Genetically optimized neural network for college English teaching evaluation method

English has long been regarded as the universal language. Countries that were earlier reluctant to learn English have also changed their stand due to its global reach. The nonnative English speaker’s proficiency largely depends on the College English Teaching (CET) and its evaluation methods. Traditional teaching evaluation models failed to consider all the difficulties of language learning. This drawback has led to researchers focusing on integrating Artificial Intelligence (AI) techniques in CET evaluation. This work is in line with such an effort to contribute a novel Teaching Evaluation Method (TEM) that employs a Machine Learning (ML) algorithm. The objective of this work is to provide colleges and other institutions with a practical framework for evaluating the CET by modifying the existing models through contemporary pedagogical approaches, thereby helping the institutions meet the complex demands of teaching and collaborative research. The study proposes a novel ML model that combines Convolutional Neural Networks (CNN) with a Grey Correlation-Based Genetic Algorithm (GCBGA) to meet the challenges in the traditional TEM. The proposed TEM model fared better than the other models in terms of English Teaching Evaluation. The performance enhancement is attributed to the novel GCBGA that optimized the proposed CNN model by avoiding it from getting stuck in local minima. This new method makes it easier for CNN to consistently judge the quality of teaching because it gets rid of the subjectivity and unpredictability of the conventional methods.