Large language models (LLMs) exhibit remarkable reasoning abilities, allowing them to generalize across a wide range of downstream tasks, such as commonsense reasoning or instruction following. However, as LLMs scale, inference costs become increasingly prohibitive, accumulating significantly over their life cycle. In this seminar we will dive into methods like quantization, pruning and knowledge distillation to optimize LLM inference. Please fill this interest form to participate in the seminar.

Course type:	Seminar
Time	Five slots, to be determined with all participants. Kick-off is likely on the 24th of October from 2-3pm
Location	in-person; SR 04-007, building 106
Organizers	Rhea Sukthanker , Arbër Zela , Mahmoud Safari
Registration	Via HISinOne (maximum nine students, registration opens 14th of October)
Language	English

---

Prerequisite

We require that you have taken lectures on or are familiar with the following:

• Machine Learning
• Deep Learning
• Automated Machine Learning

---

Organization

After the kick-off meeting, everyone is assigned a paper (one or two depending on the content). Then, everyone understands the paper(s) assigned to them and prepares two presentations.

• The first presentation will focus on establishing, the background, motivation for the work and a concise overview of the approach proposed in the paper
• The second presentation will focus on the details of the approach, the results and takeaways from the paper and an “add-on” described below

Students will contribute an "add-on" related to the paper for the final report. This includes but is not limited to a thorough literature review, reproducing some experiments, profiling inference latency of the LLMs, implementing a part of the paper or providing a colab demo on applying the method in the paper to a different LLM. Students can (e-)meet with Rhea Sukthanker for feedback and any questions (e.g., to discuss a potential "add-on").

---

Grading

• Presentations: 50% (two times 25min + 15min Q&A)
• Report: 30% (4 pages in AutoML Conf format, due one week after last end term)
• Add-on: 20%

---

List of Potential Papers

• Are Sixteen Heads Really Better than One? https://arxiv.org/pdf/1905.10650.pdf
• FLIQS: One-Shot Mixed-Precision Floating-Point and Integer Quantization Search https://arxiv.org/abs/2308.03290
• Minitron https://www.arxiv.org/abs/2407.14679
• MiniLLM https://openreview.net/pdf?id=5h0qf7IBZZ
• Compressing LLMs: The Truth is Rarely Pure and Never Simple https://arxiv.org/abs/2310.01382
• Wanda : https://arxiv.org/pdf/2306.11695
• SparseGPT: https://arxiv.org/abs/2301.00774
• On the Effect of Dropping Layers of Pre-trained Transformer Models https://arxiv.org/pdf/2004.03844.pdf
• Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned https://arxiv.org/pdf/1905.09418.pdf
• A Fast Post-Training Pruning Framework for Transformers https://proceedings.neurips.cc/paper_files/paper/2022/file/987bed997ab668f91c822a09bce3ea12-Paper-Conference.pdf
• LLM-Pruner: On the Structural Pruning of Large Language Models https://arxiv.org/pdf/2305.11627.pdf
• Compresso https://arxiv.org/pdf/2310.05015.pdf
• LLM Surgeon https://arxiv.org/pdf/2312.17244.pdf
• Shortened Llama https://arxiv.org/abs/2402.02834
• SliceGPT https://arxiv.org/abs/2401.15024
• Structural pruning of large language models via neural architecture search https://arxiv.org/abs/2405.02267
• Not all Layers of LLMs are Necessary during Inference https://arxiv.org/pdf/2403.02181.pdf
• ShortGPT: Layers in Large Language Models are More Redundant Than You Expect https://arxiv.org/abs/2403.03853
• Shortened Llama https://arxiv.org/abs/2402.02834
• FLAP: Fluctuation-based adaptive structured pruning for large language models https://arxiv.org/abs/2312.11983
• Bonsai: Everybody Prune Now: Structured Pruning of LLMs with only Forward Passes https://arxiv.org/pdf/2402.05406.pdf
• The Unreasonable Ineffectiveness of the Deeper Layers https://arxiv.org/pdf/2403.17887v1.pdf
• Sheared Llama https://arxiv.org/abs/2310.06694
• Netprune https://arxiv.org/pdf/2402.09773.pdf
• MiniLLM: Knowledge Distillation of Large Language Models https://arxiv.org/pdf/2306.08543
• A Surve...

Course type:	Block Seminar
Time:	Kickoff Session: 17.10.24 14:00 - 16:00 Presentation Sessions: 05.02.24 10:00 - 18:00
Location:	Kickoff Session: SR 02-016/18 (G.-Köhler-Allee 101) Presentation Sessions: SR 02-016/18 (G.-Köhler-Allee 101)
Organizers:	André Biedenkapp , Noor Awad , Raghu Rajan , M Asif Hasan , Baohe Zhang
Web page:	HISinOne , Local Page

You can register for the seminar via HISinOne .

Note: To ensure that we can print the posters in time, they have to be submitted by noon of 31.01.25 to Baohe Zhang (https://nr.uni-freiburg.de/people/baohe-zhang)

Background

Hyperparameter optimization (HPO) is a powerful approach to achieve the best performance on many different problems. However, reinforcement learning (RL) offers unique challenges, such that classical HPO and automated machine learning approaches are often not directly applicable. In this seminar, we will discuss the unique challenges posed by the automated reinforcement learning (AutoRL) problem and discuss a variety of solution approaches, ranging from static to dynamic configuration methods. We further explore topics in reinforcement learning, such as curriculum learning, to understand how environment design can impact the learning performance of RL agents.

Requirements

We require that you have taken lectures on

• Reinforcement Learning, and/or
• Machine Learning

We strongly recommend that you have heard lectures on

• Automated Machine Learning
• Reinforcement Learning

Organization

The seminar is intended to be held as a block seminar, with sessions at the end of the semester in which all students will present their papers. All students are required to read the relevant literature to prepare for this session. After each presentation, we will have time for a question & discussion round and all participants are expected to take part in these. Each student has to write a short paper about their assigned topic which is to be handed in one week prior to their presentation.

During the semester, the students are expected to meet once with their assigned supervisor to discuss their paper and clarify open questions. Before this meeting however, students are required to meet with a "study buddy" whose role it is to be a discussion partner to clarify the first set of questions, and give feedback for the presentation.

Grading

• Presentation: 60% (20min + 10min Q&A)
• Paper: 20% (4 pages in AutoML Conf format , due one week prior to your presentation)
• Participation in Discussions: 20%

Literature

Relevant literature can be found at https://autorl.org/ and https://autorlworkshop.github.io/ . This list contains many recent papers and blog posts in the scope, though not all of the papers that we intend to cover in the seminar. For a general overview on the topic please refer to the survey on AutoRL (https://jair.org/index.php/jair/article/view/13596 ).

Kick-Off Slides

https://docs.google.com/presentation/d/1jt-RDAZfqnPqKS5YXyuRWuqgvX3wd-kHYnIZf8uylwU/edit?usp=sharing (best viewed in presentation mode due to overlapping animations)

Paper Pool

https://docs.google.com/document/d/1Dp7edapWlOFm7Irihgx8pLJSBNwfGa3nCkspeFVTAcA/edit?usp=sharing

The field of tabular das has recently been exploding with advances through large language models (LLMs), deep learning algorithms, and foundation models. In this seminar, we want to dive deep into these very recent advances to understand them.

Course type:	Seminar
Time	Five slots, to be determined with all participants. Kick-off is likely on the 23rd of October at 10 to 11 am.
Location	in-person; Meeting Room in our ML Lab
Organizers	Lennart Purucker
Registration	Via HISinOne (maximal six students, registration opens 14th of October)
Language	English

---

Prerequisites

We require that you have taken lectures on or are familiar with the following:

• Machine Learning
• Deep Learning
• Automated Machine Learning

---

Organization

After the kick-off meeting, everyone is assigned a paper about recent advances in deep learning (one or multiple papers, depending on the content). Then, everyone is expected to understand and digest their assigned papers and prepare two presentations. The first presentation is given in midterms (two separate slots), and the second during the endterms (two separate slots).

• The first presentation will focus on the relationship between the papers, any relevant related work, any background to understand the paper, and the greater context of the work.

• The second presentation will focus on the paper's contributions, describing them in detail.

In addition to the second presentation, students are expected to contribute an "add-on" related to the paper for the final report. This includes but is not limited to reproducing some experiments, implementing a part of the paper, providing a greater literature survey, fact-checking citations, experiments, or methodology, building a WebUI or demo for the paper, etc. Students can (e-)meet with Lennart Purucker for feedback and any questions (e.g., to discuss a potential "add-on").

---

Grading

• Presentations: 40% (two times 20min + 20min Q&A)
• Report: 40% (4 pages in AutoML Conf format, due one week after last end term)
• Add-on: 20% (due with the report)

---

Short(er) List of Potential Papers / Directions:

LLMs

• https://arxiv.org/abs/2409.03946
• https://arxiv.org/abs/2403.20208
• https://arxiv.org/abs/2404.00401 , https://aclanthology.org/2024.lrec-main.1179/ , https://arxiv.org/abs/2408.09174
• https://arxiv.org/abs/2404.05047
• https://arxiv.org/abs/2404.17136
• https://arxiv.org/abs/2404.18681 , https://arxiv.org/abs/2405.17712 , https://arxiv.org/abs/2406.08527
• https://arxiv.org/abs/2405.01585
• https://arxiv.org/abs/2407.02694
• https://arxiv.org/abs/2408.08841
• https://arxiv.org/abs/2408.11063
• https://arxiv.org/abs/2403.19318
• https://arxiv.org/abs/2403.06644
• https://arxiv.org/abs/2402.17453 , https://arxiv.org/abs/2409.07703
• https://arxiv.org/abs/2403.01841

Deep Learning

• https://arxiv.org/abs/2405.08403
• https://arxiv.org/abs/2307.14338
• https://arxiv.org/abs/2305.06090 , https://arxiv.org/abs/2406.00281
• https://arxiv.org/pdf/2404.17489
• https://arxiv.org/abs/2405.14018 , https://arxiv.org/abs/2406.05216 , https://arxiv.org/abs/2406.17673 , https://arxiv.org/abs/2409.05215 , https://arxiv.org/abs/2406.14841
• https://arxiv.org/abs/2408.06291
• https://arxiv.org/abs/2408.07661
• https://arxiv.org/abs/2409.08806
• https://arxiv.org/abs/2404.00776

Foundation Models / In-Context Learning

• https://arxiv.org/abs/2406.09837 , https://arxiv.org/abs/2307.09249
• https://arxiv.org/abs/2406.06891 ,https://arxiv.org/abs/2408.17162
• https://arxiv.org/abs/2405.16156 , https://arxiv.org/abs/2402.11137 , https://arxiv.org/pdf/2406.05207...

