Das Seminar behandelt Themen der Bioethik, Medizin Ethik und KI Ethik, die für aktuelle Themen der Bioinformatik relevant sind.
Im Seminar diskutieren wir ethische Fragen, die für die Bioinformatik relevant sind. Dabei werden Grundlagen der Ethik und ihre Anwendung auf Biologie, Evolution, Medizin, und Künstliche Intelligenz behandelt. Es werden aber auch aktuelle Themen diskutiert wie z.B. CRISPR/Cas Genomeditierung, Gene Drive, gentechnisch modifizierte Pflanzen und Tiere, Gentherapie, Klonierung, Biokampfstoffe, langfristige Freisetzung neuer Arten oder neuer Pathogene, langfristige Änderungen der Biodiversität, Klimawandel, Personalisierte Genomik, Individualisierte Medizin, Medizinethik, ... diskutiert.
Weitere Information, mögliche Themen und Literaturhinweise finden sich hier: https://www.bio.ifi.lmu.de/studium/ss2026/sem_ethics/index.html
- Enseignant: Ralf Zimmer
Fehlerfreie Software ist ein Wunschtraum aus den Anfängen der Informatik. Die Fortstchritte im Bereich der künstliche Intelligenz werden zunehmend bei der Qualitätssicherung eingesetzt. Das Spektrum umfasst Ansätze zur Testgenerierung, zum Erkennen von Sicherheitslücken, zur Generierung von Invarianten und Spezifikationen, und auch zum Erzeugen von Beweisen.
Im Seminar besprechen wir unterschiedliche ausgewählte Ansätze und Tools aus der aktuellen Forschung und Praxis um Softwarequalität sicherzustellen. TeilnehmerInnen stellen jeweils eine Technik vor und demonstrieren diese anhand praktischer Beispiele.
Zeit: Di 16-18 (Ort wird noch bekannt gegeben)
Aufgaben
- Einarbeitung in das Thema
- Präsentation (ca 15 min)
- Schriftliche Ausarbeitung (8-10 Seiten LNCS)
- Enseignant: Daniel Baier
Tuesday, 10-12 Uhr c.t.
14.04.2026 - 14.07.2026
Akademiestr. 7,
1. Stock, Room 105
2 SWS
Over the past years, there has been a surge in new AI-based image generation and editing tools that do not require special computer skills, but are usable by laypersons, artists, and designers. This has been largely influenced by the publicly available, open-source "Stable Diffusion" model (https://ommer-lab.com/research/latent-diffusion-models/). Now many researchers, start-ups, and artists are investigating downstream tasks without the need for a high-performance GPU cluster to train a base model. Moreover, a number of closed-source services such as Midjourney and Open AI's Dall-E 2 have also drawn a lot of attention.
The foundation of this technology is the task of generating a single image solely based on a textual description of what should be depicted in the image. Examples of this can be found on websites such as https://lexica.art/. This technology can be extended to include additional information, such as depth maps, and allows for flexible image editing by changing existing objects based on text or removing parts of the image and filling it while paying attention to the remaining image. The latest advances also allow for generation or modification of video as well as rendering 3D scenes.
These topics and more will be covered in our seminar, where we will investigate AI-based image and video editing and generation techniques. Each student will focus on a specific topic. The objective of this seminar is then to investigate the connection between these techniques and the students' respective fields of study and the greater societal and research implications. We will explore potential applications and issues in applying this technology. Each student will give a presentation of their ideas and write a report about the technique and its potential applications and implications.
- Enseignant: Ursula Fantauzzo
- Enseignant: Olga Grebenkova
- Enseignant: Timy Phan
Im Rahmen dieses Seminars werden ausgewählte Themen aus dem Bereich der Mobilen und Verteilten Systeme behandelt, die insbesondere aus den Forschungsschwerpunkten des Lehrstuhls stammen. In den letzten Semestern führte das zu einem Fokus auf Themen aus dem Bereich des Maschinellen Lernens und Quantencomputings.
Ein Ziel des Seminars ist auch das Erlernen bzw. Üben wissenschaftlicher Arbeitstechnik. Hierzu wird im Laufe des Semesters eine Veranstaltung zu Präsentations- und Arbeitstechnik angeboten und durch individuelles Feedback ergänzt.
