Embedded Files
Learning from data in a resource-conscious manner using geometry
Geometric Green Learning
Geometric Green Learning
Using different aspects of geometry, we dive into a better understanding of application-oriented data, to provide low resource consumption solutions
Using different aspects of geometry, we dive into a better understanding of application-oriented data, to provide low resource consumption solutions
Geometry: differential geometry in finite and infinite dimensions, information geometry, Riemannian geometry, Poisson geometry,...
Green: low, justified or responsible usage of computational resources and knowledge
Learning: classical and deep learning, frugal machine learning, manifold learning, metrics learning, parameterizations learning, distributions learning, parameters learning, ...
A special session on Geometric Green Learning was organized at GSI23: https://gsi2023.org/
Session organized by: Alice-Barbara Tumpach, Diarra Fall and Guillaume Charpiat
What can you do to make your learning solution more green?
Effort 3
Effort 3
Never neglect to test a low computational coarse solution, and eventually use it to guide a more accurate but computationally heavy solution.
Image par David Mark de Pixabay
Effort 4
Effort 4
Take a step back and ask yourself if your solution is using the right amount of computing power, knowledge, and tools, or if your solution is oversized for your application. Can you measure the gain of your complex solution compared to a low computational coarse solution? If it is minimal, reserve your solution for other applications and modify your solution accordingly.
Page updated
Google Sites
Report abuse