Course type:	Lecture + Exercise
Time:	Lecture: Tuesday, 10:15 - 11:45; Optional exercises: Friday, 10:15 - 11:45
Location:	The course will be in-person. - Weekly flipped classroom sessions will be held on Tuesday in HS 00 006 (G.-Köhler-Allee 082) - Optional exercise sessions will take place on Friday in HS 00 006 (G.-Köhler-Allee 082)
Organizers:	Steven Adriaensen , Abhinav Valada , Mahmoud Safari , Rhea Sukthanker , Johannes Hog
Web page:	ILIAS - available starting 8am, 15.10.24 (please make sure to also register for all elements of this course module in HISinOne)

Foundations of Deep Learning

Deep learning is one of the fastest growing and most exciting fields. This course will provide you with a clear understanding of the fundamentals of deep learning including the foundations to neural network architectures and learning techniques, and everything in between.

Course Overview

The course will be taught in English and will follow a flipped classroom approach.

Every week there will be:

• a video lecture
• an exercise sheet
• a flipped classroom session (Tuesdays, 10:15 - 11:45)
• an attendance optional exercise session (Fridays, 10:15 - 11:45)

At the end, there will be a written exam (likely an ILIAS test).

Exercises must be completed in groups and must be submitted 2 weeks (+ 1 day) after their release.
Your submissions will be graded and you will receive weekly feedback.
Your final grade will be solely based on a written examination, however, a passing grade for the exercises is a prerequisite for passing the course.

Course Material: All material will be made available in ILIAS and course participation will not require in-person presence. That being said, we offer ample opportunity for direct interaction with the professors during live Q & A sessions and with our tutors during weekly attendance optional in-class exercise sessions.

Exam: The exam will likely be a test you complete on ILIAS. In-person presence will be required .

Course Schedule

The following are the dates for the flipped classroom sessions (tentative, subject to change):

15.10.24- Kickoff: Info Course Organisation
22.10.24 - Week 1: Intro to Deep Learning
29.10.24 - Week 2: From Logistic Regression to MLPs
5.11.24 - Week 3: Backpropagation
12.11.24 - Week 4: Optimization
19.11.24 - Week 5: Regularization
26.11.24 - Week 6: Convolutional Neural Networks (CNNs)
03.12.24 - Week 7: Recurrent Neural Networks (RNNs)
10.12.24 - Week 8: Attention & Transformers
17.12.24 - Week 9: Practical Methodology
07.01.25 - Week 10: Auto - Encoders, Variational Auto - Encoders, GANs
14.01.25 - Week 11: Uncertainty in Deep Learning
21.01.25 - Week 12: AutoML for DL
28.01.25 - Round - up / Exam Q & A

In the first session (on 15.10.24) you will get additional information about the course and get the opportunity to ask general questions. While there is no need to prepare for this first session, we encourage you to already think about forming teams.
The last flipped classroom session will be held on 28.01.25.

Competition Results

Flower Classification Challenge

This semester, we organized an optional student competition. In this challenge, students were to train a model to perform class prediction on a flower dataset (more info here ).

There were two tracks:
- Fast-track (models with less than 100k parameters)
- Large-track (models with less than 25M parameters)
The winners per track were determined based on the accuracy of the submitted models on a hidden test set.

The Fast-track podium (12 entries):
- 1st place: Jan Sander
- 2nd place: Sepuh Hovhannisyan
- 3rd place: Ole Ossen

The Large-track podium (9 entries):
- 1st place: Devika Panneer Selvam
- 2nd/3rd place : Ali Mert Aydın, Oğuz Kuyucu, Salih Bora Öztürk
- 2nd/3rd place: Ole Ossen

Congratulations!

Questions?

If you have a question, please post it in the ILIAS forum (so everyone can benefit from the answer).

Welcome to the Deep Learning Lab a joint teaching effort of the Robotics (R) , Robot Learning (RL) , Neurorobotics (NR) , Computer Vision (CV) , and Machine Learning (ML) Labs. For more details check the following link: https://rl.uni-freiburg.de/teaching/ss24/dl-lab/

Course type:	Flipped-classroom Lecture + Exercise
Time:	Mondays, 14:00 - 16:00
Location:	HS 00 036 SCHICK - SAAL (G.-Köhler-Allee 101)
Organizers:	Frank Hutter , Heri Rakotoarison , Neeratyoy Mallik , Eddie Bergman , Johannes Hog , Martin Mráz, Steven Adriaensen , Noor Awad , André Biedenkapp
Web page:	HISinOne

First Lecture

Date : Monday, April 15, 2024, 14:15.
Requirement for attending : The Overview lecture from the course website (password to be sent via HISinOne email and first in-person lecture).

Background

Applying machine learning (ML) and in particular deep learning (DL) in practice is a challenging task and requires a lot of expertise. Among other things, the success of ML/DL applications depends on many design decisions, including an appropriate preprocessing of the data, choosing a well-performing machine learning algorithm, and tuning its hyperparameters, giving rise to a complex pipeline. Unfortunately, even experts need days, weeks, or even months to find well-performing pipelines and can still make mistakes when optimizing their pipelines. The course will discuss meta-algorithmic approaches to automatically search for, and obtain well-performing machine learning systems using automated machine learning (AutoML). Such AutoML systems allow for faster development of new ML/DL applications, require far less expert knowledge than doing everything from scratch and often even outperform human developers. In this lecture, you will learn how to use such AutoML systems, develop your own systems, and understand ideas behind state-of-the-art AutoML approaches.

Requirements

We strongly recommend that you know the foundations of

• machine learning (ML)
• and deep learning (DL)

We further recommend that you have hands-on experience with:

• Python (3.8+)
• machine learning (scikit-learn)
• deep learning (PyTorch)

The participants should have attended at least one other course for ML and DL in the past.

Topics

The lectures are partitioned into several parts, including:

• Hyperparameter Optimization
• Bayesian Optimization for Hyperparameter Optimization
• Neural Architecture Search
• Dynamic Configuration
• Analysis and Interpretability of AutoML
• Algorithm Selection/Meta-Learning

Organization

The course will be taught in a flipped-classroom style. We will meet weekly once for a lecture, and once weekly (optionally) for exercise sessions.

Every week, there will be a new exercise sheet. Most exercises will be practical, and involve programming in Python and teamwork (teams of up to 3 students!) so that you learn how to apply AutoML in practice.

Lecture: Mondays , 14 - 16

Exercise sessions [optional] (hybrid format) : Thursdays , 14-16

Exercise submission deadline [strict] (via Github) : Tuesdays, 23:59

Course material : The links to access the GitHub classroom repositories, videos with subtitles, and extra lecture materials will be made available on this page . (The password to that page will be announced in the first session and via HISinOne email).

The course will be taught in English.

MOOC content: The material is publicly available via the AI-Campus platform . Please register on AI-Campus to access the materials. Grading of the exercises will be done via GitHub classroom.

The lecture materials are open-sourced via https://github.com/automl-edu/AutoMLLecture .

Course type:	Live Lectures + (optional) Exercises
Time:	Tuesday 10:00 - 12:00, Friday 10:00 - 12:00
Location:	HS 00 036 SCHICK - SAAL (G.-Köhler-Allee 101)
Organizers:	Joschka Bödecker , Tim Welschehold , Steven Adriaensen
Web page:	ILIAS

Kickoff: The first lecture will take place on Tuesday 16.04. There is nothing you need to prepare. During this lecture, we will give you an overview of the course content, its organization, and the history of AI. The first exercise sheet will be released on Friday 19.04 and is due for submission before Friday 26.04, 8:00 am (optional, to receive feedback).

Course Content: This course will introduce basic concepts and techniques used within the field of Artificial Intelligence. Among other topics, we will discuss:

• Introduction to and history of Artificial Intelligence
• Agents
• Problem-solving and search
• Board Games
• Logic and knowledge representation
• Planning
• Representation of and reasoning with uncertainty
• Machine learning
• Deep Learning

Lectures will roughly follow the book: Artificial Intelligence a Modern Approach (3rd edition)

Organization : There are two lecture slots every week, on Tuesday and Friday (10:15 - 11:45). During these slots, there will be live (in-person) lectures. These will be in-person only (not hybrid), but recordings will be made available on ILIAS.

Every week, we will also release an exercise sheet (on Friday) to be submitted before next Friday (8:00am). Your submissions will not be graded, but you will receive feedback from our tutors. On Friday, after the live lecture, one of our tutors will present the master solution and answer any questions you may have. Participation in these weekly sessions and exercises is optional. The only requirement for passing the course is passing the final exam (mode: written, in-person, open book) which will take place on 06.09.2024 at 9:00am.

