Basic Information

• Language: English
• Prerequisite courses
  • Formal Languages and Complexity (must)
  • Logic and Discrete Structures (must)
  • Formal Specification and Verification (recommended)
  • Software Verification (recommended)
  • SAT Solving (recommended)
• Date: Thursday, 10:15-11:45
• Location: Seminar Room 133, Oettingenstr. 67
• First meeting: 2024-04-18 (about the detailed seminar organization and expectations)

Content

Model checking is an important research field in computer science where automatic solutions to the following problem are studied: Given a computational model and a specification, decide whether the model satisfies the specification or not. In practice, a computational model can be a digital circuit (hardware) or a program (software). A specification can be a safety property requiring that errors never happen during the execution of the model. Compared to testing, model checking can guarantee the correctness of a computational model with mathematical rigor. Model checking is rooted in a solid theoretical foundation and requires advanced software engineering for tool implementation. The inventors of model checking won the 2007 Turing Award, and leading technology companies use model-checking techniques to assure the quality of their products.

Goal

The goal of this seminar is to learn the mathematical foundation and understand the working of state-of-the-art algorithms for hardware and software model checking. At the end of this seminar, students should become familiar with the background knowledge of model checking and able to explain a scientific publication on model checking in oral and written forms.

Organization

The seminar will consist of three phases:

1. Reading sessions: In the first phase, we will recap the mathematical foundation of model checking. Students will be divided into groups. Each group will be assigned a topic and lead the discussion on the topic in one seminar session.
2. Weekly meetings: In the second phase, each student will be assigned a publication on an algorithm for model checking. Students will have weekly meetings with the mentors to report their progress. Mentors will guide students to read and understand the algorithms in the publications.
3. Final presentation and report: In the final phase, students will present the algorithms assigned to them and write a report explaining them with examples and analysis.

Team

• Prof. Dr. Dirk Beyer
• Nian-Ze Lee, Ph.D.
• Po-Chun Chien

References

• Handbook of Model Checking, 2018

The seminar on Data Ethics is part of the MSc Data Science program at Ludwig-Maximilians-Universität (LMU) Munich. The course will be lead by Prof. Dr. Dieter Kranzlmüller and Fabio Genz.

Contents: The seminar is one part of the module Data Ethics and Data Security, which covers basic legal and ethical questions and challenges of data security. This course helps the students to prepare lectures on ethical and legal aspects of data ethics. Students are introduced to the technical, legal, and ethical issues of data security, especially when dealing with personal data or when planning experiments in Data Science.

Objective: Students will reflect on standard procedures and problems of data protection and learn technical methods to handle data responsibly.

The seminar on Data Ethics is part of the MSc Data Science program at Ludwig-Maximilians-Universität (LMU) Munich. The course will be lead by Prof. Dr. Dieter Kranzlmüller and Fabio Genz.

Contents: The seminar is one part of the module Data Ethics and Data Security, which covers basic legal and ethical questions and challenges of data security. This course helps the students to prepare lectures on ethical and legal aspects of data ethics. Students are introduced to the technical, legal, and ethical issues of data security, especially when dealing with personal data or when planning experiments in Data Science.

Objective: Students will reflect on standard procedures and problems of data protection and learn technical methods to handle data responsibly.

Some algorithms from machine learning have successfully approached many of the problems that seemed unsolvable a few decades ago (eg. computer vision, image generation). Irrespective of the successes in applied settings, the learning process, as well as the uncertainties that underlie the learning, remain open problems that challenge the foundations of statistics. Since attempts relying on the traditional (Bayesian and frequentist) frameworks have shown only limited success, some researchers have directed their efforts in questioning the foundations of statistics in a very principled manner (eg. What if there is no data generating process? What if we are not certain about our prior beliefs? Can we talk about a conditional distribution of the parameters without imposing a prior on the parameters? Are all uncertainty dimensions covered?).
In this seminar, we will explore foundations rooted in five main blocks, namely: traditional Bayesian and frequentist frameworks, imprecise probability, decision theory, information theory and compression algorithms. We will build on the following non-exhaustive list of papers:

