Seminar Assignment for Winter 2022/2023
The central registration for all computer science seminars will open on September 26th.
This system is used to distribute students among the available seminars offered by the CS department. To register for any of the seminars, you have to register here until October 26th 23:59 CET. You can select which seminar you would like to take, and will then be automatically assigned to one of them on October 30th.
Please note the following:
The assignment will be automatically performed by a constraint solver on October 30th. You will be added to the respective seminars automatically and be notified about this shortly thereafter. Please note that the assignment cannot be optimal for all students if you drop the assigned seminar, i.e., make only serious choices to avoid penalty to others.
Computer vision has led to many recent technology break-throughs and is currently one of the most demanded fields. Due to the statistical and complex nature of the world, it is also one of the hardest disciplines. Deep learning has proven to be the method of choice and nurtured the successes.
However, most deep-learning approaches for computer vision are constructed in 2D. One can argue that they do not understand the 3D world and have inherent limitations. Therefore, 3D computer vision is already becoming increasingly more important and will likely lead the next generation of algorithms.
The seminar will bring you up to speed with the concepts and state-of-the-art literature of 3D representations and 3D reconstruction with deep learning. After few introductory lectures, the seminar will continue with presentations to review the most important and most recent papers in the field, covering SDF- and occupancy-based representations as well as neural radiance fields (NeRFs). Overall, the seminar will set you up to be familiar with distinguished literature and enable you to start research work in the field.
Every student is expected to give a 45min presentation with a write-up and an implementation.
The seminar is offered by the new Computer Vision and Perception Lab (https://cvmp.cs.uni-saarland.de/) that focuses on building the next generation machine perception algorithms, which are not rigid but able to adapt to their environment and evolve. The lab is currently offering Master's and PhD positions.
Requirements: This seminar will focus on state-of-the-art research and is for advanced students who are already acquainted with machine learning. In particular, a basic understanding of projection and 3D geometry and prior experience in convolutional neural networks, as well as hands-on implementation of neural networks with pytorch is required. It is recommended to have attended the High Level Computer Vision Lecture. Prior attendance of Computer Graphics may be helpful but is not required.
Millions of users benefit from modern secure messaging apps. Since recent years, these messaging apps can offer much stronger security properties than they ever did before, for example using the Signal protocol library (as used by WhatsApp, Signal's app, and Facebook's secret conversations) and others.
In this seminar, we will explore some of the theory behind the design of these protocols through a selection of relevant research papers in this area, which is a highly active current research direction. This will include papers on the novel security properties met by these protocols, the subtle edge cases of security in group messaging, as well as recent works that show the achieved security might be lower than expected.
Most (but not all) of the papers are technical, and include security models and proofs; however, for the presentations, understanding the proofs will not be mandatory. Having attended advanced cryptography or protocol analysis lectures can be beneficial, but is not strictly required.
More background on the instructor can be found here: https://cispa.saarland/group/cremers/index.html
Requirements: Students should have completed the core Security or core Cryptography lectures, ideally both.
The recent advances in AI and other computational methods have opened new opportunities for users to interact with computing systems, creating entirely new interaction technologies, enabling novel ways to understand, model, and predict user behavior and changing even how interfaces are designed.
In this seminar we will learn about specific algorithms and AI methods from machine learning, optimization, bayesian theory, and other AI-related areas and see how they are applied to Human-Computer Interaction. This seminar is highly interactive and builds on the participation of all students. Every week you will read a paper from the HCI domain and prepare a set of discussion questions. In addition, students take turn in assuming different roles, acting as teacher, journalist or PhD student. See the seminar page for more details: https://cms.sic.saarland/ai_hci_22_23/
This seminar is held together with the proseminar AI for HCI.
Requirements: Experience with at least one of the following topics is beneficial (please indicate):
- Reinforcement learning
- Deep learning methods (e.g. CNN, RNN, Autoencoder, etc. )
- Natural language generation
- Optimization methods (e.g. integer programming, simulated annealing)
- Bayesian methods (e.g. bayesian inference, bayesian optmization)
- Statistical modeling (e.g. gaussian mixture models)
In this seminar, we discuss recently published algorithms for alignment free biological sequence analysis, mainly on DNA sequencing data. Starting with an original paper and using material from the lecture "Algorithms for Sequence Analysis" as background knowledge, participants shall explain a recent method in detail with all necessary additional background.
