Assemble-ResNet

Introduced by Lee et al. in Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network

Assemble-ResNet is a modification to the ResNet architecture with several tweaks including using ResNet-D, channel attention, anti-alias downsampling, and Big Little Networks.

Source: Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network

Read Paper See Code

Papers

Paper	Code	Results	Date	Stars

Tasks

Task	Papers	Share
Speech Synthesis	1	12.50%
Text-To-Speech Synthesis	1	12.50%
Fine-Grained Image Classification	1	12.50%
Fine-Grained Visual Recognition	1	12.50%
General Classification	1	12.50%
Image Classification	1	12.50%
Image Retrieval	1	12.50%
Retrieval	1	12.50%

Usage Over Time

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

Components

Component	Type	Add Remove
Anti-Alias Downsampling	Downsampling
Big-Little Module	Skip Connection Blocks
Dense Connections	Feedforward Networks
ResNet-D	Convolutional Neural Networks
Selective Kernel	Image Model Blocks

Categories

Add Remove

Convolutional Neural Networks