Evaluating teaching quality in colleges using combination of artificial neural networks (ANNs) and black hole optimization (BHO)

Automated evaluation of educational quality can be an effective measure of the efficiency of teaching methods, enabling the determination of the best teaching approach based on the educational environment conditions. Machine learning techniques have been used in various research studies as tools for automated evaluation of teaching quality. However, the low accuracy of classical learning models in quantitative quality estimation remains a challenge. To address this issue, one can utilize optimization strategies for learning models or combine multiple learning models as an ensemble system. These aspects are studied in the current paper. In the proposed method, automated evaluation of English teaching quality is performed through two phases: ''identification of prominent quality evaluation indicators'' and ''quality estimation based on ensemble learning.'' In the first phase, the Black Hole Optimization (BHO) algorithm is used to determine the minimum number of indicators required for accurate quality estimation. In the second phase, the proposed method combines multiple artificial neural networks (ANNs) to predict education quality. In this ensemble model, the configuration and weight vector of each ANN are adjusted by the BHO algorithm to minimize training error. Then, an averaging strategy is employed to determine the final output of the ensemble system. The performance of the proposed method is evaluated using real English teaching data. Based on the results, the proposed method can achieve an average accuracy of 98.53 % in estimating English teaching quality, which represents a 2.19 % improvement compared to previous methods.