1. Aleatoric and epistemic uncertainty in machine learning: An introduction to concepts and methods - Hüllermeier, Waegeman (2021)
2. Sources of Uncertainty in Machine Learning--A Statisticians' View - Gruber et al. (2023)
3. Minimum description length revisited - Grünwald and Roos (2020)
4. The Interplay of Bayesian and Frequentist Analysis - Bayarri and Berger (2004)
5. Robust Bayesian Analysis: sensitivity to prior - Berger (1987)
6. All models are wrong, but many are useful: Learning a variable's importance by studying an entire class of prediction models simultaneously - Fisher et al. (2019)
7. Strictly frequentist imprecise probability - Fröhlich et al. (2024)
8. Information-theoretic upper and lower bounds for statistical estimation - Zhang (2006)
9. The E-Posterior - Grünwald (2023)
10. How the game-theoretic foundation for probability resolves the Bayesian vs. frequentist standoff? - Shafer (2020)
11. Judicious Judgement Meets Unsettling Updating: Dilation, Sure Loss and Simpson’s Paradox - Gong and Meng (2021)
12. Testing by betting: A strategy for statistical and scientific communication - Shafer (2019)


We invite students to suggest their own papers that fit within the topic as well.


Who is this seminar for: Motivated students who are open to explore current trends in the foundations of statistics that might provide tools to solve open problems in uncertainty quantification and learning. The seminar is open to master’s students from statistics, mathematics, economics, data science, and similar backgrounds. For students of the Statistics and Data Science program, the seminar can be recognized for the mandatory seminar module in the Methodology & Modeling, Machine Learning and Social Statistics and Data Science tracks, as well as for the additional elective general seminar module ‘Advanced Research Methods in Theoretical Statistics’ (WP 51).

Data Science tracks, as well as for the additional elective general seminar module ‘Advanced Research Methods in Theoretical Statistics’ (WP 51). Requirements for obtaining 9 ECTS credits: Every student has to
* give a 45-60 minutes long presentation on their chosen topic supported by slides and
* write a seminar paper based on it (As a rough guideline: the typical seminar paper is 25-30 pages long; in case the paper is very technical, it can be considerably shorter, which is to be agreed on with the organisers of the seminar). The paper has to be submitted by 01.06.2025 and shall contain a deepened and extended version of the presentation, also taking up the discussion of the presentation and positioning the chosen topic in the context of the seminar. (The registration for the 9 ECTS has to be done via LSF.)

If places are left, we are happy to offer in addition a 3 ECTS version, based on a shorter term paper or presentation. The 3 ECTS could be used flexibly for one of the corresponding generic modules. To apply for the 3 ECTS option write an email to ivan.melev@lmu.de no later than September 25 (no registration via LSF!).

We expect active participation in the discussions of the seminar. This includes a quick personal preparation for each presentation based on some preparation material (1 page) the speakers are asked to provide one week in advance.

What is the format? This is a block seminar in a face-to-face format. It will take place during the week of 26.02.2025-28.02.2025; the exact dates are still to be decided.

This course covers advanced techniques for automatic software verification, especially those in the field of software model checking. It continues the Bachelor course Formal Verification and Specification (FSV). Knowledge from FSV is helpful but not mandatory. This course can be used for the specialization "Programming, Software Verification, and Logic" in the MSc computer science (cf. German site on specializations).

Topics

The course will cover the following topics:

1. Mathematical foundation for software verification
2. Configurable program analysis
3. Strongest postcondition
4. Predicate abstraction with a fixed precision
5. Craig interpolation and abstraction refinement (CEGAR)
6. Predicate abstraction with precision adjustment
7. Bounded model checking and k-induction
8. Observer automata
9. Verification witnesses
10. Test generation and symbolic execution
11. Cooperative verification
12. Project: implementing a software verifier (2 weeks)

Reference Materials

• Combining Model Checking and Data-Flow Analysis (Chapter 16 in the Handbook of Model Checking)
• A Unifying View on SMT-Based Software Verification

Organization

The course consists of weekly lectures and tutorials. Important announcements are sent via Moodle messages.

