Application of Multilayer Perceptron Genetic Algorithm Neural Network in Chinese-English Parallel Corpus Noise...

Area of Science:

• Computational Linguistics
• Artificial Intelligence
• Machine Learning

Background:

• Noise processing in Chinese-English parallel corpora is crucial for data quality.
• Standard genetic algorithms and neural networks have limitations in noise identification and training speed.

Purpose of the Study:

• To propose a fast algorithm for training neural networks using genetic algorithms for noise processing.
• To enhance the accuracy and efficiency of noise identification in Chinese-English parallel corpora.

Main Methods:

• Utilized a neural network as a predictive model and a genetic algorithm for online optimization.
• Implemented reinforcement learning with varied reward mechanisms and decoding strategies.
• Employed genetic operations and survival of the fittest for guided learning and search.

Main Results:

• The proposed genetic algorithm-based neural network method demonstrates rapid learning of network weights.
• Achieved superior performance across all aspects compared to standard genetic algorithms and neural networks.
• Showcased high recognition rates and unique application advantages, optimizing both time and efficiency.

Conclusions:

• The developed algorithm offers a significant advancement in noise processing for parallel corpora.
• It provides a robust and efficient solution for identifying isolated words and improving data quality.
• The method achieves a win-win scenario for processing time and recognition efficiency.