synapse training states and triggers reset operations when quasi-trained synapses are detected, preventing them from interfering with learning mode operations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback through a comparator circuit that continuously compares synapse training states against reference thresholds. This feedback mechanism detects when synapses are in a quasi-trained state and generates reset signals to return them to initial states, ensuring reliable operation during learning mode.

Inventive Principle:

Principle #23Feedback

2Measurement precision

If multiple comparators are added to decide quasi-learned synapses, then measurement precision is improved, but device complexity increases

Engineering Contradiction:

Improvesynapse state detection precisionVSAvoidneuron structure complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the comparator function into multiple sub-comparators, each responsible for comparing against specific reference voltages (first sub reference voltage, second sub reference voltage, and main reference voltage). This segmented approach enables precise classification of synapse training states into distinct categories while maintaining manageable circuit complexity.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements local quality by assigning different reference voltages to different comparators based on their specific comparison tasks. Each comparator uses a locally optimized reference voltage appropriate for its function, enabling precise differentiation of synapse states without requiring a single complex comparison mechanism.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS10509999B2Neuromorphic device including post-synaptic neurons having a comparator for deciding quasi- learned synapses

Publication Date: 2019.12.17 SK HYNIX INC

AI summary

A neuromorphic device may include: a pre-synaptic neuron; a plurality of post-synaptic neurons; and a plurality of synapses electrically connected to the pre-synaptic neuron and electrically connected to the plurality of post-synaptic neurons. Each of the post-synaptic neurons may include: an integrator; a main comparator having a first input port connected to an output port of the integrator; and a first sub comparator having a first input port connected to the output port of the integrator.