analiza aiqbench
This research presents a refined knowledge distillation framework designed to compress deep convolutional networks for land-use image classification without sacrificing accuracy. By combining hard ground truth supervision with soft strategies utilizing Kullback-Leibler divergence and Cosine Similarity, the method successfully transfers capabilities from a VGG16 teacher model to a lightweight MobileNetV2 student architecture. The approach achieves 99.04% accuracy on benchmark datasets while significantly reducing computational complexity, offering practical value for deploying efficient AI models in resource-constrained environments.
Improved Knowledge Distillation for Land-Use Image Classification
arXiv:2606.14886v1 Announce Type: cross Abstract: In the present article, an improved Knowledge Distillation (KD) framework has been proposed for efficient compression of deep convolutional neural networks for land-use image classification task. Motivated by the need to achieve competitive classification accuracy while reducing computational complexity, a teacher-student learning paradigm is adopted in which a VGG16 network transfers knowledge to a lightweight MobileNetV2 model. The proposed framework integrates hard supervision from ground truth labels with a soft supervision strategy that combines Kullback-Leibler divergence and Cosine Similarity losses. Experiments conducted on three land-use datasets show that the proposed KD-based method yields improved performance, and achieves an accuracy of 99.04%, outperforming both baseline student training and single-loss distillation approaches, while retaining substantial model compression.
Abstract:In the present article, an improved Knowledge Distillation (KD) framework has been proposed for efficient compression of deep convolutional neural networks for land-use image classification task. Motivated by the need to achieve competitive classification accuracy while reducing computational complexity, a teacher-student learning paradigm is adopted in which a VGG16 network transfers knowledge to a lightweight MobileNetV2 model. The proposed framework integrates hard supervision from ground truth labels with a soft supervision strategy that combines Kullback-Leibler divergence and Cosine Similarity losses. Experiments conducted on three land-use datasets show that the proposed KD-based method yields improved performance, and achieves an accuracy of 99.04%, outperforming both baseline student training and single-loss distillation approaches, while retaining substantial model compression.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2606.14886 [cs.CV]
(or arXiv:2606.14886v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2606.14886 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 215 cuvinte
Subiecte conexe
Rethinking Dataset Distillation for Classification: Do Distilled Sets Outperform Coresets?
arXiv · 4 nov. 2026
Rethinking Dataset Distillation for Classification: Do Distilled Sets Outperform Coresets?
arXiv · 17 iun. 2026
HumP-KD: A Hybrid Uncertainty-Aware Multi-Stage Progressive Knowledge Distillation Framework for Efficient Fire Classification
arXiv · 15 iun. 2026
ReCAST: Restoration-aware Cascaded Stage-wise Training for Obfuscated SMS Risk Classification
arXiv · 7 sept. 2026