Table 6.
Qualitative comparison of smart education related works.
| Use Case | Ref | Contribution | AI Role (At the Edge) |
AI Algorithm | Dataset | AI Placement | Employed Technology |
Platform | Metrics | Benefits AI-Edge |
Drawbacks | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Smart education | S. engagement monitoring | [61] | Attention detection of participants | CNN | Prediction | (DAiSEE) | Edge ( pc) | Pretrained model | Python | Accuracy | - | Accuracy needs to improve |
| [62] | Improve long-distance education | Classification | ResNet-50 | Fer2013 emotion dataset | Mobile edge computing | Hardware accelerator | / | Confusion matrix accuracy | High accuracy | Accuracy needs to improve | ||
| [63] | Real-time intervention in negative emotional contagion in a smart classroom | Classification | CNN | Fer2013 emotion dataset | Edge preprocessing | Hardware accelerator | JavaScript, TensorFlow, OpenCV | Accuracy | Less response time | Accuracy needs to improve | ||
| [64] | Multimodal engagement analysis | Prediction | DL | Real-world data | Edge server (PC) | / | JIFF, JavaScript library, TensorFlow | Average performance impact on edge device /server | Scalability | Computational overhead | ||
| [65] | Student stress monitoring and real-time alert generating | Prediction | VGG16, BiLSTM, NB | Real-world data Kaggle dataset | Fog cloud | Cloud training | Not mentioned | Specificity, sensitivity, accuracy, F-measure | High accuracy | Eliminate historical record | ||
| Skill assessment | [66] | Monitors the academic/skill of students for timely employability classification of graduation. | Resource management | K-means, PCA, KNN | Real-world dataset | Fog nodes | / | iFogSim toolkit | Mean absolute percentage error (MAPE) | Scalability | Processing overhead | |
| [67] | Education quality evaluation | ANFIS Bayesian belief network (BBN) | Environmental datasets, staff-related dataset, physical dataset, students’ academic-related historical dataset | Raspberry Pi v3 is | / | Weka | Precision, specificity, sensitivity, BBM, accuracy, RMSE, MAS | Stability, reliability | Accuracy needs to be improved | |||
| [68] | Ideology and politics education evaluation in 5G | Resource management data caching | PSO | Edge devices | Not mentioned | - | / | Energy consumption, latency | Scalability, low energy consumption, low latency | - |