Time Slots and Rooms

• Lecture: 10:15 - 12:00, Wednesday, in M 101, HGB (by Dr. Thomas Lemberger)
• Tutorial: 14:15 - 15:45, Thursday, in D Z007, HGB (by Marek Jankola)

The first lecture session on 2025-05-07 will be about the detailed course organization and expectations. The first tutorial session will be on 2025-04-24.

Enrollment

Please enroll with the following key: wr$NUB#L9Zvl0&W37Ynx

Course Description

In this lecture, we will consider various classes of stochastic processes that may differ in their state spaces and underlying index sets with a special focus on Gaussian, Lévy and Markov processes. In summary, the lecture will be divided into three core topics: the construction, the path behaviour and the probabilistic analysis of general stochastic processes.

Target Participants

Testkurs zur Einarbeitung der Mitarbeiterinnen und Mitarbeiter des Mathematischen Instituts

For more information, please see here.

Dieser öffentlich zugängliche Kurs wird für das Tutorienprojekt des Mathematischen Instituts benötigt.

Für Sie hat die automatische Einschreibung anscheinend nicht funktioniert. Das kommt häufiger vor.

Klicken Sie bitte unten unter "Selbsteinschreibung (Gast)" auf "Weiter" und nutzen Sie dann bitte den Selbsteinschreibeschlüssel "Gast".

Sollten Sie dennoch Probleme mit der Nutzung eines Links im Zusammenhang mit dem Tutorienprojekt oder MOAS haben, wenden Sie sich bitte an Stefan Ufer, ufer@math.lmu.de.

Das Modul vertieft die Kenntnisse der Grundlagen der Computerlinguistik. Diese Grundlagen umfassen linguistische Grundlagen (Morphologie, Syntax, Semantik), mathematische Grundlagen (Logik, Formale Sprachen, lineare Algebra, Wahrscheinlichkeitstheorie) und informatische Grundlagen (Komplexität, Programmieren, Parsing). Die Grundlagen in diesen drei Bereichen werden in der Übung an Beispielen konkretisiert und mittels umfangreicher Ubungsaufgaben von den Studierenden eingeübt.

Kursbeschreibung

In dieser Vorlesung werden bedingte Erwartungswerte mithilfe des Satzes von Radon-Nikodým eingeführt und ihre wesentlichen Eigenschaften hergeleitet. Zudem werden die Grundlagen der Theorie stochastischer Prozesse in diskreter Zeit, mit einem starken Fokus auf Martingale, behandelt. Dazu werden wir uns mit relevanten Methoden aus der Maß- und Integrationstheorie befassen und insbesondere das starke Gesetz der großen Zahlen, das Null-Eins-Gesetz von Kolmogorov und die Lemmata von Borel-Cantelli beweisen.

Vorgesehen für Studierende folgender Studiengänge

Modul 5: Behandlung komplexer Traumafolgestörungen einschließlich Dissoziativer Störungen Teil 1 am 15. und 16. Januar 2021

Aufbauend auf den Techniken aus Modul 2 soll das Vorgehen der DBT-PTBS vermittelt werden. Hierzu gehören folgende Aspekte: Erstellung einer Behandlungsplanung und Behandlungshierarchie unter besonderer Berücksichtigung von Suizidalität, Selbstverletzung, Substanzmissbrauch, Ess-Brech-Anfällen und anderen problematischen Verhaltensweisen; Erarbeitung von Techniken zur Behandlung von dysfunktionalen Verhaltensweisen, Etablierung von Alternativstrategien, Verbesserung der Emotionsregulation und der interpersonellen Fertigkeiten; Psychoedukation und Behandlung von Dissoziation; Erarbeitungen von Voraussetzungen für traumafokussierte Interventionen; Indikation und Kontraindikation für das imaginative Nacherleben; Durchführung des imaginativen Nacherlebens unter Berücksichtigung dissoziativer Symptome.

Darüber hinaus wird auf die Erstellung eines Behandlungsleitfadens für Patient*innen mit hoher Dissoziationsneigung eingegangen.