MachineLearning: Difference between revisions
No edit summary |
No edit summary |
||
| Line 1: | Line 1: | ||
This wiki page is the starting point for material concerning Machine Learning, Deep Learning and Artificial Intelligence. At present it is a stub. Please feel free to add content. | |||
Machine learning is a technique which exploits details of correlations between variables to distinguish classes of data. Typical cases might be separating signal from backgrounds by analyzing all the variables which show any difference between the data classes. This topic includes neural networks, boosted decision trees, kernel density, and many other techniques. Mu2e uses several neural nets in reconstruction and analysis. | Machine learning is a technique which exploits details of correlations between variables to distinguish classes of data. Typical cases might be separating signal from backgrounds by analyzing all the variables which show any difference between the data classes. This topic includes neural networks, boosted decision trees, kernel density, and many other techniques. Mu2e uses several neural nets in reconstruction and analysis. | ||
Root has a user-friendly system called TMVA. | Root has a user-friendly system called TMVA. | ||
Revision as of 20:00, 3 December 2021
This wiki page is the starting point for material concerning Machine Learning, Deep Learning and Artificial Intelligence. At present it is a stub. Please feel free to add content.
Machine learning is a technique which exploits details of correlations between variables to distinguish classes of data. Typical cases might be separating signal from backgrounds by analyzing all the variables which show any difference between the data classes. This topic includes neural networks, boosted decision trees, kernel density, and many other techniques. Mu2e uses several neural nets in reconstruction and analysis.
Root has a user-friendly system called TMVA.
Fermilab has a strong machine learning group with many resources.
The training of classifiers often involves very compute-intensive procedures which may run very effectively on specialized systems like graphics cards. The computing sector is developing access to these resources, which are becoming more widely available.
A living review of the literature for Machine Learning in Particle Physics is maintained at [https://iml-wg.github.io/HEPML-LivingReview/]