This seminar focuses on the understanding of algorithms (and their well-engineered implementations). The material can be difficult, and the seminar is not recommended for the casual student.
Both a presentation and a written report are required. An implementation can also be attempted or an existing implementation can be studied on public datasets to check whether the claims made in the paper are true.
A kick-off meeting will be held in the first week of the winter semester. A first draft of the slides and the report are required before the holiday break around Christmas. We will do a peer review, i.e. other students will write a review of your drafts and vice versa. Presentations will be late in the lecture period (the exact dates will be doodled). The final reports are required by the end of the lecture period.
Requirements: Successful completion of "Algorithms for Sequennce Analysis" lecture, ideally with a good grade (good knowledge of the material is required).
In that case, the seminar is suitable for students of Bioinformatics or Computer Science.
Note: A seminar gives 7 CP corresponding to 210 hours of work. This means, you should plan to work approximately 14 h per week every week on average on the seminar material during the lecture period, as there is no work after the lecture period. If you are not ready to invest this amount of time, do not pick this seminar.
Fully-Homomorphic Encryption (FHE) schemes and Multi-Party Computation (MPC) are fundamental tools in modern cryptography.
For decades FHE and MPC schemes have been abstract concepts living in the realm of cryptographic theory.
In recent years those systems have seen major improvements in terms of efficiency and practicality.
In short, FHE and MPC have become practical enough to be considered for applications in private delegation of machine learning models and applications to privacy-preserving distributed Genome-wide association studies.
This seminar is concerned with currently implemented applications and practical aspects of FHE and MPC.
By the end of the seminar, participants should possess fundamental knowledge about FHE and MPC and should know the
state-of-the-art of libraries and developer tools that are available nowadays.
In particular, participants should have an overview of what is currently possible to achieve and at what cost using FHE and MPC techniques.
Seminar Web page: https://cms.cispa.saarland/semifhe/
Requirements: During the seminar, students are required to read, understand and present advanced cryptography papers.
Hence, it is required that a student has completed a cryptography course and is fluent in linear and abstract algebra.
Are you passionate about logic, verification, semantics, and alike? Are you tired of thinking black-and-white, true-or-false? Then come and study the more nuanced quantitative formal program verification! In quantitative verification, properties are not just true or false. Instead, we verify quantities like runtimes, error probabilities, beliefs, etc.
Topics which we will cover include:
- Probabilistic programming (a currently trending modeling paradigm in machine learning),
- The geometry of neural networks
- Incorrectness logic (the latest creation of the former chief formal methods researcher at Facebook)
- The flow of quantitative information through programs
- Worst-case execution times
- Verification of heap-manipulating programs
- and many more
The seminar website can be found here:
Requirements: The most important requirement: You should really really like math and/or logic. This seminar covers very theoretical work.
The following courses are mandatory and/or recommended.
Mandatory: Programmierung 1, Programmierung 2, Grundzüge der Theoretischen Informatik
Recommended: Semantics; Introduction to Computational Logic; Automata, Games and Verification;
Deep learning is the predominant machine learning paradigm in natural language processing (NLP) and beyond. This approach not only gave huge performance improvements across a large variety of natural language processing, computer vision and other tasks, it also allows to integrate external knowledge sources. The dominance of deep neural networks also brought different fields closer together as many approaches and ideas are shared.
This year, there will be a special focus on data augmentation for DNNs.
For more information see: https://www.lsv.uni-saarland.de/block-seminar-machine-learning-for-natural-language-processing-spring-2023/
Reinforcement learning is a popular technique to solve decision-making problems. In combination with artificial neural networks, it has achieved huge successes in challenging domains such as mastering Atari Games, Chess, or GO.
In this seminar, we will discuss the basic concepts of deep reinforcement learning (DRL) as well as the details of popular DRL algorithms. In the final phase of the seminar, the participants will apply these algorithms to a simple race car simulation.
Each participant will give a short presentation and submit a write-up. In addition, a programming project will be solved in small groups.
In parts, this seminar will take place jointly with the reinforcement learning proseminar, which will cover the basics of classical reinforcement learning.
