Neural Network Group Switching for Multi-Setting Accuracy Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Neural networks face accuracy loss and catastrophic memory issues when operating in varying environments due to increased training data sets, leading to suboptimal performance in different settings.
Innovation Solution
A system that collects and stores training data for specific settings, allowing neural networks to adapt by switching between pre-trained coefficients and structures based on monitored environmental characteristics, using a database management system to retrieve and apply the appropriate settings dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If training data sets are increased to include data from many different settings, then the neural network can operate in more different settings, but the neural network loses accuracy and encounters catastrophic memory loss
Solution Approach 1:
The patent segments the training data into multiple separate training data sets, each corresponding to a specific setting or environment. Instead of using one large diverse training set, the system creates specialized training sets for different conditions (e.g., different lighting, temperatures, locations). This allows the neural network to maintain high accuracy for each specific setting while collectively covering multiple settings through segmentation of the overall training task.
2Adaptability or versatility
If training data sets are increased to include data from many different settings, then the neural network can operate in more different settings, but catastrophic memory loss occurs
Solution Approach 1:
The patent segments the training process into multiple separate training runs, each using a specialized training data set for a specific setting. This prevents the neural network from attempting to learn all settings simultaneously in one large training process, which causes catastrophic memory loss. By segmenting the training into manageable portions corresponding to specific settings, the system avoids memory overload while maintaining adaptability across multiple settings.
3Adaptability or versatility
If a single neural network is trained with diverse data, then it can handle multiple settings, but it cannot maintain optimal performance in any single setting
Solution Approach 1:
The patent segments the neural network system into multiple specialized neural networks, each trained on a specific training data set corresponding to a particular setting. Instead of one general-purpose network that compromises accuracy, the system creates multiple specialized networks that each excel at their specific setting. The system then selects or switches between these segmented networks based on the current operational context, maintaining optimal performance for each setting.
Solution Approach 2:
The patent creates a universal neural network system that encompasses multiple specialized networks. Each individual network is highly specialized for one setting, but the overall system is universal in its ability to handle multiple settings by selecting the appropriate specialized network. This multi-functionality is achieved through a selection mechanism that routes inputs to the most appropriate specialized network based on the current setting characteristics.
Data Source
AI summary
Methods and systems for controlling an autonomous machine. The autonomous machine has sensors generating input data, while the controller includes two or more neural networks that inference using the input data and generate output data. The neural networks can be trained using an identical set of training data set. The output data from each of the neural networks are monitored to ensure that the integrity of the operation is maintained by, for example, the output data from one neural network is compared with the output data from another neural network to verify the consistency. If the comparison yields that the integrity of the system is not maintained at an acceptable level, the controller can stop using the output in controlling the autonomous machine.