Сравнение методов

	MobileNet: Эффективные свёрточные нейронные сети для мобильного зрения	EfficientNet	Дистилляция знаний
Область	Глубокое обучение	Глубокое обучение	Глубокое обучение
Семейство	Machine learning	Machine learning	Machine learning
Год появления	2017	2019	2015
Автор метода	Andrew Howard et al. (Google)	Tan, M. & Le, Q. V.	Hinton, G., Vinyals, O. & Dean, J.
Тип	Lightweight CNN architecture	Compound-scaled convolutional neural network architecture	Neural network compression (teacher–student)
Основополагающий источник	Howard, A. G., et al. (2017). MobileNets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint.	Tan, M. & Le, Q. V. (2019). EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning (ICML 2019), PMLR 97, 6105–6114.	Hinton, G., Vinyals, O. & Dean, J. (2015). Distilling the Knowledge in a Neural Network. NeurIPS Deep Learning Workshop.
Другие названия	MobileNets, Depthwise Separable CNN, Efficient Mobile Vision Network, Mobil Evrişimli Sinir Ağı	EfficientNet, compound scaling CNN, EfficientNet-B0 through B7, EfficientNetV2	Bilgi Damıtma (Knowledge Distillation), bilgi damıtma, teacher-student distillation, model distillation
Связанные	2	4	5
Сводка	MobileNet is a family of lightweight convolutional neural network architectures introduced by Howard et al. at Google in 2017. It is designed to run image classification, object detection, and other vision tasks directly on mobile devices and embedded systems with limited computational budgets. By replacing standard convolutions with depthwise separable convolutions and exposing two global hyperparameters, MobileNet dramatically reduces multiply-add operations and model size while retaining competitive accuracy.	EfficientNet is a family of convolutional neural network architectures introduced by Mingxing Tan and Quoc V. Le (Google Brain) at ICML 2019 that systematically co-scales network depth, width, and input resolution using a single compound coefficient, achieving state-of-the-art image classification accuracy with substantially fewer parameters and FLOPs than prior networks such as ResNet and Inception.	Knowledge Distillation is a model-compression technique, introduced by Geoffrey Hinton and colleagues in 2015, that trains a small student model using the soft-label outputs of a large teacher model. Distilled models such as DistilBERT and TinyBERT reach roughly 97% of the larger model's performance while running far faster.