If you have any questions, please post in the ILIAS forum or contact us ailect24@informatik.uni-freiburg.de

Course type:	Lecture + Exercise
Time:	Lecture: Tuesday, 10:15 - 11:45; Optional exercises: Friday, 10:00 - 12:00
Location:	The course will be in-person: - Weekly flipped classroom sessions will be held on Tuesday in the Kinohörsaal (G.-Köhler-Allee 82) - Optional exercise sessions will take place on Friday in HS 00 006 (G.-Köhler-Allee 082)
Organizers:	Frank Hutter , Abhinav Valada , André Biedenkapp , Mahmoud Safari , Rhea Sukthanker
Web page:	ILIAS (please make sure to also register for all elements of this course module in HISinOne: Lecture + Exercise )

Foundations of Deep Learning

Deep learning is one of the fastest growing and most exciting fields. This course will provide you with a clear understanding of the fundamentals of deep learning including the foundations to neural network architectures and learning techniques, and everything in between.

Course Overview

The course will be taught in English and will follow a flipped classroom approach.

Every week there will be:

• a video lecture
• an exercise sheet
• a flipped classroom session (Mondays 14:15 - 15:45)
• an attendance optional exercise session (Fridays)

At the end, there will be a written exam (likely an ILIAS test).

Exercises must be completed in groups and must be submitted a week (+ 1 day) after their release.
Your submissions will be graded and you will receive weekly feedback.
Your final grade will be solely based on a written examination, however, a passing grade for the exercises is a prerequisite for passing the course.

Course Material: All material will be made available in ILIAS and course participation will not require in-person presence. That being said, we offer ample opportunity for direct interaction with the professors during live Q & A sessions and with our tutors during weekly attendance optional in-class exercise sessions.

Exam: The exam will likely be a test you complete on ILIAS. In-person presence may be required (TBA).

Course Schedule

The following are the dates for the flipped classroom sessions:

16.10.23 - Kickoff: Info Course Organisation / Team Formation
23.10.23 - ChatGPT Panel Discussion
30.10.23 - Week 1: Intro to Deep Learning
06.11.23 - Week 2: From Logistic Regression to MLPs
13.11.23 - Week 3: Backpropagation
20.11.23 - Week 4: Optimization
27.11.23 - Week 5: Regularization
04.12.23 - Week 6: Convolutional Neural Networks (CNNs)
11.12.23 - Week 7: Recurrent Neural Networks (RNNs)
18.12.23 - Week 8: Practical Methodology
08.01.24 - Week 9: Attention & Transformers
15.01.24 - Week 10: Auto - Encoders, Variational Auto - Encoders, GANs
22.01.24 - Week 11: Uncertainty in Deep Learning
29.01.24 - Week 12: AutoML for DL
05.02.24 - Round - up / Exam Q & A

In the first session (on 16.10.22) you will get additional information about the course and get the opportunity to ask general questions (and form groups!) While there is no need to prepare for this first session, we encourage you to already think about forming teams.
The last flipped classroom session will be held on 05.02.23.

Competition Results

Flower Classification Challenge

This semester, we organized an optional student competition. In this challenge, students were to train a model to perform class prediction on a flower dataset (more info here ).

There were two tracks:
- Fast-track (models with less than 100k parameters)
- Large-track (models with less than 25M parameters)
The winners per track were determined based on the accuracy of the submitted models on a hidden test set.

The Fast-track podium :
- 1st place: Simon Rauch, Philip Zimmer
- 2nd place: Aisan Fathi, Tilo Heep, Anshul Gupta
- 3rd place: Ali Elganzory, Emmanuel Hofmann, Luis Barragan

The Large-track podium :
- 1st place: Ali Elganzory, Emmanuel Hofmann, Luis Barragan
- 2nd place: Gerrit Freiwald, Lyubomir Ivanov, Yu Tung Wang
- 3rd place: Aisan Fathi, Tilo Heep, Anshul Gupta

Congratulations!

Questions?

If you have a question, please post it in the ILIAS forum (so everyone can benefit from the answer).

Course Type:	Lab Course
Time & Location:	24.11.23; MST Pool, Building 74, 14:00-15:00 Introduction to Lab 27.11.23 - 30.11.23; MST Pool, Building 74, 09:00-17:30 participation in the AutoML Fall School During the semester : A supervised projected related to AutoML; Themes relating to ensembling and capabilities of AutoML systems 07.02.24; Room 13, Building 74, 14:00-16:00 : Poster presentation
Organizers:	Eddie Bergman , Lennart Purucker , Frank Hutter
Web Page:	HISinOne , Local Page

Note: You do not need to register for the fall school if you have registered for this course. Your participation in the fall school will be free of charge. You can register for the course via HISinOne .

Background

Applying machine learning (ML) and in particular deep learning (DL) in practice is a challenging task and requires a lot of expertise. Among other things, the success of ML/DL applications depends on many design decisions, including an appropriate preprocessing of the data, choosing a well-performing machine learning algorithm and tuning its hyperparameters, giving rise to a complex pipeline. Unfortunately, even experts need days, weeks or even months to find well-performing pipelines and can still make mistakes when optimizing their pipelines.

The lab course will start with a fall school featuring lectures on hot topics in AutoML such as "automating data science", "automated reinforcement learning" or "neural architecture search", as well as present tutorials on various topics in AutoML. After this fall school, students will take on a project themed around Ensembling or capabilities of AutoML Systems.

Requirements

We require that you have heard a lecture on

• Machine Learning, and/or
• Deep Learning

We strongly recommend that you have taken the AutoML lecture.

Organization

Students take part in the AutoML Fall School (remotely, free of charge) to hear from world leading AutoML experts about current hot topics in the field. After this week, we will meet to present potential projects from which the students are free to select which one they want to tackle. We expect that students work in groups of up to three. During the semester the students will meet with a supervisor to discuss potential issues they are facing. At the end of the semester all groups will present their work during a poster presentation.

Grading

The grades are determined based on the quality of the project part.

Important Dates

• 24.11.23; MST Pool, Building 74, 14:00-15:00 Introduction to Lab
• 27.11.23 - 30.11.23; MST Pool, Building 74, 09:00-17:30 participation in the AutoML Fall School
• 07.02.24; Room 13, Building 74, 14:00-16:00 : Poster presentation

Tabular data has long been overlooked by deep learning research, despite being the most common data type in real-world machine learning applications. While deep learning methods excel on many ML applications, tabular data classification problems are still dominated by Gradient-Boosted Decision Trees. More recently, deep learning-based approaches have been proposed which showed remarkable efficiency and performance improvements. In this seminar, we will discuss this recent literature, exploring the most promising techniques and approaches for handling tabular data in deep learning.

Course type:	Seminar
Time	Every Tuesday from 14:15 - 16:00
Location	in-person; Room SR 00-006, Building 051
Organizers	Herilalaina Rakotoarison , Arbër Zela , Fabio Ferreira , Frank Hutter
Registration	Via HISinOne
Web page	Link to the seminar webpage

Welcome to the Deep Learning Lab a joint teaching effort of the Robotics (R) , Robot Learning (RL) , Neurorobotics (NR) , Computer Vision (CV) , and Machine Learning (ML) Labs. For more details check the following link: https://rl.uni-freiburg.de/teaching/ss23/laboratory-deep-learning-lab

Course type:	Live Lectures + (optional) Exercises
Time:	Tuesday 10:00 - 12:00, Friday 10:00 - 12:00
Location:	HS 00 036 SCHICK - SAAL (G.-Köhler-Allee 101)
Organizers:	Joschka Bödecker , Tim Welschehold , Steven Adriaensen
Web page:	ILIAS

Course Content: This course will introduce basic concepts and techniques used within the field of Artificial Intelligence. Among other topics, we will discuss:

• Introduction to and history of Artificial Intelligence
• Agents
• Problem-solving and search
• Board Games
• Logic and knowledge representation
• Planning
• Representation of and reasoning with uncertainty
• Machine learning
• Deep Learning

Lectures will roughly follow the book: Artificial Intelligence a Modern Approach (3rd edition)

Organization : There are two lecture slots every week, on Tuesday and Friday (10:15 - 11:45). The first lecture will be on Tuesday 18.04 (see ILIAS for all appointments). During these slots, there will be live (in-person) lectures. These will be in-person only (not hybrid), but recordings will be made available on ILIAS.

Every week, we will also release an exercise sheet (on Friday) to be submitted before next Friday (8:00am). Your submissions will not be graded, but you will receive feedback from our tutors. On Friday, after the live lecture, one of our tutors will present the master solution and answer any questions you may have. The first exercise is due on 28.04. Participation in these weekly sessions and exercises is optional. The only requirement for passing the course is passing the final exam (date/format TBD).

If you have any questions, please post in the ILIAS forum or contact us ailect23@informatik.uni-freiburg.de

Course type:	Flipped-classroom Lecture + Exercise
Time:	Monday 14:00 - 16:00
Location:	SR 01-009/13 (G.-Köhler-Allee 101)
Organizers:	Frank Hutter , André Biedenkapp , Heri Rakotoarison ,Neeratyoy Mallik , Eddie Bergman
Web page:	HISinOne

First Lecture

Date : April 17, 2023.
Requirement for attending : The Overview lecture from the AutoML MOOC .

Background

Applying machine learning (ML) and in particular deep learning (DL) in practice is a challenging task and requires a lot of expertise. Among other things, the success of ML/DL applications depends on many design decisions, including an appropriate preprocessing of the data, choosing a well-performing machine learning algorithm and tuning its hyperparameters, giving rise to a complex pipeline. Unfortunately, even experts need days, weeks or even months to find well-performing pipelines and can still make mistakes when optimizing their pipelines.The course will discuss meta-algorithmic approaches to automatically search for, and obtain well-performing machine learning systems by means of automated machine learning (AutoML). Such AutoML systems allow for faster development of new ML/DL applications, require far less expert knowledge than doing everything from scratch and often even outperform human developers. In this lecture, you will learn how to use such AutoML systems, to develop your own systems and to understand ideas behind state-of-the-art AutoML approaches.