The kick-off will take place Wednesday, November 2, 10.00 am.
Requirements: Background knowledge in statistics and machine learning.
From critical infrastructure to consumer electronics, embedded systems are all around us and underpin the technological fabric of everyday life. As a result, the security of embedded systems is crucial to us.
This seminar covers research papers addressing various topics in embedded systems security. This includes topics such as instruction profiling (template attack), fault injection and side-channel attacks, firmware static and dynamic analysis, intrusion detection in embedded systems, automotive/space systems security, and fuzzing embedded systems.
In this seminar, students learn to work independently on an assigned single topic consisting of multiple papers. Each week, one or two students present a paper on the topic of interest, followed by a discussion of the leading paper. All participants are expected to participate in the discussion by asking questions actively.
Students must send questions about the main paper the day before the seminar.
Requirements: This seminar aims primarily at master students in Computer Science/Embedded Systems. Previous experience in embedded systems and computer security is beneficial.
In this seminar you will plan and conduct a user study in the context of games and game culture. We have 16 spots in this seminar. In groups of four (the groups are formed in the kick-off), participants will - based on the related research in this area - decide on the research question they want to explore and develop a research methodology on how to investigate this question.
All (!) information are presented here upfront: https://umtl.cs.uni-saarland.de/teaching/winter-2022/2023/seminar-games-user-research.html
Please read the page carefully before deciding to pick this seminar.
We gave this seminar in 2018 and according to the students, it was a great experience. We hope to replicate this in this term as well.
As you have seen on the page, based on the nature of tasks to do, the seminar covers the lecture and lecture-free period. We meet irregularly on Thursdays, 12-2pm (12-14 Uhr).
Requirements: You should have passed the HCI lecture (or comparable, if you are coming from a another university). Ideally, you also have passed the Statistics with R lecture.
While the Internet has started as a research effort, it has consistently evolved throughout the decades into the largest commercial network. Today, many research efforts focus on understanding the Internet's structural trends, optimizing its packet delivery, and laying the foundations for its future developments. In this seminar, you will receive a closer look at bleeding-edge research published at the top conferences in our domain. Discussing state-of-the-art approaches and recent findings from a broad range of network-related topics with your peers and instructors will provide you with a deep understanding of your assigned topic. Preparing the accompanying research survey and topic presentation will not only strengthen your academic writing but also help you to present your future work in a more accessible and structured way.
Requirements: Participants should have successfully participated in "Data Networks" or an equivalent course.
This seminar (HyLEAR) is concerned with selected hybrid intelligent systems that combine techniques from subsymbolic learning (deep learning within area of machine learning) with symbolic techniques for reasoning, planning or learning. Both types of AI techniques have their strengths and limitations.
While deep learning systems have been quite successfully applied to, for example, pattern recognition, image interpretation, speech recognition and translation, they can be characterized as overly data hungry, susceptible to adversarial attacks, opaque (non-interpretable by humans), and not informed by general principles such as causality or commonsense and domain knowledge. The successes of symbolic reasoning techniques (“good old fashioned AI”) are in such applications as automated (human-understandable, traceable) planning, diagnosis, design tasks, and question answering by cognitive virtual assistants but are often quite limited by the need of expensive, explicit knowledge acquisition and modelling, inefficient logic-based reasoning, and instability in the presence of noisy data. There is a consensus in the AI community that symbiotic, profound integration or combination of machine learning and reasoning is essential for human-level AI in general.
In this seminar, we will take a closer look at selected techniques and systems for hybrid learning and reasoning or planning, and discuss their strengths and weaknesses.
The seminar is held on Wednesdays from 10:15am to 12:00.
First session (introduction with topic assignments) on Wednesday 9.11.2022
The seminar takes place at DFKI Saarbrücken, SIC Bldg. D3.2, in room “Leibniz” (please ask for directions at DFKI reception desk) or virtually via Zoom. Please check the seminar webpage frequently for changes.
Requirements: This seminar aims primarily at advanced master students in Computer Science who preferably hold a B.Sc. degree in this or related field.
Good knowledge in AI -- symbolic knowledge representation and reasoning, machine learning (symbolic and deep learning), planning -- required.
