Neural Network Learning Cells for Efficient Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep learning requires a large amount of high-quality training data and is time-consuming, with existing methods not adequately addressing the need for a simpler algorithm to reduce system load and improve recognition rates.
Innovation Solution
A learning device and method utilizing a neural network with learning cells that update weighting coefficients or add new cells based on output values, allowing for simpler algorithmic learning and high recognition rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning using multilayer neural network is used, then recognition rate is improved, but learning time and system load increase significantly
Solution Approach 1:
The patent divides the neural network into multiple independent learning cells, where each cell processes a subset of input data independently. This segmentation allows parallel processing and reduces the overall learning time while maintaining recognition accuracy, as each cell can be trained separately and contributes to the final recognition result.
Solution Approach 2:
The patent extracts and removes unnecessary computational components from traditional deep learning models. By using a simplified learning cell structure that only performs essential weighted summation and threshold comparison operations, the system eliminates complex multilayer transformations while achieving comparable recognition rates with reduced learning time.
2Measurement precision
If complex multilayer neural network is constructed, then recognition capability is improved, but device complexity and operational processing load increase
Solution Approach 1:
The neural network is segmented into multiple simple learning cells rather than using a complex multilayer structure. Each learning cell is an independent unit with simple weighted summation and threshold comparison operations, making the overall system less complex while maintaining recognition capability through parallel cell operations.
Solution Approach 2:
The patent changes the operational parameters of the neural network by using fixed threshold values and simple weighting schemes instead of complex adaptive parameters. This parameter simplification reduces device complexity and operational processing load while preserving recognition capability through the collective output of multiple learning cells.
3Measurement precision
If large amount of training data is used, then learning accuracy is improved, but learning time and computational resources increase
Solution Approach 1:
Training data is distributed across multiple learning cells, with each cell processing a portion of the data independently. This segmentation enables parallel training operations, improving learning efficiency while maintaining learning accuracy as each cell contributes to the overall recognition model through its independent learning process.
Data Source
AI summary
An example embodiment includes a neural network unit to which a plurality of element values based on learning target data are input, and a learning unit that trains the neural network unit. The neural network unit has a plurality of learning cells each including a plurality of input nodes that perform predetermined weighting on each of the plurality of element values and an output node that sums the plurality of weighted element values and outputs the sum, and in accordance with an output value of each of the learning cells, the learning unit updates weighting coefficients of the plurality of input nodes of each of the learning cells or adds a new learning cell to the neural network unit.