Spike Neural Network Bias Control for Environmental Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Spike neural network circuits implemented in semiconductor devices are susceptible to errors due to environmental factors such as temperature and humidity changes, leading to discrepancies in operation results.
Innovation Solution
Incorporation of a self-correcting control circuit that adjusts bias voltages based on comparisons between designed and actual output signals, using control codes and bias voltage generation circuits to maintain accurate operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a spike neural network circuit is implemented using a semiconductor device, then the circuit can process data in a manner similar to biological neural networks, but the circuit is influenced by surrounding environment changes (temperature, humidity, power supply voltage) causing errors in operation results
Solution Approach 1:
The patent implements a self-correcting control circuit that generates control codes based on comparing actual output signals with reference values. This feedback mechanism continuously monitors and adjusts bias voltages to compensate for environmental variations, thereby maintaining reliable operation results while preserving the biological neural network processing capability
Solution Approach 2:
The patent dynamically adjusts bias voltages applied to synaptic circuits based on detected errors and generated control codes. By changing the bias voltage parameter in response to environmental conditions, the system maintains accurate operation results despite temperature, humidity, or power supply variations
2Productivity
If bias voltage is applied to synaptic circuits for operation, then the circuit can perform neural network computations, but environmental changes cause the actual output signal to differ from the designed output signal
Solution Approach 1:
The self-correcting control circuit compares the actual output signal from synaptic computations with a reference number and generates appropriate control codes. This feedback loop ensures that environmental variations do not compromise the precision of computation results while maintaining full neural network productivity
Solution Approach 2:
The control circuit generates control codes in advance based on the comparison between actual and reference output signals. These pre-generated control codes are then used to adjust bias voltages before subsequent computations, preventing accuracy degradation before it occurs
Data Source
AI summary
Disclosed is a spike neural network circuit according to an embodiment of the present disclosure, which includes a self-correcting control circuit that generates an input signal and a first control code, a bias voltage generation circuit that generates a first bias voltage based on the first control code, a synaptic circuit including a first synaptic column that performs an operation of the input signal and a first weight signal and generates a first operation signal, a neuron circuit including a first neuron that generates a first output signal based on a comparison of the first operation signal and a threshold voltage, and a spike comparison circuit that generates a first comparison signal corresponding to a difference between the first output signal and a reference number, and the self-correcting control circuit further generates a second control code for correcting the first bias voltage.