Biological neural network system and methods
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for artificial neural networks (ANNs) in complex tasks is hindered by high time, energy, and data requirements for training, necessitating more efficient computational methods.
Innovation Solution
A biological and artificial neural network (BANN) system utilizing a multi-electrode array (MEA) with biological neural networks (BNNs) and trained statistical models to encode and decode input signals, reducing the power required for training and improving computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If artificial neural networks are used to solve complex tasks, then task solving capability is improved, but training time and energy consumption increase
Solution Approach 1:
The patent introduces biological neural networks as an intermediary component between traditional ANNs and the task environment. The BNN performs preliminary processing of input data through biologically-inspired mechanisms, reducing the computational burden on the ANN and thereby lowering training energy consumption while maintaining task solving capability
Solution Approach 2:
The patent replaces purely computational mechanical systems (traditional ANN processing) with a hybrid system incorporating biological neural networks. The BNN uses biologically-based processing mechanisms that are more energy-efficient for certain types of pattern recognition and feature extraction, substituting high-energy computational operations with lower-energy biological processing
2Adaptability or versatility
If artificial neural networks are used to solve complex tasks, then task solving capability is improved, but training time increases
Solution Approach 1:
The biological neural network performs preliminary processing of input data before it reaches the ANN for full training. By pre-processing features and patterns using biologically-inspired mechanisms, the system reduces the amount of work the ANN must do during training, thereby reducing training time while maintaining task solving capability
Solution Approach 2:
The patent segments the neural processing task into two parts: the BNN handles initial feature extraction and pattern recognition using biologically-inspired mechanisms, while the ANN focuses on higher-level task-specific learning. This segmentation allows parallel processing and reduces the overall training time required
3Measurement precision
If traditional ANNs are used for data processing, then computational accuracy is achieved, but power consumption is high
Solution Approach 1:
The BNN serves as an intermediary that performs energy-efficient preliminary processing using biologically-inspired mechanisms. This reduces the computational load on the power-consuming ANN components while preserving the accuracy needed for task completion, thereby reducing overall power consumption without sacrificing computational accuracy
Data Source
AI summary
Techniques for calibrating a system comprising a multi-electrode array (MEA); a biological neural network (BNN) comprising neurons arranged on the MEA, and a processor. The method comprises using the system to select a subset of a electrodes of the MEA by stimulating the BNN by using the electrodes of the MEA to generate electrical signals in accordance with a calibration stimulation pattern; measuring a response of the BNN to the stimulating; selecting, based on the measured response of the BNN, the subset of electrodes based on an amount of neuronal activity induced by respective ones of the electrodes; receiving an input signal to be processed by the BNN; encoding the input signal to generate a stimulation pattern for stimulating the BNN; and stimulating the BNN using only the selected subset of the electrodes to generate electrical signals in accordance with the stimulation pattern.