SqueezeNeXt

Introduced by Gholami et al. in SqueezeNext: Hardware-Aware Neural Network Design

SqueezeNeXt is a type of convolutional neural network that uses the SqueezeNet architecture as a baseline, but makes a number of changes. First, a more aggressive channel reduction is used by incorporating a two-stage squeeze module. This significantly reduces the total number of parameters used with the 3×3 convolutions. Secondly, it uses separable 3 × 3 convolutions to further reduce the model size, and removes the additional 1×1 branch after the squeeze module. Thirdly, the network use an element-wise addition skip connection similar to that of ResNet architecture.

Source: SqueezeNext: Hardware-Aware Neural Network Design

Read Paper See Code

Papers

Paper	Code	Results	Date	Stars

Tasks

Task	Papers	Share
Quantization	1	33.33%
Clustering Multivariate Time Series	1	33.33%
Multivariate Time Series Imputation	1	33.33%

Usage Over Time

This feature is experimental; we are continuously improving our matching algorithm.

Components

Component	Type	Add Remove
Batch Normalization	Normalization
Convolution	Convolutions
Dense Connections	Feedforward Networks
Global Average Pooling	Pooling Operations
Max Pooling	Pooling Operations
ReLU	Activation Functions
Softmax	Output Functions
SqueezeNeXt Block	Skip Connection Blocks
Xavier Initialization	Initialization

Categories

Add Remove

Convolutional Neural Networks