Multi-Channel CNN Text Classification Model Architecture
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional text classification methods require heavy manual workload and are not accurate, especially with the limitations of existing topic models like Latent Dirichlet Allocation (LDA).
Innovation Solution
A text classification method using a convolutional neural network (CNN) model with multiple channels, each channel corresponding to a sub-text classification model, which calculates and integrates sub-classification parameters to improve classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual text classification is used, then flexibility in handling diverse text types is maintained, but workload increases and accuracy decreases
Solution Approach 1:
The patent replaces manual mechanical classification with an automated CNN-based text classification system. The model automatically processes text through convolutional layers, pooling layers, and fully connected layers to generate classification results, eliminating manual workload while maintaining high accuracy through learned feature representations.
Solution Approach 2:
The patent transforms text data into numerical vectors and processes them through multiple transformation stages (embedding, convolution, pooling, normalization). These parameter transformations enable the system to handle diverse text types automatically while achieving high classification accuracy through learned transformations.
2Extent of automation
If conventional topic models like LDA are used, then automation is achieved, but classification accuracy remains insufficient
Solution Approach 1:
The patent segments the text classification process into multiple specialized components: embedding layer for feature extraction, convolutional layers for local pattern recognition, pooling layers for feature aggregation, and fully connected layers for final classification. This segmented architecture enables automated processing while achieving high accuracy through specialized feature extraction at each stage.
Solution Approach 2:
The patent transforms text from one-dimensional sequences into multi-dimensional feature spaces through embedding layers and convolutional operations. This dimensional transformation enables the model to capture complex relationships and patterns that conventional topic models miss, achieving higher accuracy while maintaining full automation.
3Device complexity
If a single-channel text classification model is used, then model complexity is reduced, but classification accuracy is limited
Solution Approach 1:
The patent divides the classification task into multiple parallel channels, each specializing in different text patterns or features. Each channel processes text independently through its own convolutional and pooling layers, then results are aggregated. This segmentation enables the model to capture diverse features without excessive complexity in any single pathway.
Solution Approach 2:
The patent combines multiple classification channels into a composite model architecture, where each channel contributes specialized feature detection capabilities. The final classification integrates results from all channels, creating a robust system that achieves high accuracy while managing complexity through modular design.
Data Source
AI summary
This application relates to a text classification method. The method includes obtaining, by a computer device, a to-be-classified text, and calculating an original text vector corresponding to the text; determining, by the computer device according to the original text vector, an input text vector corresponding to each channel of a trained text classification model; inputting, by the computer device, the input text vector corresponding to each channel into a convolution layer of the corresponding channel of the trained text classification model, the trained text classification model comprising a plurality of channels, each channel being corresponding to a sub-text classification model, and the trained text classification model being used for determining a classification result according to a sub-classification parameter outputted by each sub-text classification model; and obtaining, by the computer device, a classification result outputted by the trained text classification model, and classifying the text according to the classification result.