Industrial control systems (ICS) are complex systems able to monitor and control an industrial process autonomously. ICSs are designed for safety and availability; however, security is recently being considered, given the number of cyber and physical threats that are menacing the ICS space. A wide range of attack and defence proposals are presented for ICS security, and in this seminar, we will review the state-of-the-art attack and defence scenarios.
The seminar takes place virtually via Zoom. Please visit the CMS webpage for the most recent news.
Inpainting has been introduced as a technique to restore missing or deteriorated regions by using information from within the image. There are various ways to fill these missing parts such as simple interpolation, diffusion-based methods, or by comparing image patches. Nowadays, specialised inpainting methods can remove objects seamlessly or can even reconstruct an image using only a tiny fraction of the original information. In this seminar, we firstly cover basic concepts and move on to forefront research in the field of inpainting.
Further information on the seminar webpage: https://www.mia.uni-saarland.de/Teaching/iia22.shtml
Requirements: The seminar is for advanced bachelor or master students in Visual Computing, Mathematics, or Computer Science. Basic mathematical knowledge (e.g. Mathematik für Informatiker I-III) and some knowledge in image processing and computer vision is required (IPCV is necessary, DIC is useful but not required).
As robots are becoming pervasive, a question of ever increasing importance is how people can best interact with robots, to solve tasks intuitively and effectively. Within the past few years, the wide availability of accessible hardware platforms for prototyping and digital fabrication tools have made it easier to develop interactive robotic systems. A variety of robotic devices and applications are now finding their way to end-users. Robots have the potential to enhance human performance, augment the human body and offer unique new opportunities for interaction with our physical environment and other people. This course focuses on state-of-the-art technologies for human-robot interaction alongside key questions regarding how to interact with robots in our day-to-day life.
The goal of this seminar is to acquire basic conceptual and practical skills in developing an interactive robotic system for human use. Participants will work in small groups on a practical project. They can choose either a software-focused or a hardware-focused topic. Software projects will use existing robotic hardware to design and implement innovative human-robot interactions in software. In contrast, hardware projects will focus on developing a visionary robotic hardware system, e.g., with interactive e-textiles or 3D-printed meta-materials.
For further information please refer to our course page:
Requirements: - For the registration, we require you to submit a brief motivation statement that elaborates on why you want to take this class. Please indicate if you prefer to do a software project or a hardware project, or any of them.
- For our hardware-related topics, you are required to have basic experience in hardware programming (e.g., Arduino) and simple electronics. It's advantageous but not strictly required to have passed Interactive Systems.
- For our software-related topics, we don't expect any prior experience with hardware.
- This is a Master-level seminar, which can also be taken by Bachelor students from the 5th study semester and above.
While deep learning based Natural Language Processing (NLP) has achieved great progress during the past decade, one of the major bottlenecks in training deep neural networks (DNNs) is the requirement of substantial amounts of labeled training data. This can make deploying NLP models to real-world applications challenging, as data creation can be costly, time-consuming and/or labor intensive.
In the last years, there has been an increased interest in building NLP models that are less data-demanding and significant progress has been made there. Recent advances in this field enable efficient learning from just a handful of samples (few-shot learning). In addition, large-scale pre-trained language models can often achieve non-trivial performance on unseen NLP tasks (zero-shot learning).
This seminar aims to provide a board and up-to-date overview about recent progress on zero- and few-shot learning in NLP. In particular, we will study recent papers to understand the challenges of learning from limited data and how to leverage pre- trained language models to make efficient learning in low-resource settings possible.
For more information see: https://www.lsv.uni-saarland.de/learning-from-limited-data-in-nlp-seminar-wese-2022-23/
Requirements: Elements of Machine Learning or core course ML or Neural Networks.master students only.
Das Seminar Legal Tech und eJustice ist ein interdisziplinäres Seminar für Informatiker und Juristen.
