Meta-Learning Strategies for Engineering Education Reform: Leveraging Convolutional Neural Networks and Reinforcement Learning Perspectives

There are many barriers in the use of technological education like low inclusive learning education resources; better ways of educating students and outcome assessment technology -based teaching and learning approach also poses a challenge in education since traditional teaching designs meet the demands in a way that enhances poor students’ attendance in class as well as poor performance. Inasmuch as the conventional approaches for curriculum development as well as manual grading approaches are rigid and cannot be adapted to other settings. This paper introduces a new meta-learning strategy to enhance a variety of machine learning; convolutional neural networks to network visual learning aids, grading tools, and instructional information to set up learning environments. RL algorithms are used in creating student learning paths or programmers, and smart tutoring systems that adapt with the students’ performance and behavioral characteristics. Briefly, the Meta-learning strategies such as MAML and few-shot learning improve the flexibility of these models and make them capable of Rapid Adaptability to perform new learning tasks with a small amount of data. The approach consists of recognizing educational problems, obtaining and preprocessing data, choosing the suitable CNN and RL models, using meta-learning approaches. Ongoing monitoring and feedback are included with the tools in order to maintain update and modify the tools as needed. Evaluation of the performance shows a substantial enhancement in engagement of students, adoption of learning, and grading with the accuracy of 92% for AAC, precision of 90%, recall of 88%, and f1-score are 89%.