Winter 2021/2022
Bachelor / Undergraduate
Projektgruppen
Wissensentdeckung und Maschinelles Lernen
Bachelor BA-INF 051
In dieser Projektgruppe werden grundlegende Algorithmen aus den Bereichen Wissensentdeckung und Data Mining erarbeitet, diskutiert, implementiert und empirisch evaluiert.
Lecturers: Prof. Dr. Stefan Wrobel, Dr. Tamas Horvath, Dr. Pascal Welke, Florian Seiffarth
Contact: Florian Seiffarth
Supervisors: tba
Details
Time: Vorbesprechung am 13.10.2021 14:00-16:00
Participants: Maximal 12
Beschreibung:In den Projektgruppen werden grundlegende Algorithmen aus den Bereichen Maschinelles Lernen, Wissensentdeckung und Data Mining vorgestellt. Die Aufgabe der Studierenden ist es, in Kleingruppen jeweils einen Algorithmus zu erarbeiten und einen wissenschaftlichen Vortrag darüber zu halten. Im Anschluss soll der Algorithmus implementiert und evaluiert werden. Neben einem Abschlussvortrag soll eine schriftliche Ausarbeitung erstellt werden.
Kursmaterialien
Wichtige Termine
-
Vorbesprechung: 13.10.2021 14:00-16:00 (für beide Projektgruppen)
- Bitte registrieren Sie sich im ecampus Kurs bis zum 13.10.2021, um an der Vorbesprechung teilzunehmen. Wir werden dort einen Link und alle weiteren Informationen bereitstellen.
Master / Graduate
Lectures/Vorlesungen
Intelligent Learning and Analysis Systems: Machine Learning
Master: MA-INF 4111
Introductory course to machine learning.
Lecturers: Prof. Dr. Stefan Wrobel, Dr. Tamas Horvath
Exercises: Eike Stadtländer
Contact: Dr. Tamas Horvath
Details
Start date: 2021-10-22
Time: Fr 12:15-13:45
Place: online
Exercise time: Fr 14:15-15:45
Exercise place: online (perhaps on-site for one exercise class)
Malte Boßert (Exercise Class I)
Ben Rank (Exercise Class II)
Luca Scharr (Exercise Class III)
Description:
With more and more data available for analysis and decision making - from web documents and digital media to sensory data from cameras, microphones, and ubiquitous devices - it becomes increasingly more important to understand how such large volumes of data can be analyzed by computers and used as the basis for new intelligent services, for decision making, and for making computers learn from experience. In companies around the world, from retail and banks all the way to Google, intelligent learning and analysis techniques are used to improve business decisions. Likewise, in science, important discoveries are made easier by automated learning methods, and games and other artifacts are being made adaptive with learning technology.
Within the intelligent systems track of the computer science Master's program, intelligent learning and analysis systems are one of the two major topics. This module (Machine Learning) is one of the two modules that are offered as an introduction for master's students to Intelligent Learning and Analysis Systems. The other is the Data Mining module taught in the summer semester. Both modules can be selected in either order, and you may choose to attend one or both of them. For a complete introduction to the topic, it is recommended to attend both modules.
In the Machine Learning module in particular, we will give a practically oriented introduction into the most popular methods from Machine Learning as a subfield of Intelligent Learning and Analysis Systems. We will get to know decision tree methods, instance-based learning, artificial neural networks, probabilistic learning, regression methods, kernel methods and support vector machines, and reinforcement learning for intelligent agents. This will be complemented with lectures on the most important approaches within computational learning theory. Within the exercises, it is possible to try out the most important methods and popular Machine Learning systems.
Course Organization & Materials
- Materials are available via eCampus. You must enroll in eCampus on or before October 25 (Read how to access eCampus)
-
Important: Please read carefully the "orga_slides.pdf" uploaded to eCampus. If you are currently not in Germany and don't have a Uni-ID yet, please inform Eike Stadtländer via e-mail.
Important Dates
-
exam (1st try): 2022-02-11, 10:00-12:00 (ONLINE)
- exam (2nd try): 2022-03-25, 10:00-12:00 (ONLINE)
Learning from Non-Standard Data
Master: MA-INF 4303
This lecture introduces a selection of machine learning and data mining algorithms developed for graphs and relational structures.
Lecturer: Dr. Tamas Horvath
Contact: Dr. Tamas Horvath
Details
Start date: 2021-10-20
Time: Wed 14:15-15:45
Exercise time: Wed 16:00-17:30
Exercise place: likely online
Tutor: Malte Boßert
Description:
Traditional machine learning and data mining algorithms are resorted to data that can be represented by a single table of fixed width; the rows and the columns correspond to objects and object attributes, respectively. This assumption turns out to be quite restrictive in numerous practical applications involving structured data, such as graphs or relational structures. The lecture will cover the basics of learning and mining graph structured data and relational structures. We will present various algorithms for this setting and analyse their computational properties. We will also discuss some interesting applications in bioinformatics, computational chemistry, and natural language processing.