Die Endnote des Seminars ergibt sich aus der Qualität der wissenschaftlichen Arbeit und der abschließenden Präsentation.
- Enseignant: Philipp Altmann
- Enseignant: Leo Sünkel
- Enseignant: Maximilian Zorn
This seminar covers a selection of current topics from the areas of High Performance Computing, Quantum Computing, Virtual Reality and Cryptography for students to work on.
Participants will work on one topic throughout the semester (master students on their own, bachelor students in teams of two). The goal is to write a paper, submit it to a fictitious conference committee, review each other's work and present the findings at an end-of-term "conference" in the seminar.
The task is supported by a lecture about scientific writing and presentations and by an individual supervisor for each topic.
The final seminar presentations will take place in block from at the of the semester.
Here is a tentative list of topics covered in the seminar:
- CPU vs. GPU Architectures - Contrasting the Latency-Optimized design of CPUs (complex branching) with the Throughput-Optimized design of GPUs (massive parallelism)
- Evolution of GPU Hardware - Tracing the path from early fixed-function graphics chips to the general-purpose GPUs used in modern AI
- TPUs (Tensor Processing Units) - Google’s custom ASICs that use Systolic Array designs to pass data through a grid of processors without constant memory access
- Vector and Matrix Engines (AVX/SVE/AMX) - Specialized units within a CPU designed to perform vector and matrix operations to accelerate deep learning
- Neuromorphic Computing - Computer chips that mimic how the brain uses "spiking neurons" to process information
- Using the Cerebras AI Chip for Scientific Computing - Strategies for developing scientific computing applications on this non-traditional hardware platform
- Julia for HPC - Utilizing the Julia programming language for high performance numerical and scientific computing
- Domain-Specific Languages (DSLs) - Specialized languages like Halide that simplify writing high performance code for specific hardware targets
- Lightweight Virtualization (Firecracker) - Using microVMs to provide the isolation of traditional virtual machines with the speed and efficiency of containers
- Mixed-Precision Computing - Utilizing lower-bit formats (such as BF16) to accelerate scientific simulations while maintaining accuracy
- In-Situ Analysis and Visualization - Processing and visualizing data in real-time while it is still in the memory of the supercomputer, avoiding the bottleneck of writing to disk
- DNA Data Storage - The experimental use of synthetic DNA to store massive amounts of data in a biological format that can last for centuries
- In-Network Computing with SmartNICs - Offloading computational tasks (like data reduction or encryption) directly to smart Network Interface Cards to improve HPC efficiency
- Ultra Ethernet - An effort to evolve standard Ethernet into a high-speed, reliable fabric suitable for massive AI training clusters and HPC
- Zero Trust Networking - A security philosophy where no device, user, or connection is trusted by default
- Trusted Execution Environments (TEEs) - Hardware vaults such as Intel SGX that protect sensitive data and code even from the computer’s own OS
- Digital Sovereignty - The movement by nations to build independent hardware and cloud infrastructures to ensure data remains within their physical borders
- Post-Quantum Cryptography Hardware - New chip designs built to run the complex math needed to stay safe from future quantum computer attacks
- Processing-in-Memory (PIM) - While TPUs use systolic arrays to minimize memory access, PIM goes a step further by integrating logic directly into the memory chips (DRAM or SRAM)
- Quantum Accelerators for HPC - Hybrid classical-quantum computing models and how emerging quantum processors may be integrated into traditional supercomputing workflows
- Enseignant: Sergej-Alexander Breiter
- Enseignant: Karl Fürlinger
Clustering is an unsupervised learning task that aims to partition data into different groups without label information. Learning a high-quality data representation is crucial for clustering algorithms. Recently, deep clustering, which learns clustering-friendly representations using deep neural networks, has been widely applied to a broad range of clustering tasks. Existing surveys on deep clustering mainly focus on single-view settings and network architectures, while overlooking complex real-world clustering scenarios. To address this issue, this seminar explores the technological landscape of deep clustering, ranging from deep single-view clustering to deep multi-view single clustering and deep multi-view multiple clustering.