Ablauf des Seminars:
Studierende erhalten während der Vorbesprechung ein Thema und müssen eine Seminararbeit hierzu anfertigen. Vor der Abgabe der fertigen Seminararbeit ist die Einreichung eines Abstracts vorgeschrieben. Nach der Abgabe des Abstracts kann bei Bedarf bei den Betreuern Feedback zu diesem eingeholt werden. Danach ist die Abgabe eines „Preprints“ der Seminararbeit erforderlich. Dies ist ein Vorabgabe der finalen Seminararbeit. Daraufhin findet eine Review der abgegebenen Preprints durch die weiteren Teilnehmer des Seminars statt. Jedem Studierenden werden hierfür 3-4 Paper anderer Studierender zugeteilt, welche von Ihnen reviewed werden. Die Review jedes Papers muss sich inhaltlich auf das gesamte Paper erstrecken und mindestens 400 Zeichen (für jedes Paper) enthalten. Studierende erhalten somit schon vor der endgültigen Abgabe der Seminararbeit Feedback von den anderen Teilnehmern des Seminars. Nach den Reviews finden die Vorträge statt. Abschließend werden die fertigen Seminararbeiten abgegeben.
Die Gesamtnote ergibt sich aus dem Preprint, den verfassten Reviews zu den Papern der anderen Studierenden, dem Vortrag und der finalen Seminararbeit.
Die Nichtabgabe des Abstracts, des Preprints, der Reviews oder der finalen Seminararbeit sowie ein Nichterscheinen zum Vortragstermin führt zum Nichtbestehen des Seminars.
Vorbesprechung und Themenvergabe: 09.11.2022, 18.00 Uhr (s.t.)
Anmeldung zur Besprechung der Abstracts: 27.11.2022
Abgabe Abstracts: 04.12.2022
Abgabe Preprint: 04.01.2023
Abgabe Reviews: 15.01.2023
Vorträge: tba (KW 5 2023)
Finale Abgabe der Seminararbeiten: 19.02.2023
Requirements: Es wird erwartet, dass Teilnehmer in der Lage sind, Vorträgen in deutscher Sprache zu folgen.
Program analysis is an interdisciplinary topic encompassing programming languages, formal methods, and software engineering, with the overarching goal to build algorithms and methods to find unwanted program behaviors like vulnerabilities and bugs. Program analysis techniques are not perfect, and their shortcomings have limited their impact on modern programs. Over the past years, machine learning has shown enormous potential in solving many complex tasks, drawing the attention of the program analysis community and, hopefully, program analysis. The research community has started investigating applications of machine learning to program analysis tasks, and this seminar intends to explore this nascent research area.
This seminar covers research papers addressing key challenges at the intersection of these two fields. How do we represent code for ML models? What tasks can ML solve? Can we use classifiers as vulnerability detectors? This seminar will explore these research questions and more.
In this seminar, students learn to work independently on an assigned single topic consisting of multiple papers. Each week, one or two students present a paper on the topic of interest, followed by a discussion of the main paper. All participants are expected to participate in the discussion by asking questions actively. Moreover, students are required to send questions about the main paper the day before the seminar.
More info: https://cms.cispa.saarland/ml4pa/
Requirements: The course has no formal requirements but preference will be given to students with background in machine learning. Background in program analysis is not required but it is beneficial.
Wouldn't it be cool if one could teach computers to help developers in coding, testing, and debugging software? In this seminar, we will discuss current results and new problems by applying techniques from machine learning to software development, based on relevant scientific papers. We will explore techniques such as
* Generating software tests
* Generating test oracles
* Generating graphical user interfaces
* Code search and code fill
* Locating faults
* Program repair using neural networks
* ... and more!
The general process will be as follows: Each week, you get 1-2 reading assignments and write an abstract about them. You may also be asked to give an (ungraded) five-minute short presentation to kick off the discussion and improve your presentation skills. At the end of the seminar, you give a 15-20 minute presentation on one of the techniques, preferably including small experiments or demonstrations on how well they work; these will then be graded.
More details can be found at https://cms.cispa.saarland/mlse_2223/
Requirements: Prior knowledge in machine learning and software engineering will be beneficial. For experiments and evaluations, programming knowledge will be helpful, too. We recommend doing experiments and evaluations in Jupyter Notebooks – so don't be afraid of them.
