CoShNet: A Hybrid Complex Valued Neural Network using Shearlets
Not on the map. Why?
Abstract (arXiv)
In a hybrid neural network, the expensive convolutional layers are replaced by a non-trainable fixed transform with a great reduction in parameters. In previous works, good results were obtained by replacing the convolutions with wavelets. However, wavelet based hybrid network inherited wavelet's lack of vanishing moments along curves and its axis-bias. We propose to use Shearlets with its robust support for important image features like edges, ridges and blobs. The resulting network is called Complex Shearlets Network (CoShNet). It was tested on Fashion-MNIST against ResNet-50 and Resnet-18, obtaining 92.2% versus 90.7% and 91.8% respectively. The proposed network has 49.9k parameters versus ResNet-18 with 11.18m and use 52 times fewer FLOPs. Finally, we trained in under 20 epochs versus 200 epochs required by ResNet and do not need any hyperparameter tuning nor regularization. Code: https://github.com/Ujjawal-K-Panchal/coshnet
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 20 of its 53 references that are in the map. First posted on arXiv 14 Aug 2022.
Builds on
20 of 53 references.
References · 20
- 2013Intriguing properties of neural networkscited by 5,155
- 2017Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithmscited by 4,349
- 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,262
- 2015Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)cited by 1,070
- 2017Opening the Black Box of Deep Neural Networks via Informationcited by 503
- 2021ResNet strikes back: An improved training procedure in timmcited by 368
- 2017How to Escape Saddle Points Efficientlycited by 285
- 2015Tensorizing Neural Networkscited by 250
- 2017Deep Complex Networkscited by 217
- 2021Torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer visioncited by 81
- 2014Deep Roto-Translation Scattering for Object Classificationcited by 50
- 2016Gradient Descent Converges to Minimizerscited by 25
8 more
- 2018DCFNet: Deep Neural Network with Decomposed Convolutional Filterscited by 23
- 2020Complex-Valued vs. Real-Valued Neural Networks for Classification Perspectives: An Example on Non-Circular Datacited by 12
- 2018Evaluation of Complex-Valued Neural Networks on Real-Valued Classification Taskscited by 9
- 2015Deep Convolutional Neural Networks Based on Semi-Discrete Framescited by 6
- 2019Edge, Ridge, and Blob Detection with Symmetric Moleculescited by 6
- 2019Shearlets as Feature Extractor for Semantic Edge Detection: The Model-Based and Data-Driven Realmcited by 5
- 2018Learning Multi-Layer Transform Modelscited by 3
- 2021ChaRRNets: Channel Robust Representation Networks for RF Fingerprintingcited by 3
Built on by
Citing papers per year: 2022: 1.
1 paper.