Deep Neural Network Category Identification via Auto-Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep neural networks fail to identify data category information during data identification due to the absence of category information in unsupervised auto-encoding models used in their training process.
Innovation Solution
A method is introduced to establish a deep neural network by integrating data category information, using a linear category analysis function and an optimization function that combines unsupervised auto-encoding and linear category analysis, allowing the network to identify data categories through a multi-layer structure with input and output layers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an unsupervised auto-encoding model is used to train the deep neural network, then the reconstruction error is minimized, but the network cannot acquire category information corresponding to input sample data
Solution Approach 1:
The patent combines the unsupervised auto-encoding model with a supervised classification model into a unified deep neural network. The auto-encoding layers minimize reconstruction error while the classification layers acquire category information, allowing both objectives to be achieved simultaneously through joint training with a combined loss function.
Solution Approach 2:
The deep neural network is designed to perform multiple functions: feature extraction through auto-encoding for minimizing reconstruction error, and category classification through supervised learning. This multi-functional architecture allows the network to both reconstruct input data and identify category information from the same trained model.
2Adaptability or versatility
If a deep neural network with many layers is constructed, then feature extraction capability is improved, but the network structure becomes more complex
Solution Approach 1:
The deep neural network is segmented into distinct functional modules: unsupervised auto-encoding layers for feature extraction and reconstruction, and supervised classification layers for category identification. This segmentation allows each module to be optimized independently while maintaining overall network effectiveness.
Solution Approach 2:
The auto-encoding layers are trained first to perform preliminary feature extraction and dimensionality reduction on the input data. This preliminary action simplifies the subsequent classification task, allowing the classification layers to focus on category identification without dealing with raw high-dimensional input features.
Data Source
AI summary
A deep neural network to which data category information is added is established locally, to-be-identified data is input to an input layer of the deep neural network generated based on the foregoing data category information, and information of a category to which the to-be-identified data belongs is acquired, where the information of the category is output by an output layer of the deep neural network. A deep neural network is established based on data category information, such that category information of to-be-identified data is conveniently and rapidly obtained using the deep neural network, thereby implementing a category identification function of the deep neural network, and facilitating discovery of an underlying law of the to-be-identified data according to the category information of the to-be-identified data.