Excitable Semiconductor Material for STDP Timing Evaluation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current neuromorphic hardware implementing spike-timing dependent plasticity (STDP) in artificial synapses requires precise timing of electrical pulses, which is challenging due to system delays and energy inefficiencies, especially when aiming for high processing speed and throughput, and is further complicated by the need for external electronics that occupy significant space in large neural networks.

Innovation Solution

A method using an excitable semiconductor material that changes conductivity over time, allowing for a temporal evaluation of the interval between excitations, enabling a gradual STDP behavior without the need for complex timing electronics, by varying the voltage threshold over time, thus influencing the resistance of artificial synapses based on the timing of electrical signals.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If precise timing of electrical pulses is implemented in neuromorphic hardware, then STDP behavior can be achieved, but system complexity increases due to required external electronics

Engineering Contradiction:
ImproveSTDP behavior implementationVSAvoidexternal electronics
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The excitable material inherently performs the timing function through its biophysical properties. The material's excitability changes over time automatically, eliminating the need for external timing electronics. The system uses its own material properties to achieve the timing-dependent plasticity behavior.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces electronic timing mechanisms with a biophysical material system. Instead of using electronic circuits to control pulse timing, the invention uses the natural time-dependent excitability characteristics of the excitable material to achieve timing-dependent synaptic plasticity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If external electronics are used for pulse timing, then timing precision can be achieved, but space consumption increases in large neural networks

Engineering Contradiction:
Improvetiming precisionVSAvoidspace consumption
Core Design Contradiction:
Measurement precisionVSArea of stationary object

Solution Approach 1:

The excitable material inherently performs the timing function through its biophysical properties. The material's excitability changes over time automatically, eliminating the need for external timing electronics. The system uses its own material properties to achieve the timing-dependent plasticity behavior.

Inventive Principle:
Principle #25Self-service

3Reliability

If traditional pulse superposition method is used, then STDP can be implemented, but energy efficiency decreases due to required pulse coordination

Engineering Contradiction:
ImproveSTDP implementationVSAvoidenergy efficiency
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent replaces electronic timing mechanisms with a biophysical material system. Instead of using electronic circuits to control pulse timing, the invention uses the natural time-dependent excitability characteristics of the excitable material to achieve timing-dependent synaptic plasticity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The invention changes the excitability parameter of the material over time. The material's ability to respond to pulses naturally decays according to its biophysical properties, creating the time window for STDP without requiring active control. This parameter change is inherent to the material rather than externally imposed.

Inventive Principle:
Principle #35Parameter changes

4Measurement precision

If pulse timing is coordinated externally, then correlation detection can be achieved, but processing speed decreases due to system delays

Engineering Contradiction:
Improvecorrelation detectionVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSSpeed

Solution Approach 1:

The excitable material inherently performs the timing function through its biophysical properties. The material's excitability changes over time automatically, eliminating the need for external timing electronics. The system uses its own material properties to achieve the timing-dependent plasticity behavior.

Inventive Principle:
Principle #25Self-service

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach allows for efficient and space-saving implementation of STDP behavior in artificial neural networks, enhancing processing speed and data throughput by dynamically adjusting synaptic weights based on the timing of signals without requiring extensive external timing mechanisms.

Implementation Method 1

an excitable semiconductor material which can be brought at least partially into an excited amorphous state by a first excitation, with the electrical excitability of the semiconductor material in the excited state changing over time

Methodology Applied
Scientific EffectTime-dependent conductivity change:

Implementation Method 2

a first excitation is applied to an excitable semiconductor material in order to bring this material into an excited amorphous state

Methodology Applied
Scientific EffectPhase transition: Phase Change

Data Source

PatentEP3103069B1Device for the evaluation of a time-interval between two excitations
Publication Date: 2020.07.29 SALINGA MARTIN
  • EP3103069B1 patent drawingFigure 1
  • EP3103069B1 patent drawingFigure 2
  • EP3103069B1 patent drawingFigure 3

AI summary

The invention relates to an apparatus for analyzing a time interval between two excitations, comprising at least one excitable material which can be transformed at least in part into an excited state by an excitation, the excitability of the material changing over time in said excited state. The invention further relates to a method for analyzing a time interval between two excitations, the use of an apparatus according to the invention, and an artificial neural network comprising at least one apparatus according to the invention.