COURSE REGISTRATION: CLOSED !! The course was in 2023
Course in english
DESCRIPTION :
- Introduction to the capabilities and strengths of ML/DL with concrete examples of successful implementations; Big Data
- How to implement a data analysis procedure:
- Real data and simulated data
- Real Data Acquisition and preparation
- Formats of data and simulations: column data, images, time series;
- Labelled and unlabelled datasets: supervised and unsupervised learning
- Data quality, filtering, re-sampling
- Missing data: interpolation, multiple imputation
- Weighting input datasets
- Data splitting
- Feature-based supervised learning
- Supervised Deep Learning
- Regression and Classification
- Feature-based supervised learning
- Feature engineering;
- Examples of algorithms: Boosted Decision Trees (BDTs) and Multilayer Perceptron (MLP)
- Supervised Deep Learning
- Image transformations
- Network architecture
- Data mining: Unsupervised learning for parameter space investigation through dimension reduction and visualization
- Acceleration of simulations through Generative Adversarial Networks (GANs);
- Analysis efficiency (ROC curve), definition of analysis cuts, extraction of searched signal from datasets, evaluation of the performance of the analysis
- Final lecture on final considerations and suggestions
- Two-hour session where each student presents her/his analysis challenges with 2 slides
SCHEDULE:
November 2022
Email (contact pédagogique) : yvonne.becherini@apc.in2p3.fr
2 points
REGISTRATION:
For doctoral school STEPUP : please register through the usual google sheet with code: ED-SPU30-STE30
( In 2022 also through Ametis : https://amethis.app.u-paris.fr/amethis-client/formation/gestion/formation/4852
For non UPC students, you will need a certificate (« certificat de scolarité 2022-2023 ») proving that you belong to the doctoral school. Once you have it, please sent it to ced.formation.drive@u-paris.fr and sevrine.frimat@u-paris.fr to ask for your registration. )