Requirements

We strongly recommend that you know the foundations of

• machine learning (ML)
• and deep learning (DL)

We further recommend that you have hands-on experience with:

• Python (3.8+)
• machine learning (scikit-learn)
• deep learning (PyTorch)

The participants should have attended at least one other course for ML and DL in the past.

Topics

The lectures are partitioned in several parts, including:

• Hyperparameter Optimization
• Bayesian Optimization for Hyperparameter Optimization
• Neural Architecture Search
• Dynamic Configuration
• Analysis and Interpretability of AutoML
• Algorithm Selection/Meta-Learning

The weekly slides and exercises are available on https://ml.informatik.uni-freiburg.de/github-classroom-automl-2023

Organization

The course will be taught in a flipped-classroom style. We will meet weekly once for a combined Q/A session and an exercise. Roughly every week, there will be a new exercise sheet. Most exercises will be practical, involve programming in python and teamwork (teams of up to 3 students!) so that you learn how to apply AutoML in practice.

Lecture/Exercise: Monday 14 - 16

The material is publicly available via the AI-Campus platform . Please register on AI-Campus to access the materials. Grading of the exercises will be done via GitHub classroom.

The links to access the GitHub classroom repositories will be made available on this page . (The password to that page will be announced in the first session).

The course will be taught in English.

The lecture materials are open sourced via https://github.com/automl-edu/AutoMLLecture

Course type:	Lecture + Exercise
Time:	Monday, 14:15 - 15:45, first meeting: Oct. 17
Location:	The course will be Hybrid: - Weekly flipped classroom sessions will be held on Monday at HS 00 026 µ - SAAL (G.-Köhler-Allee 101) and via Zoom. See ILIAS for Zoom link. - Optional exercise sessions will take place on Friday 10:15-11:45 at HS 00 006 (G.-Köhler-Allee 082)
Organizers:	Frank Hutter , Abhinav Valada , Steven Adriaensen , Mahmoud Safari
Web page:	ILIAS (please make sure to also register for all elements of this course module in HISinOne)

Foundations of Deep Learning

Deep learning is one of the fastest growing and most exciting fields. This course will provide you with a clear understanding of the fundamentals of deep learning including the foundations to neural network architectures and learning techniques, and everything in between.

Course Overview

The course will be taught in English and will follow a flipped classroom approach.

Every week there will be:

• a video lecture
• an exercise sheet
• a flipped classroom session (hybrid, Mondays 14:15 - 15:45)
• an attendance optional exercise session (in-class/offline, Fridays 10:15 - 11:45)

At the end, there will be a written exam (likely an ILIAS test).

Exercises must be completed in groups and must be submitted a week (+ 1 day) after their release.
Your submissions will be graded and you will receive weekly feedback.
Your final grade will be solely based on a written examination, however, a passing grade for the exercises is a prerequisite for passing the course.

Hybrid course: All material will be made available online and course participation will not require in-person presence. That being said, we offer ample opportunity for direct interaction with the professors during live Q & A sessions (HS 00 026 µ - SAAL, G.-Köhler-Allee 101) and with our tutors during weekly attendance optional in-class exercise sessions (HS 00-006, G.-Köhler-Allee 082).

Exam: The exam will likely be a test you complete on ILIAS. In-person presence may be required (TBA).

Course Schedule

The following are the dates for the flipped classroom sessions:

17.10.22 - Kickoff: Info Course Organisation / Team Formation
24.10.22 - Week 1: Intro to Deep Learning
31.10.22 - Week 2: From Logistic Regression to MLPs
07.11.22 - Week 3: Backpropagation
14.11.22 - Week 4: Optimization
21.11.22 - Week 5: Regularization
28.11.22 - Week 6: Convolutional Neural Networks (CNNs)
05.12.22 - Week 7: Recurrent Neural Networks (RNNs)
12.12.22 - Week 8: Attention & Transformers
19.12.22 - Week 9: Practical Methodology
09.01.23 - Week 10: Hyperparameter Optimization
16.01.23 - Week 11: Neural Architecture Search
23.01.23 - Week 12: Auto-Encoders, Variational Auto-Encoders, GANs
30.01.23 - Week 13: Uncertainty in Deep Learning
06.02.23 - Round-up / Exam Q & A

The course material (lecture video, slides, exercise sheet) for "Week N" will be made available a week before the flipped classroom session for "Week N". For example, the material for Week 1 will be available on 17.10.22 and solutions to the exercises must be submitted latest 25.10.22 at 23:59. Virtual participation in flipped classroom sessions will be enabled using Zoom and the meeting link can be found on ILIAS in the "Flipped Classroom" folder.

In the first session (on 17.10.22) you will get additional information about the course and get the opportunity to ask general questions (and form groups!) While there is no need to prepare for this first session, we encourage you to already think about forming teams.
The last flipped classroom session will be held on 06.02.23.

Competition Results

Flower Classification Challenge

This semester, we organized an optional student competition. In this challenge, students were to train a model to perform class prediction on a flower dataset (more info here ).

There were two tracks:
- Fast-track (models with less than 100k parameters)
- Large-track (models with less than 25M parameters)
The winners per track were determined based on the accuracy of the submitted models on a hidden test set.

The Fast-track podium:
- 1st place: Jelle Dehn, Soham Basu, Laura Neschen
- 2nd place: Adithya Anoop Thoniparambil, Daniel Rogalla
- 3rd place: Dominika Matus, M'Saydez Campbell, Florian Vogt

The Large-track podium:
- 1st place: Muhammad Ali
- 2nd place: Jelle Dehn, Soham Basu, Laura Neschen
- 3rd place: Premnath Srinivasan, Rishabh Verma, Ali Sarlak

Congratulations!

Tree Segmentation Challenge

After the success of the flower classification challenge, we organized a second optional student competition, spanning two semesters. In this challenge, students were to train a model to perform semantic segmentation on a tree dataset (more info here ).

The podium:
- 1st place: Rishabh Verma, Premnath Srinivasan
- 2nd place: Ali Sarlak, Elham Elyasi
- 3rd place: Noah Lenagan

Congratulations!

Questions?

If you have a question, please post it in the ILIAS forum (so everyone can benefit from the answer).

Alternatively, you can also email dl-orga-ws22@cs.uni-freiburg.de

Course Type:	Lab Course
Time & Location:	10.10.22 - 13.10.22 participation in the AutoML Fall School 19.10.22; 14:00-16:00; 082 HS 00 006 Introduction to Projects During the semester : Implement your own AutoML system TBA : Poster presentation
Organizers:	André Biedenkapp , Rhea Sukthanker , Frank Hutter
Web Page:	HISinOne , Local Page

Note: You do not need to register for the fall school if you have registered for this course. Your participation in the fall school will be free of charge. You can register for the course via HISinOne .

Background

Applying machine learning (ML) and in particular deep learning (DL) in practice is a challenging task and requires a lot of expertise. Among other things, the success of ML/DL applications depends on many design decisions, including an appropriate preprocessing of the data, choosing a well-performing machine learning algorithm and tuning its hyperparameters, giving rise to a complex pipeline. Unfortunately, even experts need days, weeks or even months to find well-performing pipelines and can still make mistakes when optimizing their pipelines. The lab course will start with a fall school featuring lectures on hot topics in AutoML such as "automating data science", "automated reinforcement learning" or "neural architecture search", as well as present tutorials on various topics in AutoML. After this fall school, students are tasked with implementing their own AutoML system for a particular task or problem domain.

Requirements

We require that you have heard a lecture on

• Machine Learning, and/or
• Deep Learning

We strongly recommend that you have taken the AutoML lecture.

Organization

This lab course officially starts one week before the start of the lecture period. During that week, students take part in the AutoML fall school , to be held in Freiburg, to hear from world leading AutoML experts about current hot topics in the field. After this week, we will meet to present potential projects from which the students are free to select which one they want to tackle. We expect that students work in groups of up to three to implement their AutoML system. During the semester the students will meet with a supervisor to discuss potential issues they are facing when implementing their system. At the end of the semester all groups will present their work during a poster presentation.

Grading

The grades are determined based on the quality of the project part.

Important Dates

• 10.10.22 - 13.10.22: AutoML Fall School
• 19.10.22 14:00 - 16:00: Introduction to the topics In: HS 00 006 (G.-Köhler-Allee 082)
• TBD: Poster Session

Kickoff Slides

https://docs.google.com/presentation/d/1xZ2L8OQpycTzP8E4V-jhjowilh1xZGa4l3YUAzcd7jc/edit?usp=drivesdk

Course type:	Seminar
Time:	Tuesdays 14:00 - 16:00
Location:	G.-Köhler-Allee 051, R 03 026
Organizers:	André Biedenkapp , Noor Awad , Frank Hutter
Web page:	HISinOne , Local Page

You can register for the seminar via HISinOne .

Background

Hyperparameter optimization is a powerful approach to achieve the best performance on many different problems. However, automated approaches to solve this problem tend to ignore the iterative nature of many algorithms. With the dynamic algorithm configuration (DAC) framework we can generalize over prior optimization approaches, as well as handle optimization of hyperparameters that need to be adjusted over multiple time-steps. In this seminar, we will discuss applications (such as temporally extended epsilon greedy exploration in RL ) and domains (e.g., reinforcement learning , evolutionary algorithms or deep learning ) that can benefit from dynamic configuration methods. A large portion of the seminar will be dedicated to discussing papers that describe DAC methods that employ reinforcement learning to learn hyperparameter optimization policies for various domains.

