|
ИСТИНА |
Войти в систему Регистрация |
ИПМех РАН |
||
Ultimate tensorization: convolutions and FC alikeстатья
- Авторы: Timur Garipov, Dmitry Podoprikhin, Alexander Novikov, Dmitry Vetrov
- Сборник: Advances in Neural Information Processing Systems 29 (NIPS 2016)
- Том: 29
- Год издания: 2016
- Первая страница: 348
- Последняя страница: 354
- Аннотация: Convolutional neural networks excel in image recognition tasks, but this comes at the cost of high computational and memory complexity. To tackle this prob- lem, [ 1 ] developed a tensor factorization framework to compress fully-connected layers. In this paper, we focus on compressing convolutional layers. We show that while the direct application of the tensor framework [ 1 ] to the 4-dimensional kernel of convolution does compress the layer, we can do better. We reshape the convolutional kernel into a tensor of higher order and factorize it. We combine the proposed approach with the previous work to compress both convolutional and fully-connected layers of a network and achieve 80 network compression rate with 1 : 1% accuracy drop on the CIFAR-10 dataset.
- Добавил в систему: Журавлев Юрий Иванович