Text Classification Model Based on Long Short-Term Memory with L2 Regularization
Ning Chen · 2024
Machine translation (MT) is the automatic, nonhuman translation of one text into another within the field of computer languages. However, as people used a variety of texts, it is crucial to translate textual data through the languages in order to convey and express ideas. This research proposed the Long Short-Term Memory with L2 Regularization (LSTM-L2) framework, which effectively restores the text’s original significance in its context and facilitates text categorization. The corpus dataset is then utilized to gather the data, then stop word removal and tokenization is utilized to improve the quality of the raw data. The method of feature extraction involves the usage of Bag of Words (BOW) to extract features. Finally, the proposed LSTM-L2 regularization is used to the classification process to improve accuracy and classify the text in corrected form. According to the results it illustrates that the proposed LSTM-L2 regularization method achieves better performance accuracy of 95.32%, precision of 92.16%, Recall of 90.25% and F1-score of 95.36% when compared with the existing models such as Universal Language Model Fine Tuning (ULM -FiT), Neural Machine Translation (NMT) and Stochastic Gradient Descent (SGD).