A Framework for Text Classification Using Evolutionary Contiguous Convolutional Neural Network and Swarm Based Deep...

Area of Science:

• Natural Language Processing (NLP)
• Machine Learning
• Deep Learning

Background:

• Accurate text classification is crucial for various NLP applications.
• Deep learning models often outperform traditional machine learning techniques in pattern classification.
• Selecting appropriate deep learning architectures for text classification remains a significant challenge.

Purpose of the Study:

• To propose and evaluate novel deep learning models for enhanced text classification.
• To introduce the Evolutionary Contiguous Convolutional Neural Network (ECCNN) model.
• To introduce the Swarm Deep Neural Network (Swarm DNN) model.

Main Methods:

• Developed ECCNN by integrating Contiguous Convolutional Neural Networks (CCNN) with Differential Evolution (DE) for deeper data understanding.
• Developed Swarm DNN by combining Deep Neural Networks (DNN) with Particle Swarm Optimization (PSO).
• Validated both models on the BBC newsgroup and 20 newsgroup text datasets.

Main Results:

• Swarm DNN achieved a classification accuracy of 97.32% on the BBC newsgroup dataset and 87.99% on the 20 newsgroup dataset.
• ECCNN achieved a classification accuracy of 97.11% on the BBC newsgroup dataset and 88.76% on the 20 newsgroup dataset.
• The Swarm DNN model demonstrated slightly superior performance compared to ECCNN on the tested datasets.

Conclusions:

• The proposed Swarm DNN and ECCNN models offer effective approaches for text classification.
• Swarm DNN, in particular, shows high potential for achieving accurate text classification.
• These advanced deep learning techniques contribute to improving NLP task performance.