With the deployment of 5G and time-sensitive networks, communication systems are now able to support more demanding applications than ever. Providing such strict performance guarantees (i.e., predictable delay and reliability) comes at the cost of increased complexity, since network functions must be more reactive to adverse conditions. This opens many opportunities to apply Machine Learning (ML), which has proven to be very valuable in solving complex problems by finding patterns in large collections of data. Deep reinforcement learning, imitation learning, binary neural networks, transfer learning, and other techniques can be applied to solve the challenges of providing a predictable, high quality-of-service in telecommunications. This seminar gives an overview of ML solutions in different layers of the protocol stack. The discussion of proposed solutions in scientific papers aims to shed light on the questions of what the benefit of ML techniques is, why ML replaces algorithmic solutions, and what metrics are used to justify the use of ML.
Requirements: It's advantageous but not strictly required to have passed Digital Transmission & Signal Processing (DTSP) or Audio/Visual Communication and Networks (AVCN).
You should be ready to read and review three scientific papers.
The intersection between security and machine learning can be viewed from two perspectives: The security of machine learning algorithms and systems, e.g., adversarial examples and poisoning attacks. Second is the use machine learning methods to improve and analyze the security of a system, e.g., malware detection or decompilation. In this seminar, we will cover recent publications from both sides by reading and summarizing the state-of-the-art on these two topics and performing an artefact evaluation of their code to verify and comprehend the practical implementations of the latest scientific publications.
The seminar is structured into two parts. In both parts, you will work in groups of two:
- you will write a short survey paper on the main topic of your assigned paper.
- you will evaluate the code of the paper during an artefact evaluation.
Additional information can be found at https://cms.cispa.saarland/ml_sec_sem/
Requirements: This seminar aims primarily at master students in Computer Science or related fields. Previous experience in machine learning and computer security is beneficial.
Spectre, Meltdown, and other microarchitectural attacks have been in the limelight in recent years. These attacks exploit subtle timing and behavioral differences of processors that are caused by microarchitectural optimizations such as caches and speculative execution to gain access to secret information.
The vulnerabilities exploited by microarchitectural attacks are not captured by today's hardware-software contracts, i.e. instruction-set architectures (ISAs). Traditionally, ISAs only capture the "functional" behavior of a system and thus have a blind spot when it comes to side channels. Recently, there has been a push to augment conventional ISAs with a formal specification of information leakage, resulting in more general hardware-software contracts. Such contracts enable writing secure code, e.g. implementing cryptographic algorithms, in a rigorous manner.
In this seminar, we are going to study
- novel hardware-software contracts that capture microarchitectural vulnerabilities,
- verification of hardware-software contracts,
- fuzz testing of hardware-software contracts,
- techniques to automatically synthesize hardware-software contracts from hardware models, and
- techniques to analyze security properties of software on top of contracts.
Each participant will give a presentation of an assigned paper, followed by a group discussion. All students are expected to read each paper carefully and to actively participate in the discussions. Each student will write a summary of the paper they have presented, including a general overview of the topic and reflecting the group discussion.
The seminar website can be found here: https://cms.sic.saarland/contracts2223/
Requirements: Basic knowledge of computer architecture (e.g. due to Systemarchitektur) is required. Knowledge of security and formal methods is a plus.
Machine learning has witnessed tremendous progress during the past decade, and data is the key to such success. However, in many cases, machine learning models are trained on sensitive data, e.g., biomedical records, and such data can be leaked from trained machine learning models. In this seminar, we will cover the newest research papers in this direction.
Requirements: Students are required to have basic knowledge of data mining and machine learning.
Deep learning has achieved major breakthroughs in a variety of tasks. Yet, it comes at a considerable computational cost, which is exaggerated by the recent trend towards ever wider and deeper neural network architectures. Instead, many problems can be solved with the help of extremely sparse neural network architectures but finding and training them is a non-trivial task. According to the recent lottery ticket hypothesis, such sparse architectures can be identified by pruning large randomly initialized neural networks. In this seminar, we will present recent algorithmic advancements in this direction, gain theoretical insights into the existence of lottery tickets, identify open problems, and discuss common challenges in the quest for winning lottery tickets.
-Kick-off meeting at 16:00 on Monday, November 7th, 2022 (if feasible).
-We will have a block course in the spring break (date and time tbd at kick-off meeting) with one presentation and discussion of a topic per person.
-Note that some (light load of) written deliverables will be due before the block course meeting and a seminar paper one week after the meeting.
More information can be found here:
Requirements: The course has no formal requirements but preference will be given to Master students in Computer Science and related fields with prior knowledge in deep learning.