Requirements

We require that you have taken lectures on

• Machine Learning, and/or
• Deep Learning

We strongly recommend that you have heard lectures on

• Automated Machine Learning
• Reinforcement Learning

Organization

Every week all students read the relevant literature. Two students will prepare presentations for the topics of the week and present it in the session. After each presentation, we will have time for a question & discussion round and all participants are expected to take part in these. Each student has to write a short paper about their assigned topic which is to be handed in one week after their presentation.

Grading

• Presentation: 40% (20min + 20min Q&A)
• Paper: 40% (4 pages in AutoML Conf format , due one week after your presentation)
• Participation in Discussions: 20%

Schedule

Welcome to the Deep Learning Lab a joint teaching effort of the Robotics (R) , Robot Learning (RL) , Neurorobotics (NR) , Computer Vision (CV) , and Machine Learning (ML) Labs. For more details check the following link: https://rl.uni-freiburg.de/teaching/ss22/laboratory-deep-learning-lab

Fair and Interpretable Machine Learning

Course type:	Seminar
Time:	Wednesday 16:00 - 18:00
Location:	G.-Köhler-Allee 051, SR 00 034
Organizers:	Janek Thomas , Noor Awad
Web page:	HisInOne

Seminar on Fair and Interpretable Machine Learning

The seminar language will be English (even if everyone is German-spoken, to practice presentation skills in English). 

*First meeting:

• 27th April, 16:00-18:00, SR 00 034 (G.-Köhler-Allee 051).

*Regular meetings:

• Every Wednesday, 16:00-18:00, SR 00 034 (G.-Köhler-Allee 051).

Background

The seminar focuses on interpretable machine learning in the first half of the semester and fair machine learning in the second half.

We will discuss model-agnostic tools for fairness and interpretability as well as specific algorithms. Further topics include measuring fairness, a causal perspective on fairness and how to consider fairness in AutoML.

Organization

Each week: All Students read relevant literature. Three students prepare the topic with slides and applications. Three other students are assigned as discussants. Discussants have to meet with the group presenting prior to the session, give feedback and prepare critical discussion points and open questions. 

End of the semester: Each student has to write a short paper (10 pages) about their topic. 

Requirements

We strongly recommend that you know the foundations of

• Machine Learning
• For some topics: Deep Learning

Main Literature

[1] Molnar, Christoph. Interpretable machine learning. Lulu. com, 2020. - https://christophm.github.io/interpretable-ml-book/

[2] Mehrabi, Ninareh, et al. "A survey on bias and fairness in machine learning." ACM Computing Surveys (CSUR) 54.6 (2021): 1-35.

[3] Barocas, Solon, Moritz Hardt, and Arvind Narayanan. "Fairness in machine learning." Nips tutorial 1 (2017): 2.

Schedule

Date	Topic	Main Ref.	Further Refs.
27.04.2022	Introduction, groups and topic assignments	-
04.05.2022	-
11.05.2022	-	[1] Chapter 1-3
18.05.2022	Interpretable Machine Learning Methods	[1] Chapter 5	Friedman, Jerome H., and Bogdan E. Popescu. "Predictive learning via rule ensembles." The annals of applied statistics 2.3 (2008): 916-954.
			Lou, Yin, et al. "Accurate intelligible models with pairwise interactions." Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining. 2013.
			Hofner, Benjamin, et al. "A framework for unbiased model selection based on boosting." Journal of Computational and Graphical Statistics 20.4 (2011): 956-971.
25.05.2022	Global Model-Agnostic Interpretability Methods	[1] Chapter 8	Apley, Daniel W., and Jingyu Zhu. “Visualizing the effects of predictor variables in black box supervised learning models.” Journal of the Royal Statistical Society: Series B (Statistical Methodology) 82.4 (2020): 1059-1086.
			Wei, Pengfei, Zhenzhou Lu, and Jingwen Song. "Variable importance analysis: a comprehensive review." Reliability Engineering & System Safety 142 (2015): 399-432.
			Kim, Been, Rajiv Khanna, and Oluwasanmi O. Koyejo. "Examples are not enough, learn to criticize! criticism for interpretability." Advances in neural information processing systems 29 (2016).
01.06.2022	Local Model-Agnostic Interpretability Methods	[1] Chapter 9	Goldstein, Alex, et al. "Peeking inside the black box: Visualizing statistical learning with plots of individual conditional expectation." journal of Computational and Graphical Statistics 24.1 (2015): 44-65.
			Ribeiro, Marco Tulio, Sameer Singh, and Carlos Guestrin. "" Why should I trust you?" Explaining the predictions of any classifier." Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. 2016.
			Karimi, Amir-Hossein, et al. "Model-agnostic counterfactual explanations for consequential decisions." International Conference on Artificial Intelligence and Statistics. PMLR, 2020.
08.06.2022	-	[3] Chapter 1
15.06.2022	Interpretability Methods for Neural Networks	[1] Chapter 10	Zeiler, Matthew D., and Rob Fergus. "Visualizing and understanding convolutional networks." European conference on computer vision. Springer, Cham, 2014.
			Kim, Been, et al. "Interpretability beyond feature attribution: Quantitative testing with concept activation vectors (tcav)." International conference on machine learning. PMLR, 2018.
			Jain, Sarthak, and Byron C. Wallace. "Attention is not explanation." arXiv preprint arXiv:1902.10186 (2019).
22.06.2022	Multi-Objective and constrained Optimization and Model Selection	TBD	Deb, Kalyanmoy, et al. "A fast and elitist multiobjective genetic algorithm: NSGA-II." IEEE transactions on evolutionary computation 6.2 (2002): 182-197.
			Knowles, Joshua. "ParEGO: A hybrid algorithm with on-line landscape approximation for expensive multiobjective optimization problems." IEEE Transactions on Evolutionary Computation 10.1 (2006): 50-66.
			Gardner, Jacob R., et al. "Bayesian Optimization with Inequality Constraints." ICML. Vol. 2014. 2014.
29.06.2022	Measures for Fairness	[2] Section 4.1	Hardt, Moritz, Eric Price, and Nati Srebro. "Equality of opportunity in supervised learning." Advances in neural information processing systems 29 (2016).
			Kearns, Michael, et al. "An empirical study of rich subgroup fairness for machine learning." Proceedings of the conference on fairness, accountability, and transparency. 2019.
			Chouldechova, Alexandra. "Fair prediction with disparate impact: A study of bias in recidivism prediction instruments." Big data 5.2 (2017): 153-163.
06.07.2022	Debiasing Methods	[2] Section 5.1	Mehrabi, Ninareh, et al. "Debiasing community detection: The importance of lowly connected nodes." 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM). IEEE, 2019.
			Calmon, Flavio, et al. "Optimized pre-processing for discrimination prevention." Advances in neural information processing systems 30 (2017).
			Kamiran, Faisal, and Toon Calders. "Data preprocessing techniques for classification without discrimination." Knowle...

Course type:	Flipped-classroom Lecture + Exercise
Time:	Wednesday 12:00 - 14:00
Location:	HS 00 036 SCHICK - SAAL (G.-Köhler-Allee 101)
Organizer:	Janek Thomas , André Biedenkapp , Edward Bergman , Neeratyoy Mallik , Frank Hutter
Web page:	Link to Lecture Page

Course type	Oberseminar: Reading Group (no ECTS) -- Automated Machine Learning
Time:	Wednesdays, 13:00
Web page:	Reading Group

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	online
Contact:	Samuel Müller ,

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	TBA
Location:	online only
Web page:	Open Projects

Update Oct. 14: We added details about the seminar mode.
Update Oct. 8: Title and time of the seminar changed & added the link to HISinOne.

Course type:	Seminar
Time:	Tuesday, 10:00 - 11:30 am, first meeting: Oct. 19
Location:	The course will be fully virtual/online. See ILIAS for Zoom link.
Topic & Mode:	In this year's seminar, we will discuss papers about Self-Supervised Learning with strong focus on computer vision. The details about the seminar mode will be discussed in the kick-off meeting on Oct 19th. We will begin our seminar with a background video that we discuss together on Oct 2nd. For the following sessions students are asked to: 1) present a paper, 2) lead the moderation on the paper, 3) actively participate in the discussion of other paper presentations. We will try to assign the papers based on your preference but there's no guarantee you will receive the paper you requested.
Organizer:	Frank Hutter , Fabio Ferreira , Samuel Müller , Robin Schirrmeister
Web page:	ILIAS , HISInOne

Course type:	Lecture + Exercise
Time:	Wednesday, 12:15 - 13:45, first meeting: Oct. 20
Location:	The course will be fully virtual/online. Weekly flipped classroom sessions will be held on Zoom. See ILIAS for Zoom link.
Organizers:	Frank Hutter , Abhinav Valada , Steven Adriaensen , Samuel Müller , Yash Mehta , Niclas Vödisch
Web page:	ILIAS (please also register for all elements of this course module in HISinOne )

Foundations of Deep Learning

Deep learning is one of the fastest growing and most exciting fields. This course will provide you with a clear understanding of the fundamentals of deep learning including the foundations to neural network architectures and learning techniques, and everything in between.

Course Overview

The course will be taught in English and will follow a flipped classroom approach. Every week there will be:
- a video lecture
- an exercise sheet
- a flipped classroom session (virtual/online, Wednesdays 12:15 - 13:45)
- an attendance optional exercise session (in-class/offline, Thursdays 10:15 - 11:45)
At the end, there will be a written exam (likely ILIAS test).
Exercises must be completed in groups and must be submitted a week (+ 1 day) after their release. Your submissions will be graded and you will receive weekly feedback. Your final grade will be solely based on a written examination, however a passing grade for the exercises is a prerequisite for passing the course.
Online course: All material will be made available online and course participation will not require in-person presence. That being said, we offer the opportunity for direct interaction with our tutors during weekly attendance optional in-class exercise sessions (building 82, HS 00-006). In addition, it is possible to attend the digital flipped classroom sessions on campus using your own laptop + headphones (Building 101 - HS 00 036).
Exam: The exam will likely be a test you complete on ILIAS. In person presence may be required (tba).

