Machine Learning for Photon Science


<link rel="stylesheet" type="text/css" href="/static/wp/style.css" />

<h1>Machine Learning for Photon Science</h1>

<b>Modulnummer:</b> Q10-41<br>
<b>Englischer Titel:</b> Machine Learning for Photon Science<br>
<b>Leistungspunkte:</b> 10<br>
<b>Lehrperson:</b> <a href="/?teacher=Feuer-Forson">Feuer-Forson</a>

<h3>Empfohlene Vorkenntnisse</h3>
- Python, Numpy, Matplotlib<br>- PyTorch or equivalent deep learning framework<br>- Experience with optimisation frameworks such as Evotorch or Optuna<br>- Basic machine learning and deep learning knowledge will be useful but is not required

<h3>Zwingende Voraussetzungen</h3>
keine

<h3>Inhalt</h3>
Machine Learning and Deep Learning methods have in recent years seen widespread application across many fields of research. In areas where large quantities of complex data are generated, employing AI methods for data analysis, parameter optimisation, and process automation has proven highly successful. At the research institute Helmholtz-Zentrum Berlin, which operates the synchrotron raditation source BESSY II, machine learning methods are being developed and applied to a range of research activities, including materials science, accelerator physics, and life sciences.<br><br>This course is designed to introduce students to the application of machine learning methods to real-world challenges. In contrast to generative AI, the machine learning methods covered in this course will focus on analytical approaches to optimise parameters and extract information from large, complex datasets. The course will be structured around current research in photon science at BESSY II and will use both simulated and real-world data. It will cover both the training and application of various types of deep neural networks, data handling and pre-processing, and the development of optimisation strategies to solve ill-posed inversion problems.<br><br>The lectures will provide all the necessary fundamentals for completing the course, whilst also introducing current research topics in applied machine learning in photon science. The coursework will consist of practical, group-based assignments in which students are required to demonstrate both their understanding of the methods and their ability to solve problems by developing their own solutions. The assignments will be undertaken using Python and various machine learning frameworks.

<h3>Erforderliche Arbeitsleistungen für LP-Vergabe und Prüfungszulassung</h3>
- schriftlich eingereichte und/oder mündlich vorgetragene Lösungen zu Aufgaben<br>- Abschlusspräsentation und schriftlicher Ergebnisbericht

<h3>Lehrveranstaltungen</h3>
Vorlesung:      4 SWS  6 LP<br>Übung:          2 SWS  3 LP<br>MAP: 1 LP


    <h3>Zugeordneter Vertiefungsschwerpunkt</h3>
    Algorithmen und Modelle: nein<br>
    Modellbasierte Systementwicklung: nein<br>
    Daten- und Wissensmanagement: ja<br>
    Ohne Vertiefungsschwerpunkt: nein

    <h3>Sprache im Modul</h3>
    Deutsch: nein<br>
    Englisch: ja

    <h3>Angeboten für Studiengänge</h3>
    M. Sc.: ja<br>
    M. Ed.: ja<br>
    Wirtschaftsmaster: ja


    <h3>Angeboten im</h3>
    Wintersemester: ja<br>
    Sommersemester: nein

    <h3>Turnus</h3>
    Unregelmäßig