Course Materials
- Materials are available via eCampus. You must enroll in eCampus on or before 2021-10-21 (Read how to access eCampus)
- You are allowed to form excercise groups (of at least four and at most five students) if you know other students participating in this course. (We don't release the list of participants for privacy reasons.) In this case, please send an email with subject "LNSD2122: Excercise Groups" to Malte Boßert with the name (surname, given name) and email addresses of all group members before 2021-10-22, 23:59 o'clock. Otherwise, you will be added to some group at random and informed about your group via email on 2021-10-25. Please check your emails on 2021-10-25 evening for your exercise group.
Important Dates
-
exam (1st try): 2022-02-09, 10:00-12:00 (ONLINE)
- exam (2nd try): 2022-03-23, 10:00-12:00 (ONLINE)
Graph Representation Learning
MA-INF 4316
We will discuss general approaches for machine learning (ML) on graph structured data. In particular, computational methods for graph representation learning such as graph neural networks (GNNs), graph kernels, as well as graph mining techniques will be discussed, analyzed, and applied. Regarding GNNs and graph kernels, we will discuss the expressive power and how these concepts are related, as well as several specific examples. In the area of graph mining, we will likely investigate fast (approximate) algorithms to count small patterns, such as triangles, or trees. If time permits, we might venture into the realm of ranking on large-scale graphs, with applications such as recommender systems. The exercises will focus on practical implementations and the application of these methods to real world examples.
Lecturer: Dr. Pascal Welke
Contact: Dr. Pascal Welke
Exercises: Till Schulz
Details
Start date: 2021-10-11
Time: Mondays 14.15-15.45
Place: Hybrid (HS 5+6, Hörsaalzentrum Poppelsdorf, Friedrich Hirzebruch Allee 5)
Exercise time: Tuesdays 16.15-17.45
Exercise place: tba
Course Materials
I am planning to give this lecture in a hybrid fashion: I will be present in the lecture hall and will stream the presentation and audio live via zoom. You may join physically or virtually.
If you join physically, be prepared to prove 3G when entering the lecture hall building.
Important Dates
- Don't forget the course registration deadline in BASIS, which is usually in November.
- first exam: 21.2.2022 10h-12h (tentative!)
- second exam: 21.3.2022 10h-12h (tentative!)
Mining Media Data
The main focus of this introductory course to focus on established and state-of-the-art data mining methods suited to analyze media data that usually are collections of behavioral telemetry, transactional, textual, social network, financial as well as temporal datasets. The topics we will investigate are grouped into three categories including methods for affinity mining, latent pattern mining and predictive data mining to analyze media data. That is, we will start with the notion affinity mining covering some fundamental methods for frequent and high utility itemset mining as well as methods for association rule mining. Following this multiple scalable latent pattern mining methods will be covered to come up with interpretable ways of describing datasets. Finally, we will introduce the notion of predictive modeling, present a respective data mining taxonomy, study the methods for predicting minority entities and lay out details of a list of established algorithms. In addition to that, we will put emphasis on mathematical optimization, representation learning and model interpretability as well as give practical examples of the covered methods on a variety of (new) media applications.
Prerequisites and sequel course: Good programming skills and basic knowledge of linear algebra, statistics, AI, data science, machine learning, and pattern recognition. This course is intended for the students, who are planning to get an initial exposure to the methods of data mining and data science as well as their applications for analyzing media data. This course is strongly recommended for its sequel b-it course Mining Media Data II: Advanced Methods.
Lectures will be online (check the course website https://sites.google.com/view/bitmmd for related updates and the URLs to the lectures) every Thursday morning between 8:30 am - 10:00 am (CEST). Exercises will be on Mondays between 8:30 am - 10:00 am (CEST).
Study Period: Graduate / Hauptstudium
Lecturer: Dr. Rafet Sifa and Prof. Dr. Christian Bauckhage
Contact: Dr. Rafet Sifa
Details
Start date: 2021-11-04
Time: Thu 8.30 - 10.00
Place: online
Exercise time: tba
Course Materials
Materials are available here
Image Processing, Search and Analysis I
MA-INF 2314
This course will teach you the theoretical and mathematical basics of digital image processing and raster graphics editing. Starting with the technical foundations and hardware aspects of digital photography, we will talk about the mathematical representations of digital images as well as different coordinate systems and how to transform between them. We will talk about different kinds of filtering algorithms (low- band, and high pass filtering, mean- and Gaussian filtering, median filtering and morphological operations) and how to implement them efficiently. Moreover, you will learn how to perform image warping and morphing, learn the physiological foundations of color perception and apply color and intensity manipulations.