Course Schedule

The following are the dates for the flipped classroom sessions:
20.10.21 - Kickoff: Info Course Organisation / Team Formation
27.10.21 - Week 1: Overview of Deep Learning
03.11.21 - Week 2: From Logistic Regression to MLPs
10.11.21 - Week 3: Backpropagation
17.11.21 - Week 4: Optimization
24.11.21 - Week 5: Regularization
01.12.21 - Week 6: Convolutional Neural Networks (CNNs)
08.12.21 - Week 7: Recurrent Neural Networks (RNNs)
15.12.21 - Week 8: Practical Methodology & Architectures
22.12.21 - Week 9: Hyperparameter Optimization
12.01.22 - Week 10: Neural Architecture Search
19.01.22 - Week 11: Attention & Transformers
26.01.22 - Week 12: Auto-Encoders, Variational Auto-Encoders, GANs
02.02.22 - Week 13: Uncertainty in Deep Learning
09.02.22 - Round-up / Exam Q & A

The course material (lecture video, slides, exercise sheet) for "Week N" will be made available a week before the flipped classroom session for "Week N". For example, the material for Week 1 will be available on 20.10.21 and solutions to the exercises must be submitted latest 28.10.21 at 23:59. We will be using Zoom and the meeting link can be found on ILIAS in the "Flipped Classroom" folder.

In the first session (on 20.10.21) you will get additional information about the course and get the opportunity to ask general questions (and form groups!) While there is no need to prepare for this first session, we encourage you to already think about forming teams. The last flipped classroom session is held on 09.02.22.

Competition Results

This semester, we organised an optional student competition. In this challenge, students were to train a model to perform class prediction on a flower dataset (more info here ).

There were two tracks:
- Fast-track (models with less than 100k parameters)
- Large-track (models with less than 25M parameters)
The winners per track were determined based on the accuracy of the submitted models on a hidden test set.

The Fast-track podium:
- 1st place: Nisarga Nilavadi Chandregowda, Pablo Marhoff, Tidiane Ndir (accuracy: 90.49%)
- 2nd place: Bijay Gurung, Caoting Li, Kartik Yadav (accuracy: 82.80%)
- 3rd place: Uygar Akkoc, Aron Bahram, Samir Garibov (accuracy: 73.12%)

The Large-track podium:
- 1st place: Paweł Bugyi, Abhijeet Nayak, Preethi Sivasankaran (accuracy: 95.15%)
- 2nd place: Akshay Chandra Lagandula, Sai Prasanna Raman, John Robertson (accuracy: 92.49%)
- 3rd place: Bijay Gurung, Caoting Li, Kartik Yadav (accuracy: 91.28%)

Congratulations!

Questions?

If you have a question, please post it in the ILIAS forum (so everyone can benefit from the answer).
Alternatively, you can also email dl-orga-ws21@cs.uni-freiburg.de

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	TBA
Location:	online only
Web page:	Open Projects

Course type:	Lab Course
Time:	Tuesday, 14:00-16:00 (Beginning Apr 20, 2021)
Location:	online only
Organizer:	Frank Hutter , Fabio Ferreira , Jörg Franke , Arbër Zela , Samuel Müller , Yash Mehta
Web page:	Website , HISinOne

Course type:	Flipped-classroom Lecture + Exercise
Time:	Wednesday 14:00 - 16:00
Location:	online only
Organizer:	Frank Hutter André Biedenkapp , Gresa Shala
Web page:	AutoML

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	online
Contact:	Samuel Müller ,

Course type	Oberseminar: Reading Group (no ECTS) -- Automated Machine Learning
Time:	Wednesdays, 13:00
Web page:	Reading Group

Course type:	B.Sc./M.Sc. projects
Location:	Kitchen, Building 074
Web page:	Open Projects

Course type:	Seminar
Time:	Wednesday, 12:30 - 2pm, first meeting: Nov. 4
Location:	The seminar will be fully virtual/online and held on Zoom. See ILIAS for Zoom link.
Organizer:	Frank Hutter , Noor Awad , Fabio Ferreira
Web page:	ILIAS , HISinOne

Organizational

• we will meet every Wednesday, 12:30-2pm, starting Nov 4
• the seminar will be fully virtual and carried out with Zoom
• we will announce the Zoom meeting link on ILIAS
• link to ILIAS course
• link to HisInOne course
• list of papers released at end of the kick-off meeting

General

Welcome to the Automated Machine Learning Seminar webpage. The seminar language will be in English and it will be comprises of papers from three areas: learning to learn (L2L), hyperparameter optimization (HPO), and neural architecture search (NAS). We will start on Wednesday, Nov. 4 at 12:30-2pm on Zoom. Check out the ILIAS course (link above) for the Zoom link. The papers discussed in this seminar will be announced at the end of the kick-off meeting. We will assume all participants have understood the fundamental concepts presented in the AutoML course .

Procedure

We will meet weekly to discuss research papers from a yet to be released list of papers. Every week, two students present each one paper and will lead the succeeding discussion. All other participants read the paper and submit 3 questions due the Monday before via ILIAS. The presenter will be given access to the questions before the presentation. The discussion after the presentation will be based but not limited to the submitted questions. During the discussion, all participants are asked to be involved as much as possible. We will discuss the paper, its merits, limitations, etc.. In the end, the participants are asked to provide constructive feedback to the presenter. The feedback will not be considered for grading. The final grade takes the oral presentation, the questions submitted and class participation into account. Besides the seminar topic, you will learn several skills necessary not only in academia:

• read and understand research papers
• assessing the strengths and weaknesses
• oral presentation in front of your peers
• discussion with your peers
• high level summary of research with which you are not intimately familiar.

Schedule

The schedule is yet to be defined.

Material

All relevant material will be uploaded to the ILIAS course.

Contact

For questions, please email the organizers.

Course type:	Lecture + Exercise
Time:	Wednesday, 14:15 - 15:45, first meeting: Nov. 4 Exception: Week 8, Thursday, 14:15 - 15:45, Jan. 7
Location:	The course will be fully virtual/online. Weekly flipped classroom sessions will be held on Zoom. See ILIAS for Zoom link.
Organizers:	Frank Hutter , Abhinav Valada , Joschka Bödecker , Steven Adriaensen , Tim Frederic Runge
Web page:	ILIAS , HISinOne (lecture) , HISinOne (exercises, please register for both!)

Foundations of Deep Learning

Deep learning is one of the fastest growing and most exciting fields. This course will provide you with a clear understanding of the fundamentals of deep learning including the foundations to neural network architectures and learning techniques, and everything in between.

Course Overview

The course will be taught in english and will follow a flipped classroom approach. Every week there will be:
- a video lecture
- an exercise sheet
- a flipped classroom session (virtual/online, Wednesdays 14:15-15:45) At the end, there will be a written exam (online ILIAS test). Exercises must be completed in groups and must be submitted a week (+ 1 day) after their release. Your submissions will be graded and you will receive weekly feedback. Your final grade will be solely based on a written examination, however a passing grade for the exercises is a prerequisite for passing the course. Online course: All material will be made available online and course participation will not require in-person presence.
Online exam: The exam is a test you complete on ILIAS and does not require in-person presence.

Course Schedule

The following are the dates for the release of video lectures: 04.11.20 - Week 1: Overview of Deep Learning
11.11.20 - Week 2: From Logistic Regression to MLPs
18.11.20 - Week 3: Backpropagation
25.11.20 - Week 4: Optimization
02.12.20 - Week 5: Regularization
09.12.20 - Week 6: Convolutional Neural Networks (CNNs)
16.12.20 - Week 7: Recurrent Neural Networks (RNNs)
06.01.21 - Week 8: Practical Methodology & Architectures (flipped classroom on Thursday 07.01!)
13.01.21 - Week 9: Hyperparameter Optimization
20.01.21 - Week 10: Neural Architecture Search
27.01.21 - Week 11: Auto-Encoders, Variational Auto-Encoders, GANs
03.02.21 - Week 12: Uncertainty in Deep Learning On the same day, there is a flipped classroom session about the material released the week before. An exception is Week 8, where the flipped classroom session on RNNs will take place on Thursday 07.01 (14:15-15:45) instead of Wednesday 06.01 (holiday). We will be using Zoom and the meeting link can be found on ILIAS in the Flipped Classroom folder. In the first session (on 04.11.20) you will instead get additional information about the course and get the opportunity to ask general questions (and form groups!) While there is no need to prepare for this first session, we encourage you to already think about forming teams. The first exercise on 'Week 1: Overview of Deep Learning' is due on 12.11.20 at 23:59. The last flipped classroom session is held on 10.02.21.

Competition Results

This semester, we organised an optional student competition. In this challenge, students were to train a model to perform class prediction on a flower dataset (more info here ).
There were two tracks:
- Fast-track (models with less than 100k parameters)
- Large-track (models with less than 25M parameters)
The winners per track were determined based on the accuracy of the submitted models on a hidden test set.
The Fast-track podium:
- 1st place: Giulio Neusch-Frediani (accuracy: 76.25%)
- 2nd place: Guri Zabergja, Jeta Bekteshi, Albanot Makolli (accuracy: 72.75%)
- 3rd place: Florian Diederichs, Lukas Koenig, Nina Pant (accuracy: 71.75%)
The Large-track podium:
- 1st place: Guri Zabergja, Jeta Bekteshi, Albanot Makolli (accuracy: 98.5%)
- 2nd place: Florian Diederichs, Lukas Koenig, Nina Pant (accuracy: 97.0%)
- 3rd place: Samuel Boehm, Narges Dastanpour, Vytautas Jankauskas (accuracy: 95.0%)
Congratulations!

