Machine-Learning

MC-Toolkit - Machine Learning

The goal of this group is to discuss about topic related to Machine Learning.

For easily editing a README: https://stackedit.io/

Next meetings (calendar) :

• Thursday 4/11 at 10:00 am : Machine learning #3: Physical constraints to machine learning methods Jordi Bolibar (Material : Paper, Video, Book)

Invited speaker

• Louis Le Toumelin (CEN) RG git : Use ML to do downscaling of wind prediction in mountain environment.
• Ravi raj purohit Purushottam Raj Purohit (ESRF) RG : Use ML to index diffraction pattern. He trains its model using simulated data.

Possible next meetings / Topic:

Past meetings:

• Tuesday 22/06 at 10:30 am : Machine learning #2: Presentation of a ML probleme from scratch (Clara Burgard)
• Tuesday 11/05 at 10:30 am: Machine learning #1: Creation of a new discussion group within MCToolkit, first meeting (Thomas Chauve)

Here are some ideas of topics/tools to talk about:

Methods

• Tree base machine learning (Random Forest, Gradient Boost)
• Neural Network
• Convolutional Neural Network (CNN)
• Neural Ordinary Differential Equations (Neural ODEs)
• Universal Differential Equations (UDEs)
• Physics-Informed Neural Networks (PINNs)

Tutorials

• FIDLE

Environement

• scikit-learn framework
• TensorFlow framework
• pyTorch framework
• pySindy

Explainable AI

• SHAP
• LIME
• Recursive Feature Elimination (RFE)
• Permutation Importance
• Universal Differential Equations

Paper discussion

• Padarian et al. 2020 - Game theory interpretation of digital soil mapping convolutional neural networks
• Valmonte Riel et al. 2021 - Data-Driven Inference of the Mechanics of Slip Along Glacier Beds Using Physics-Informed Neural Networks
• Rackauckas et al., 2020 - Universal Differential Equations for ScientificMachine Learning
• Brinkerhoff et al., 2020 - CONSTRAINING SUBGLACIAL PROCESSES FROM SURFACE VELOCITY OBSERVATIONS USING SURROGATE-BASEDBAYESIAN INFERENCE

Present your work

The goal is to present a tool and exchange with others + know who to ask when we need help.

Contacts for proposing new topics:

• thomas.chauve@univ-grenoble-alpes.fr

PS: you can also make directly a pull request on this file in the Possible unplanned meetings section or use directly the Slack #machine-learning channel to contact us. We will then add you as a contributor so that you can share some code.