The course covers a variety of deep clustering approaches and how they partition data. It also fosters critical discussions on the rationale behind algorithmic choices, exploring key research questions and design trade-offs
- Enseignant: Mamdouh Aljoud
- Enseignant: Zhicong Xian
Das Erlernen des Programmierens fällt vielen Menschen nicht leicht und benötigt viel Übung. Nicht immer sind Lehrpersonen in der Lage alle Studierende adäquat zu betreuen und ihnen detailliertes und schnelles Feedback zu geben. Ein Ansatz dieses Problem zu lösen, besteht darin Systeme zur (semi-)automatischen Bewertung und/oder Generierung von Feedback einzusetzen.
Im Seminar werden technische und didaktische Herausforderungen und Lösungsansätze diskutiert, die Lernende und Lehrende beim Lehren und Lernen unterstützen.
Die Prüfungsleistung besteht aus einem Vortrag und aktiver Teilnahme im Seminar sowie einer schriftlichen Ausarbeitung.
Das Seminar findet als Präsenzveranstaltung statt, damit gute Diskussionen möglich sind.
- Enseignant: Sven Strickroth
Malware (d.h. Viren, Trojaner und Würmer) entwickelt sich stetig weiter. Waren die ersten Viren in den 1970er Jahren noch eher vereinzelte Experimente, schafften es in den 90er Jahren einige Schadprogramme in die Medien, bevor sich in den 2000ern vor allem E-Mail-Würmer explosionsartig ausbreiten konnten. Mittlerweile dominieren kriminelle Organisationen und Geheimdienste das Geschehen und entwickeln hochspezialisierte Malware, von Verschlüsselungs-Trojanern zur Erpressung von Lösegeldern bis zu unauffälligen Hintertüren für Spionage und Sabotage.
In diesem Seminar betrachten wir sowohl die historische Entwicklung von Malware als auch aktuelle Trends. Verfügbare Themen werden akademische Arbeiten zur Malware-Analyse und Erkennung sein, aber auch detaillierte Berichte zu einzelnen prominenten Malware Exemplaren.
Voraussetzung für das Seminar ist ein Interesse für Maschinencode und Details von Betriebssystemen.
Neben den eigentlichem Inhalt sollen auch Grundlagen des wissenschaftlichen Arbeitens vermittelt werden.
- Enseignant: Johannes Kinder
- Enseignant: Tong Liu
- Enseignant: Volker Tresp
- Enseignant: Yao Zhang
Course Overview
This course provides a comprehensive introduction to the rapidly evolving field of Large Vision-Language Models (LVLMs). The curriculum explores the fundamental concepts, historical evolution, and cutting-edge research directions of multimodal tasks.
The course will guide students through key task scenarios, including but not limited to:
Visual Question Answering (VQA)
Multimodal Chain-of-Thought (CoT) Reasoning
Visual Grounding and Instruction Tuning
Instructional Method
The course is conducted entirely in English. We emphasize active, research-oriented learning. Students will:
Engage with state-of-the-art (SOTA) academic papers.
Participate in Collaborative Group Study and intensive seminars.
Develop communication skills through In-class Presentations to showcase their research findings and project progress.
Pre-requisites & Requirements
To ensure a high-quality learning experience, students are expected to meet the following criteria:
Fundamental Knowledge: A strong interest in and basic understanding of Large Vision-Language Models. Prior experience using or experimenting with models (e.g., CLIP, LLaVA, GPT-4V) is highly preferred.
Academic Literacy: The ability to read, analyze, and synthesize technical academic papers efficiently.
highly self-motivated. We are looking for individuals who are willing to explore the complexities of vision-language tasks and are ambitious enough to develop their own hands-on projects by the end of the course.
- Enseignant: Jian Lan
This course on the application of algebra to computer science aims to introduce students to the different subjects of computer science that make use of algebraic theories such as groups, rings, fields and many other objects. This means that we cover a large spectrum of subjects within computer science. The goal of this seminar is twofold: first to learn about these algebraic structures and second to understand how they can be used in computer science.
- Enseignant: Xavier Genereux
- Enseignant: Max Barth
- Enseignant: Marie-Christine Jakobs
- Enseignant: Evi Sinn
- Enseignant: Ralf Zimmer