Questions?

If you have a question, please post it in the ILIAS forum (so everyone can benefit from the answer).
Alternatively, you can also email dl-orga-ws20@cs.uni-freiburg.de

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	online

Course type:	Oberseminar: Reading Group (no ECTS) -- Automated Machine Learning
Location:	online
Web page:	Reading Group

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	TBA
Location:	online only
Web page:	Open Projects

Course type:	Lab Course
Time:	Wednesday, 14:00-16:00
Location:	online only
Organizer:	Fabio Ferreira , Jörg Franke , Arbër Zela , Frank Hutter (Frank is on parental leave this summer, so please contact the other organizers for questions)
Web page:	Website

Course type:	Flipped-classroom Lecture + Exercise
Time:	Tuesday 14:00 - 15:30
Location:	online only
Organizer:	André Biedenkapp , Arbër Zela , Katharina Eggensperger , Matthias Feurer , Frank Hutter (Frank is on parental leave this summer, so please contact the other organizers for questions)
Web page:	AutoML

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	Building 052, 01-033

Course type	Oberseminar: Reading Group (no ECTS) -- Automated Machine Learning
Time:	Thursdays, 10:45
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	TBA
Location:	Kitchen, Building 074
Web page:	Open Projects

Course type:	Seminar
Time:	Thursday, 12:30-14:00 (first meeting, October 24th, 12:15-13:45)
Location:	Building 101, Room 01 018
Organizers:	Frank Hutter , Katharina Eggensperger , Matthias Feurer , Noor Awad , Arbër Zela
Web page:	, HisInOne

Seminar on Bayesian Optimization

The seminar language will be English (even if everyone is German-spoken, to practice presentation skills in English). *First meeting:

• 24th October, 12:15-13:45, SR 01 018 (Building 101).

*Regular meetings:

• Every Thursday, 12:30-14:00, SR 01 018 (Building 101).

Background

Bayesian optimization is a popular method for blackbox function optimization. Blackbox function are functions for which no assumptions are made, which means that neither the derivatives nor the smoothness are known. Furthermore, function evaluations might be noisy and are typically assumed to be expensive. Finally, the only knowledge about blackbox functions is the call signature which allows one to query the function for different input values (and observe the outcome). These properties make Bayesian optimization an ideal method for hyperparameter optimization. In this seminar we will read papers on both the foundations of Bayesian optimization and recent research aiming to apply Bayesian optimization to state-of-the-art deep learning models.

Schedule

##

14.11

## 1. Practical Bayesian Optimization of Machine Learning Algorithms - Spearmint 2. A Tutorial on Bayesian Optimization - with focus on the knowledge gradient and entropy search ##

21.11

## 1. Scalable Bayesian Optimization Using Deep Neural Networks 2. Automating Bayesian optimization with Bayesian optimization ##

28.11

## 1. Portfolio Allocation for Bayesian Optimization 2. Meta-Learning Acquisition Functions for Transfer Learning in Bayesian Optimization ##

05.12

## 1. Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets 2. BOHB: Robust and Efficient Hyperparameter Optimization at Scale ##

09.01

## 1. Neural Architecture Search with Bayesian Optimization and Optimal Transport 2. Scalable Hyperparameter Transfer Learning ##

16.01

## 1. A Flexible Framework for Multi-Objective Bayesian Optimization using Random Scalarizations 2. High-Dimensional Bayesian Optimization via Additive Models with Overlapping Groups ##

23.01

## 1. Constrained Bayesian Optimization with Noisy Experiments 2. Hyperparameter Importance Across Datasets ##

13.02

## 1. Scalable Meta-Learning for Bayesian Optimization using Ranking-Weighted Gaussian Process Ensembles 2. Scalable Global Optimization via Local Bayesian Optimization

Requirements

• Machine Learning
• Statistical Pattern Recognition
• Automated Machine Learning would be good

Material

• How to give a good presentation slides

• Seminar structure

• Summary template

• Brief introduction to Bayesian optimization

• Slides for Brochu et al.

Further information

For questions, please send an email to one of the organizers: Arbër Zela

Course type:	Lecture + Exercise
Time:	Lecture: Monday 14:15 - 15:45; Exercise: Thursday 10:00 - 11:30
Location:	Lecture: Building 101, HS 00 026 (μ-SAAL); Exercise: Building 082, HS 00 006 (KinoHörsaal)
Organizers:	Frank Hutter , Joschka Bödecker , Abhinav Valada , Arber Zela , Raghu Rajan , Jörg Franke , Andreas Sälinger
Web page:	, ILIAS

Foundations of Deep Learning

Deep learning is one of the fastest growing and exciting fields. This course will provide you with a clear understanding of the fundamentals of deep learning including the foundations to neural network architectures and learning techniques, and everything in between.

Course Overview

The following are the dates for the in-class discussions (The video lectures will be released on the Tuesday the week earlier): 21.10.19 - Week 1: Overview of Deep Learning
04.11.19 - Week 2: From Logistic Regression to MLPs
11.11.19 - Week 3: Backpropagation
18.11.19 - Week 4: Optimization
25.11.19 - Week 5: Regularization
02.12.19 - Week 6: Convolutional Neural Networks (CNNs)
09.12.19 - Week 7: Recurrent Neural Networks (RNNs)
16.12.19 - Week 8: Practical Methodology & Architectures
13.01.19 - Week 9: Hyperparameter Optimization & Neural Architecture Search
20.01.20 - Week 10: Uncertainty in Deep Learning
27.01.20 - Week 11: Auto-Encoders, Variational Auto-Encoders, GANs
03.02.20 - Week 12: Group Presentations; Project Kickoff
The course will be taught in english and we will follow a flipped classroom approach.

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	Building 052, 01-033

Course type:	Oberseminar: Reading Group (no ECTS) -- Automated Machine Learning
Time:	Thursdays, 10:45
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	Lab Course
Time:	Wednesday, 14:00-16:00
Location:	SR-00-019, Building 74 (MST Pool)
Organizer:	Frank Hutter , Aaron Klein , Matilde Gargiani , Jörg Franke , Arbër Zela
Web page:	Website

Course type:	Lecture + Exercise
Time:	Monday 14:00 - 16:00 and Friday 14:00 - 16:00
Location:	SR-00-007, Building 106
Organizer:	Marius Lindauer , Frank Hutter , André Biedenkapp , Arbër Zela , Katharina Eggensperger
Web page:	AutoML

Course type:	Lecture + Exercise
Time:	Monday 16:00 - 18:00 and Thursday 16:00 - 18:00
Location:	Building 101 - HS 00 026
Organizers:	Frank Hutter , Michael Tangermann , Marius Lindauer
Web page:	ILIAS

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	Building 052, 01-033

Course type:	Oberseminar: Reading Group (no ECTS) -- Machine Learning and Automated Algorithm Design
Time:	Thursdays, 10:45
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	TBA
Location:	Kitchen, Building 074
Web page:	Open Projects

Course type:	Lab Course
Time:	Tuesday 14:00 - 16:00
Location:	Building 82, room 00-006 (Kinohörsaal)
Organizers:	Aaron Klein , Matilde Gargiani , Frank Hutter ,
Web page:	Website

Course type:	Lecture + Exercise
Time:	Lecture: Monday 14:15 (s.t.) - 15:45; Exercise: Tuesday 12:30 - 14:00
Location:	Lecture: Building 106 SR 00-007; Exercise: Building 051 SR 00 006
Organizers:	Marius Lindauer , André Biedenkapp , Frank Hutter
Web page:	, ILIAS

Machine Learning for Automated Algorithm Design

News

• All news, slides and exercise sheets will be online at ILIAS

Background

Automated algorithm design gets more and more important because algorithms get more complex and a developer has to make calls about many aspects, e.g., which subroutine to use or how to set parameters (also known as magic constants). From the perspective of a user, there often exist more than one algorithm to solve a given problem. So, how can we efficiently choose a well performing algorithm? In practice, it is even worse: There exist no single well-performing parameter setting or best algorithm for all kind of possible inputs. Therefore, we have to determine the best settings for new inputs again and again which can be (i) a really time-consuming and tedious task and (ii) a human is often biased by her own experience which is often not optimal. To automatically solve all these problems, automated algorithm design can be used to determine well-performing algorithm parameter settings, to select the best algorithm for a given input, or even to help to develop and implement better algorithms. In our course "ML4AAD", we will discuss all these problems and how to solve them. To this end, we will use a lot machine learning and optimization techniques to solve them effectively. Furthermore, we will focus on hard combinatorial problems and machine learning as exemplary classes of algorithms to apply automated algorithm design in the course. We strongly recommend that you know the foundations of (i) artificial intelligence (AI) and (ii) machine learning (ML) in order to attend the course. The participants should have attended at least in one other course for AI and ML in the past. In particular, the lectures are partitioned in several parts:

• Algorithm selection

• Algorithm design for solving combinatorial problems and machine learning

• Empirical evaluation of algorithms

• Hyperparameter optimization

• Neural architecture search

• Algorithm configuration

• Combinations of algorithm configuration and algorithm selection

• Algorithm analysis

Dynamics

__The course will be in English.__ We will meet weekly for a lecture and an exercise. Roughly every week, there will be a new exercise sheet. Most exercises will be practical and involve teamwork (teams of 2 students!) so that you learn how to apply automatic algorithm design in practice. The exercises are no requirement for the exam (however highly recommended) but to get a grade in the end, you have to pass the exercises.

• Lecture: Monday 14:15 (s.t.) - 15:45 in Building 106 SR 00 007

• Exercise: Tuesday 12:30 (s.t.) - 14:00 in Building 051 SR 00 006

Exam

In the end, everyone (no team work!) has to implement a larger project which is the base of the final oral exam. In the first 15 minutes of the oral exam, you have to present your project and in the second 15 minutes, we will ask you to answer questions about further course material.