Lecturer: Prof. Dr. Christian Bauckhage
Exercises: Vanessa Toborek
Contact: Vanessa Toborek
Details
Start date: 2021-10-18
Time: Monday and Thursday 10.15-11.45
Place: online
Exercise place: online
Course Organization & Materials
- Are available via eCampus: https://ecampus.uni-bonn.de/goto_ecampus_crs_2310612.html
- Read how to access eCampus
- You must enroll in eCampus before October 11th.
Important Dates
- Don't forget the course registration deadline in BASIS, which is usually in November.
-
first exam: 2022-02-03 10.00
-
second exam: 2022-03-31 10.00
Explainable AI and Applications
MA-INF 4326
Most of the deep-learning systems are black-box systems and lack explainability and trustworthiness. In this lecture, I will start by introducing the achievements and limitations of large Deep-Learning systems, e.g. Watson, GPT, self-driving cars. Then I introduce two different neural approaches to syllogistic reasoning – one from the black box perspective, the other from the white box perspective. Background theories behind the two perspectives will be introduced, e.g., the dual-system theories (System 1 and 2), laws of cognition. The state-of-the-art XAI approaches will be applied for the understanding of the black-box neural system. The white-box system will be followed by an introduction of spatial representation and reasoning, and applications, e.g., knowledge graph reasoning, computational humor.
Lecturer: Dr. Tiansi Dong
Contact: Dr. Tiansi Dong
Exercises: Dr. Tiansi Dong and Siba Mohsen
Details
Start date: Oct. 18, 2021
Time: Monday 12.30 - 14
Place: online
1st Exam: Feb 25, 90 Minutes within the time period from 11:00 to 14:00, location: Friedrich-Hirzebruch Allee 5 - Hörsaal 2
Exercise place: online
Course Materials
Important Dates
- Don't forget the course registration deadline in BASIS, which is usually in November.
- first exam: tba
- second exam: tba
Seminars/Seminare
Principles of Data Mining and Learning Algorithms - Selected Papers from EMNLP
Master: MA-INF 4209
In this seminar we will focus on a selected few papers from the Empirical Methods in Natural Language Processing (EMNLP) conference, which is one of the leading conference in the area of natural language processing and artificial intelligence. The seminar consists of one session on how to give a talk at a conference, a discussion meeting, and one final presentation meeting, which will simulate a scientific conference with paper presentations.
Lecturers: Prof. Dr. Stefan Wrobel, Dr. Tamas Horvath, Dr. Pascal Welke, Ewald Bindereif
Contact: Ewald Bindereif
Details
Time: (see below)
Place: online
Participants: Max. 6
Course Materials
- Are available via ecampus
Important Dates
-
preliminary meeting: 2021-10-13 14:00h - 15:00h (online)
-
seminar meetings: tba
Principles of Data Mining and Learning Algorithms: Reading Seminar
Master: MA-INF 4209
In this seminar we will read several selected topics concerning machine learning and data mining theory and applications from the deep learning book by Goodfellow, Bengio and Courville. We will meet every two weeks via Zoom to summarize the topic from two weeks before and present a new topic.
Lecturers: Prof. Dr. Stefan Wrobel, Dr. Tamas Horvath, Dr. Pascal Welke, Fouad Alkhoury
Contact: Fouad Alkhoury
Details
Start date: 2021-10-25
Time: Mon 12:15 - 13:45 (every two weeks)
Place: online
Participants: Max. 16
Course Materials
Important Dates
-
preliminary meeting: 2021-10-13 14:00h - 15:00h (online)
-
seminar meetings: Mon 12:15 - 13:45 (every two weeks)
Labs/Praktika
Lab Development and Application of Data Mining and Learning Systems:
Data Science and Big Data
Master: MA-INF 4306
In this lab, machine learning and data mining techniques are implemented and used in a wide range of applications.
Lecturers: Prof. Dr. Stefan Wrobel, Dr. Tamas Horvath, PD Dr. Michael Mock, Dr. Pascal Welke
Contact: Sebastian Müller
Supervisors: Michael Mock, Sebastian Müller
Prerequisites: MA-INF 4212 highly recommended
Details
Time: (see below)
Place: online
Participants: Max. 6
Course Materials
Important Dates
-
Please register in the ecampus course until 2021-10-18, to be able to attend the preliminary meeting. We will make a link available there.
-
Preliminary Meeting: 2021-10-19 13.00h - 15.00h (online).