search

unizard / AwesomeArxiv

89 stars 14 forks source link

[2018.07.31] SqueezeNext: Hardware-Aware Neural Network Design #210

Open unizard opened 5 years ago

unizard commented 5 years ago

SqueezeNet New Version

Institute: UC Berkeley URL: https://arxiv.org/pdf/1803.10615.pdf Code: https://github.com/amirgholami/SqueezeNext https://github.com/luuuyi/SqueezeNext.PyTorch (He is not the author.), pytorch Author: http://amirgholami.org/

Summary In this work, we introduce SqueezeNext, a new family of neural network architectures. SqueezeNext matches AlexNet's accuracy on the ImageNet with 112x fewer parameters, and its deeper variant exceeds VGG-19's accuracy with only 4.4 Million parameters, (31x smaller). SqueezeNext also achieves better top-5 classification accuracy with 1.3x fewer parameters as compared to MobileNet, while avoiding depthwise-separable convolutions that have poor arithmetic intensity. Using hardware simulation results for power and inference speed on an embedded system, guided us to optimize the baseline model that is 2.59x/8.26x faster and 2.25x/7.5x more energy efficient as compared to SqueezeNet/AlexNet without any accuracy degradation.

unizard commented 5 years ago

Architecture

image image

unizard commented 5 years ago

Results

image image