TinyML Based Edge Intelligent English Classroom Quality Assessment Scheme

Abstract

With the escalating demand for intelligent educational evaluation driven by the advancement of artificial intelligence and edge computing, traditional English classroom quality assessment methods, characterized by subjectivity, inefficiency, and lack of real-time feedback, struggle to meet modern educational needs. Moreover, cloud-based AI solutions pose risks to student data privacy and suffer from high latency. TinyML, a lightweight machine learning paradigm, is inherently compatible with edge intelligence due to its ability to run efficiently on resource-constrained edge devices and perform local inference, thereby reducing latency and safeguarding data privacy. This paper presents an edge-intelligently assisted English classroom quality evaluation scheme based on the TinyML model, which utilizes edge computing devices to analyze classroom interaction data in real time, improving the objectivity and timeliness of teaching quality assessment. The scheme employs lightweight deep learning models for various analyses and conducts localized data processing to avoid cloud-related issues. Experimental results demonstrate its superiority over traditional cloud AI schemes in accuracy, real-time performance, and resource utilization, offering a viable approach for intelligent education evaluation.