gaussic/text-classification-cnn-rnn
Text Classification Cnn Rnn
This project is a TensorFlow-based supervised text categorizer designed for Chinese natural language processing. It utilizes a hybrid neural network architecture that combines convolutional and recurrent layers to map raw Chinese text to predefined categories.
The system integrates convolutional neural networks for local feature extraction and recurrent neural networks for analyzing sequential dependencies. It employs character-level tokenization and word embeddings to represent text as numerical tensors.
The implementation covers the end-to-end machine learning pipeline, including text preprocessing, iterative model training with dropout regularization, and predictive performance measurement using loss and accuracy metrics. It also includes utilities for tracking training progress and maintaining model state through checkpoints.
Features
- Chinese Text Classifiers - Provides a specialized machine learning tool for assigning categories to Chinese text documents.
- Convolutional Neural Networks - Uses convolutional neural networks to extract local features and n-gram patterns from text.
- Hybrid CNN-RNN Architectures - Employs a hybrid CNN-RNN architecture to analyze both local features and long-term sequential dependencies.
- Recurrent Neural Networks - Implements text classification using recurrent neural networks to capture long-term sequential dependencies.
- Model Training Pipelines - Implements a complete training pipeline to optimize neural network weights for predictive accuracy.
- Recurrent Neural Networks - Utilizes recurrent neural networks to capture sequential dependencies and temporal patterns in text.
- Supervised Text Categorizers - Implements a supervised training and inference pipeline for predicting text labels.
- Text Classification Models - Implements a TensorFlow-based model for categorizing Chinese text documents.
- Text Preprocessing Pipelines - Provides a pipeline to convert raw Chinese text into fixed-length numerical sequences.
- Word Embeddings - Maps Chinese characters to numerical vectors for representation within the neural network.
- Text Classification Training - Trains convolutional neural networks specifically for the task of text document classification.
- Dropout Regularization - Employs dropout regularization to prevent overfitting and improve model generalization.
- Iterative Training - Implements iterative training loops to optimize model weights over multiple epochs.
- Model Performance Evaluators - Quantifies classifier quality using loss and accuracy metrics across training and validation sets.
- Model Prediction Evaluation - Calculates loss and accuracy metrics by comparing model predictions against ground truth labels.
- Model Training - Follows iterative procedures to train predictive models on prepared text datasets.
- Training Weight Adjustments - Updates internal connection strengths using gradient descent to minimize prediction error.
- TensorFlow Graph Execution - Defines the mathematical operations and layer connectivity using TensorFlow's computational graphs.
- Convolutional Text Features - Extracts hierarchical spatial patterns and local features from text using CNNs.
- LLM Category Assignments - Assigns predefined categories to input text using a trained neural network model.
- Text Classification - Implementation examples for news classification using neural networks.
