Approach We present INQ which incorporates three interdependent operatio...
Approach By properly defining attention for convolutional neural network...
简介 细粒度分类问题通常是很困难的,这是因为判别区域定位和细粒度特征学习是很具有挑战性的。现有方法常常忽略了区域检测和细粒度特征学习之间的相互关...
SSD的网络分为两部分,前半部分是用于图像分类的标准网络(去掉了分类相关的层),后面的网络是用于检测的多尺度特征映射层,从而达到检测不同大小的目...
Approach Knowledge Transfer (KT), which aims at training a smaller stude...
Approach Given a trained CNN model, we propose an iterative two-step alg...
Approach We propose a simple two-step approach for speeding up convoluti...
Approach We introduce “deep compression”, a three stage pipeline: prunin...
Approach We introduce ternary weight networks (TWNs) - neural networks w...