ED352 - Course 'Advanced statistical methods for high energy physics', FELIGIONI/COADOU

After an introduction to probability theory and basic statistics, more advanced topics are presented together with recent developments in statistical methods used for particle physics, in particular for data analysis at the LHC. The complementarity between developing a machine learning approach and optimising the statistical treatment of the dataset to extract a measurement will be explored. An emphasis will be given to practical sessions where the student will exercise the acquired notions.

* Statistical methods for data analysis:
Basic concepts of probability, hypothesis tests, exclusion limits and discovery, CLs method, profile likelihood.
* Machine learning for HEP analysis:
Concepts of machine learning, boosted decision trees, deep learning (DNN, RNN, LSTM, CNN, GNN), data handling, feature selection, figures of merit, hyperparameter optimisation, systematic uncertainties.

Basic knowledge in programming language (C++, python)

- Being able to comprehend the statistical challenges of a data analysis
- Understanding how to optimize the analysis strategy with machine learning tools
- Being able to build, train and apply a machine learning algorithm
- Being able to extract a limit or perform a measurement from a dataset

---

---

---

---

---

---

---

---

---

---