Transposed Convolution

Transposed convolution, a crucial operation in deep learning, reverses the downsampling process of standard convolution, enabling the generation of higher-resolution outputs from lower-resolution feature maps. Current research focuses on optimizing transposed convolution for efficiency and effectiveness within various architectures, including its integration into models like Conv-TasNet and variational autoencoders, as well as developing novel algorithms to handle non-integer strides and improve hardware acceleration. These advancements are significantly impacting diverse fields, from medical image analysis and audio processing to generative modeling and 3D image reconstruction, by improving the speed, accuracy, and resource efficiency of these applications.

Papers

November 3, 2023

Capturing Local and Global Features in Medical Images by Using Ensemble CNN-Transformer
Javad Mirzapour Kaleybar, Hooman Saadat, Hooman Khaloo
Convolutional Neural Network Medical Image Local Feature Global Feature Convolutional Block Transformer Ensemble Transposed Convolution Ensemble CNN

October 26, 2023

Real-time Neonatal Chest Sound Separation using Deep Learning
Yang Yi Poh, Ethan Grooby, Kenneth Tan, Lindsay Zhou, Arrabella King, Ashwin Ramanathan, Atul Malhotra, Mehrtash Harandi, Faezeh Marzbanrad
Deep Learning 1D Convolution Transposed Convolution Heart Sound Auscultation Lung Sound

October 25, 2023

ClearMark: Intuitive and Robust Model Watermarking via Transposed Model Training
Torsten Krauß, Jasper Stang, Alexandra Dmitrienko
Watermarking Method Model Watermarking Visible Watermark Transposed Convolution DNN Watermarking

June 19, 2023

Algorithms of Sampling-Frequency-Independent Layers for Non-integer Strides
Kanami Imamura, Tomohiko Nakamura, Norihiro Takamune, Kohei Yatabe, Hiroshi Saruwatari
Practical Algorithm Music Source Separation Audio Source Separation Transposed Convolution Independent Layer

October 30, 2022

Transposed Variational Auto-encoder with Intrinsic Feature Learning for Traffic Forecasting
Leyan Deng, Chenwang Wu, Defu Lian, Min Zhou
Graph Attention Network Traffic Forecasting Traffic Pattern Traffic4cast Competition Intrinsic Feature Transposed Convolution

October 17, 2022

Deformably-Scaled Transposed Convolution
Stefano B. Blumberg, Daniele Raví, Mou-Cheng Xu, Matteo Figini, Iasonas Kokkinos, Daniel C. Alexander
Direct Convolution Convolution Layer Transposed Convolution

September 8, 2022

Kernel-Segregated Transpose Convolution Operation
Vijay Srinivas Tida, Sai Venkatesh Chilukoti, Xiali Hei, Sonya Hsu
Convolution Operation Transposed Convolution

August 30, 2022

A Learning-Based 3D EIT Image Reconstruction Method
Zhaoguang Yi, Zhou Chen, Yunjie Yang
Image Reconstruction Electrical Impedance Tomography Transposed Convolution

May 2, 2022

Efficient Accelerator for Dilated and Transposed Convolution with Decomposition
Kuo-Wei Chang, Tian-Sheuan Chang
Microbial Decomposition Reconfigurable Intelligent Transposed Convolution

February 22, 2022

Transition Matrix Representation of Trees with Transposed Convolutions
Jaemin Yoo, Lee Sael
Tree Specie Tree Representation Tree Model Structure Discovery Transition Matrix Transposed Convolution

February 4, 2022

EcoFlow: Efficient Convolutional Dataflows for Low-Power Neural Network Accelerators
Lois Orosa, Skanda Koppula, Yaman Umuroglu, Konstantinos Kanellopoulos, Juan Gomez-Luna, Michaela Blott, Kees Vissers, Onur Mutlu
Direct Convolution Dilated Convolution Inference Accelerator Transposed Convolution Dataflow Based