We are proud to announce that a team of undergraduate students from the COMP 4436 course (Artificial Intelligence of Things) has been awarded the Best Student Paper Award at the 22nd IEEE International Conference on Networking, Sensing, and Control (ICNSC), held from October 1–3, 2025 in Oulu, Finland. The winning paper, titled “Energy-Efficient and High-Accuracy AIoT Visual Sensing Using K-Vote, Transfer-Spiking Net, and Reconstructive Classifier,” was developed as part of the students’ semester project. The student team comprises: Abstract of the Paper: The increasing deployment of Artificial Intelligence of Things (AIoT) applications necessitates efficient and accurate visual sensing models, yet a comprehensive comparison of machine learning (ML), deep learning (DL), and spiking neural networks (SNN) within this domain remains limited, creating a gap in understanding their relative effectiveness for resource constrained environments. This study aims to evaluate and compare representative models across five paradigms: supervised ML, unsupervised ML, supervised DL, unsupervised DL, and SNN; by implementing novel approaches such as K-Vote, Reconstructive Classifier, and Transfer Spiking Net, on simple binary image classification datasets. Employing performance metrics including accuracy, precision, recall, F1 score, and runtime efficiency, our results demonstrate that the Transfer Spiking Net achieves an accuracy of 99.2%, approaching the state-of-the-art, with comprehensive visualizations and analysis revealing the strengths and limitations of each method. The contributions of this work include a systematic performance comparison, the proposal of innovative neural network models, and the release of open-source code to support future research in AIoT visual sensing applications.