Monte Carlo ray tracing is a popular technique to render realistic images. It is used for movies, architecture, video games, and product design. This seminar looks at a broad range of methods to make rendering via Monte Carlo ray tracing more efficient. We will look at rendering algorithms, from a basic path tracer to photon mapping, at sampling and material models, Markov chains and path guiding, and more. Each student will give a presentation on an assigned paper, write a short summary, and reproduce the key idea in a simplified setting.
Requirements: Basic understanding of computer graphics is assumed, for example through our Computer Graphics core lecture.
Are you interested in the current hype around space activities such as emerging satellite mega-constellations in low-Earth orbit or scheduled interplanetary missions?
Be aware that these are no longer aerospace-specific topics. Modern space systems are deeply rooted in recent advances in computing and miniaturization promising massively increased scale, complexity, and autonomy. This, in turn, poses unprecedented challenges to the informatics domain, where state-of-the-art algorithms and techniques spanning optimization, verification, and machine learning need to be applied and adapted to the space context. We will approach the contents of Space Informatics in three phases: 1) Physics, 2) Technology and 3) Informatics.
The seminar will run in full virtual mode. Periodic synchronous Zoom encounters and asynchronous interactions via Discord will be leveraged. Students will work on reports of assigned topics in the form of draft book chapters, including sample exercises, with support from the instructor. Topics will be presented and revised by other seminar students.
Requirements: There is no specific background required, but students are expected to be open to investigating orbital mechanics models, as well as space technology aspects like power and communications.
Personalized user interfaces are a core AI topic and relevant across all domains - from web interfaces and support chatbots, over home automation and driver assistance systems, to industrial robots and medical support utilities. The task of Artificial Intelligence is to ensure that such systems not only provide accurate information or functionality, but that it is relevant to their users and in the current situation, and presented in a way that fits its purpose. This process is realized mainly through knowledge about the user, available in either symbolic (e.g. preferences) or subsymbolic (e.g. eye-gaze data) manner, and algorithms which use this knowledge to realize the adaption (e.g. to provide proactive assistance). For example, a robot which can perform a particular task could optimize its interaction with a user who requests the task if it knows about the typical preferences of the user, and keep learning based on further feedback.
In this practical seminar, we will look at the different AI components that make up a system that can continuously adapt to the user:
(1) Acquisition of knowledge about the user, primarily focused on Natural Language (speech) input
(2) Presentation of a personalized user interface (can be pseudo-functional)
(3) Continuous learning algorithms for adapting the user interfaces
You will work both on an INDIVIDUAL topic (talk) and on practical GROUP projects, where each group looks at a speech-enabled adaptive system from a different application scenario, including the domains listed above. Grading criteria are an individual presentation, practical group work results, and active participation.
*** The seminar presented by the UMTL chair of Prof. Antonio Krüger will take place TUESDAYS 10-12h as an ONLINE class in ENGLISH (7 CP). ***
More info: https://umtl.cs.uni-saarland.de/teaching/winter-2022/2023/seminar-speech-based-adaptation-of-personalized-user-interfaces.html
Requirements: This course assumes basic programming skills. This class covers multiple facets of AI - knowledge in at least one of (1) user interface design, (2) machine learning, or (3) natural language modeling is recommended (not all 3 are required!).
If your major is Cybersecurity, you might not be able to grade in this seminar - check your study regulations first!
In this seminar, students will learn to present, discuss, and summarize papers in different areas of Web security. The seminar is taught as a combination of a reading group with weekly meetings and a regular seminar, where you have to write a seminar paper. Specifically, each student will get a single topic assigned to them, consisting of two papers (a lead and follow-up paper).
For the weekly meetings, all students have to have read the lead paper and must state at least three questions before the meeting. In the meeting, the assigned student will present the follow-up paper (20 minute presentation + 10 minute Q/A). Afterward, the entire group will discuss both papers.
Moreover, each student will write a seminar paper on the topic assigned to them, for which the two papers on the topic serve as the starting point.
The seminar slot is fixed on Monday 2-4pm.
For more details and the timeline, see https://cms.cispa.saarland/websecsem22/
Note that we will *not* offer a hybrid solution. We plan to have in-person meetings as long as possible and switch to fully online if the need arises.
