KingJamesSong/DifferentiableSVD

A collection of differentiable SVD methods and ICCV21 "Why Approximate Matrix Square Root Outperforms Accurate SVD in Global Covariance Pooling?"

33
/ 100
Emerging

This project provides various methods to improve the stability and performance of Singular Value Decomposition (SVD) in deep learning models. It helps machine learning engineers and researchers who are developing computer vision systems, especially for image classification, by providing more reliable ways to handle covariance pooling layers. You input your image classification model and it outputs a model with enhanced accuracy and stability, particularly for fine-grained recognition tasks.

No commits in the last 6 months.

Use this if you are a machine learning engineer or researcher working on deep learning models for computer vision and need more stable and performant SVD computations within your network architectures.

Not ideal if you are looking for a general-purpose SVD library for mathematical computations outside of a deep learning context or for applications other than computer vision.

deep-learning computer-vision image-classification fine-grained-recognition neural-network-optimization
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 9 / 25
Maturity 16 / 25
Community 8 / 25

How are scores calculated?

Stars

80

Forks

4

Language

Python

License

Apache-2.0

Last pushed

Oct 20, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/KingJamesSong/DifferentiableSVD"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.