Study on Optimizing BP Neural Network Based on Genetic Algorithm to Predict Wordle Report Results

• Published in: 2023 IEEE International Conference on Image Processing and Computer Applications (ICIPCA) pp. 403 - 408
• Main Author: Xu, Weiqian
• Format: Conference Proceeding
• Language: English
• Published: IEEE 11.08.2023
• Subjects: Computational modeling; Correlation; Data models; Genetic algorithm optimization of BP neural network; Image processing; mean absolute error; Neural networks; Prediction algorithms; Predictive models; root mean square error; Wordle
• DOI: 10.1109/ICIPCA59209.2023.10257918

To predict the distribution of reported results for a given word in Wordle difficulty mode, i.e., to predict the correlation percentage of (1, 2, 3, 4, 5, 6, X), this paper uses the number of repeated letters, the ranking of word usage, and the number of commonly used letters as the correlation attributes of words, and uses a data regression forecasting model of genetic algorithm optimized BP neural network to predict the distribution of reported results with root mean square error and mean absolute error as the evaluation criteria. Meanwhile, the model was used to compare with the BP neural network data regression prediction model without genetic algorithm optimization, and it was found that the optimization model has smaller root mean square error and mean absolute error, i.e., the optimized model has better accuracy. Finally, the word attributes associated with the word EERIE were brought into the genetic algorithm optimized BP neural network data prediction model, and the distribution of the reported results was (0%, 2%, 11%, 24%, 29%, 26%, and 8%).