Requirements: A basic understanding of Web Security, e.g., from having taking Foundations of Cybersecurity 1, the Security Core Lecture, Foundations of Web Security, or Secure Web Development is beneficial.
Optimization lies at the heart of many machine learning algorithms. This seminar teaches an overview of modern mathematical optimization methods, for applications in machine learning and data science. In particular, we will discuss the theoretical basics of stochastic optimization, scalability of algorithms to large datasets, and challenges in distributed optimization, such as for instance in decentralized or federated machine learning. We will cover a set of foundational papers, but also a selection of recent publications.
- There is no weekly meeting. The presentations will take place in (6-7) x 2 hour slots, roughly between end of November - January.
- A kick-off meeting will be held virtually during the first week of the semester.
Note that - besides the block format - there are some written deliverables due during the semester, as well as a (mandatory) meeting with the tutor to get feedback on your report and presentation slides.
Additional information can be found at https://cms.cispa.saarland/optml_seminar_22/.
Requirements: This seminar aims primarily at master students in Computer Science or related fields. Previous experience in machine learning, data analysis, or optimization is beneficial.
The ability to acquire knowledge through learning and adapt it quickly to new environments is the hallmark of intelligence. Despite recent successes of machine learning (ML) based models such as Deep Neural Network (DNN), Transformer, GPT-3, and DALL-E, they still lack the ability to generalize out of training distributions. Traditionally, most ML algorithms were developed under the identically and independently distributed (i.i.d) assumption, i.e., test data come from the same distribution as training data. As these models are increasingly trained and deployed across heterogeneous and potentially massive environments, the i.i.d. assumption is almost always violated in practice. This problem significantly limits the scope of real-world applications of machine learning.
In this seminar, we will explore the research frontier that aims to push machine learning beyond the i.i.d. setting. We will study state-of-the-art theories and algorithms that enable machine learning models to generalize out of distribution (OOD). Topics of interest include deep learning, causality, meta-learning, reinforcement learning, domain adaptation, domain generalization, and robustness.
Requirements: This seminar aims primarily at master's students in Computer Science or related fields. Students are expected to know the basics of machine learning.
Most recent software systems offer the end-user many knobs to personalize the software system. The sheer size of different configurations that arise from combining different knobs leads often to unforeseen behavior of the software system. Even understanding the impact of changing a single knob on the behavior, e.g., runtime, reliability, or robustness, of the system imposes a major challenge.
In this seminar, we address and discuss possible solutions and trade-offs, theoretically, by addressing the state of the art in research regarding black-box, white-box, and causal analyses.
In this seminar, each participant has to perform a literature search and present the state-of-the-art in research on configurable software systems.
Subsequently, the topic and the results of the literature search have to be incorporated into a presentation and a written thesis.
To aid the literature search and the presentation, this seminar includes multiple preparatory sessions at the beginning of the semester.
All sessions will take place on-site at the university (under the caveat that the pandemic situation admits in-person sessions) on Thursdays 12:15 PM - 2:00 PM.
Participation to all sessions is mandatory.
The topic assignment will take place on Thursday November 03, at 12:15 PM. Further information will be provided via e-mail after registration.
Requirements: Basic knowledge on software engineering
Most of our information-processing systems nowadays are distributed - be it large-scale data centers that replicate data in different physical locations, sensor networks that collect data from different places, or simply multi-threaded programs that run on different cores of your CPU. Due to their composition from multiple interacting components, ensuring correctness of these systems is a major challenge. In many application areas, faulty behavior or even short outages of these systems can have severe consequences. Therefore, both industrial and academic research has in recent years developed a range of new methods to formally guarantee certain properties of a given distributed system.
In this seminar, students will learn to present, discuss, and summarize research papers that aim at formalizing and verifying distributed systems. The seminar is split into two parts. The first part will take the form of reading sessions, where we lay the foundations of the topic. For the second part, each student is assigned a recent paper from the research area. Students will present their paper and will write a seminar paper on the topic assigned to them, taking into account connections to the topics discussed in the seminar.
Requirements: Interest in formal methods and basic knowledge about concurrent or distributed systems. Some knowledge of formal methods (e.g. Verification or Automated Reasoning) is recommended.