Course type:	Lecture + Exercise
Time:	Lecture: Monday 12:15 - 13:45
Location:	Building 82, room 00-006 (Kinohörsaal)
Organizers:	Frank Hutter , Raghu Rajan (exercises)
Web page:	ILIAS

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 15:00
Location:	Building 052, 01-033

Course type:	Oberseminar: Reading Group (no ECTS) -- Machine Learning and Automated Algorithm Design
Time:	Mondays, 10:45
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	Lab Course
Time:	Monday, 10:00-12:00
Location:	SR-00-019, Building 74 (MST Pool)
Organizer:	Frank Hutter , Marius Lindauer
Web page:	Machine Learning for Automated Algorithm Design (Lab Course)
	ILIAS

Course type:	Seminar
Time:	Friday, 14:00-16:00
Location:	SR-01-016, Building 101
Organizer:	Frank Hutter , Marius Lindauer
Web page:	Advanced Topics in Automated Algorithm Design
	ILIAS

Course type:	Seminar
Time:	Tuesday, 14:00-16:00 The seminar will start on April 24th
Location:	SR-00-007, Building 106
Organizer:	Frank Hutter , Matthias Feurer , Aaron Klein , Mathilde Gargiani ,
Web page:	Advanced Deep Learning

Course type:	Lecture + Exercise
Time:	Tuesday 12:15 - 13.45 and Thursday 12:15 - 13:45
Location:	Building 101 - HS 00 036
Organizers:	Michael Tangermann , Frank Hutter
Web page:	ILIAS

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	Building 052, 01-033

Course type:	Oberseminar: Reading Group (no ECTS) -- Machine Learning and Automated Algorithm Design
Time:	Fridays, 10:45
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	TBA
Location:	Kitchen, Building 074
Web page:	Open Projects

Course type:	Lab Course
Time:	Monday 14:00 - 16:00
Location:	Kinohörsaal, Building 082
Organizers:	Aaron Klein , Frank Hutter , Joschka Bödecker
Web page:	Machine Learning and Computer Vision Track

Course type:	Lecture + Exercise
Time:	Lecture: Friday 14:00 - 15:30
Location:	Building 106 SR 00-007 (Friday)
Organizers:	Frank Hutter , Marius Lindauer , Joschka Bödecker , Gabriel Kalweit
Web page:	, ILIAS

Reinforcement Learning

News

• All news and exercise sheets will be online at ILAS

Background

The lecture deals with methods of Reinforcement Learning that constitute an important class of machine learning algorithms. Starting with the formalization of problems as Markov decision processes, a variety of Reinforcement Learning methods are introduced and discussed in-depth. The connection to practice-oriented problems is established throughout the lecture based on many examples.

Dynamics

__The course will be in English.__ The course will be held in a flipped classroom manner. The students have to watch a lecture video each week at home and we will meet weekly to answer questions, discuss the new content and start to solve exercises. Roughly every week, there will be a new exercise sheet. Most exercises will be practical (teams of 2-3 students!) so that you learn how to apply reinforcement learning in practice. You must get 50% points from the exercise to participate in the final exam (see below).

• Building: 106; Room: SR 00 007

• Session: Friday 14:00 - 15:30

Exam

In the end, everyone has to implement a larger project which is the base of the final oral exam. In the first 15 minutes of the oral exam, you have to present your project and in the second 15 minutes, we will ask you to answer questions about further course material.

Course type:	Lecture + Exercise
Time:	Lecture/Exercise: Tuesday 12:30 - 14:00
Location:	Building 051 SR 03-026
Organizers:	Frank Hutter , Stefan Falkner
Web page:	ILIAS

Course type:	Lecture + Exercise
Time:	Lecture: Monday 12:20 - 13:50; Lecture/Exercise: Wednesday 12:20 - 13:50
Location:	Building 106 SR 00-007
Organizers:	Frank Hutter , Marius Lindauer , André Biedenkapp
Web page:	, ILIAS

Machine Learning for Automated Algorithm Design

News

• The first session will be at the Oct 18th. We will not meet at the Oct 16th.

• All news and exercise sheets will be online at ILAS

Background

Automatic algorithm design gets more and more important because algorithms get more complex and a developer has to make calls about many aspects, e.g., which subroutine to use or how to set parameters (also known as magic constants). From the perspective of a user, there often exist more than one algorithm to solve a given problem. So, how to choose a well performing algorithm? In practice, it is even worse: There exist no single well-performing parameter setting or best algorithm for all kind of possible inputs. Therefore, we have to determine the best settings for new inputs again and again which can be (i) a really time-consuming and annoying task and (ii) a human is often biased by her own experience which is often not optimal. To automatically solve all these problems, automatic algorithm design can be used to determine well-performing algorithm parameter settings, to select the best algorithm for a given input, or even to help to develop and implement better algorithms. In our course "ML4AAD", we will discuss all these problems and how to solve them. To this end, we will use a lot machine learning and optimization techniques to solve them effectively. Furthermore, we will focus on hard combinatorial problems as an example class of algorithms to apply automatic algorithm design in the course. We strongly recommend that you know the foundations of (i) artificial intelligence (AI) and (ii) machine learning (ML) in order to attend the course. The participants should have attended at least in one other course for AI and ML in the past. In particular, the lectures are partitioned in 8 modules:

• Introduction to NP-Hard problems

• Methods for solving combinatorial problems

• Empirical evaluation of algorithms

• Statistical models of the empirical hardness of NP-hard problems

• Algorithm selection

• Algorithm configuration

• Meta-learning

• Hyperparameter optimization

Dynamics

__The course will be in English.__ We will meet weekly for a lecture and an exercise. Roughly every second week, there will be a new exercise sheet. Most exercises will be practical (teams of 2 students!) so that you learn how to apply automatic algorithm design in practice. You must get 50% points from the exercise to participate in the final exam (see below).

• Building: 106; Room: SR 00 007

• Lecture: Monday 12:20 - 13:50

• Exercise: Wednesday 12:20 - 13:50

Exam

In the end, everyone (no team work!) has to implement a larger project which is the base of the final oral exam. In the first 15 minutes of the oral exam, you have to present your project and in the second 15 minutes, we will ask you to answer questions about further course material.

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	Building 052, 01-033

Course type:	Oberseminar: Reading Group (no ECTS) -- Machine Learning and Automated Algorithm Design
Time:	Mondays, 10:45
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	Wednesday, May 3rd 2017 at 1pm (s.t.)
Location:	Kitchen, Building 074
Web page:	Open Projects

Course type:	Lecture + Exercise
Time:	Lecture: Monday 12:30 - 14.00; Exercise: Thursday 12:30 - 14:00
Location:	Building 101 - HS 00 036 SCHICK - SAAL
Organizers:	Michael Tangermann
Web page:	ILIAS

Course type:	Seminar
Time:	Friday, 14:00-15:30
Location:	SR-00-007, Building 106
Organizer:	Frank Hutter , Stefan Falkner , Marius Lindauer
Web page:	Advanced Topics in Deep Learning
	ILIAS

Course type:	Oberseminar: Reading Group (no ECTS) -- Reinforcement Learning
Time:	Monday, 16:00
Location:	Kitchen, Building 074

Course type:	Oberseminar: Reading Group (no ECTS) -- Deep Learning and Hyperparameter Optimization
Time:	Wednesdays, 16:00
Location:	Building 052, 01-033

Course type:	Oberseminar: Reading Group (no ECTS) -- Machine Learning and Automated Algorithm Design
Time:	Wednesdays, 13:15
Location:	Kitchen, Building 074
Web page:	Reading Group

Course type:	B.Sc./M.Sc. projects
Kickoff meeting:	Monday, October 24th 2016
Location:	Kitchen, Building 074
Web page:	Open Projects

Course type:	Lecture + Exercise
Time:	Friday, 14:00-16:00. Beginning: Friday, October 21, 2016
Location:	Kinohörsaal, Building 082
ECTS Credits:	4 or 6
Requirements:	Bachelor or Master students. Fundamental programming skills in Python are recommended.
Organizer:	Ilya Loshchilov , Frank Hutter
Web page:	Deep Learning Course, Automated Algorithm Design Section

Course type:	Seminar
Time:	Monday, 10:15-11:45
Location:	SR-00 031, Building 51
Organizer:	Frank Hutter , Aaron Klein , Stefan Falkner
Web page:	Deep Learning and Hyperparameter Optimization
	ILIAS

Course type:	Lecture + Exercise
Time:	Tuesday 12.30 - 14.00, Thursday 12.30 - 13.15 (Lecture); Thursday 13.15 - 14.00 (Exercise)
Location:	Building 101 - HS 00 036 SCHICK - SAAL
Organizers:	Michael Tangermann , Frank Hutter
Web page:	Ilias
	Daphne
	Setup

Course type:	Lab course (Praktikum)
Time:	Monday 14.15-15.45
Location:	TBA
Organizers:	Frank Hutter
Web page:	Lab Course Automated Machine Learning applied to Machine Learning Challenges

Course type:	Lecture + Exercise
Time:	Tuesday 08:15 - 09:45 (Lecture); Thursday 08:30 - 10:00 (Exercise)
Location:	Building: 106; Room: SR 00 007
Organizers:	Frank Hutter
Web page:	Machine Learning and Optimization for Algorithm Design

Course type:	Praktikum
Time:	Mondays, 14:15-15:45 (first meeting: April 20)
Location:	Building 074 - 00-023
Organizer:	Frank Hutter
Web page:	Praktikum Bayesian Optimization

Course type:	Seminar
Time:	Wednesday, 10:00-12:00 (first meeting: 22th October)
Location:	Building 051 - HS 03 026
Organizers:	Frank Hutter
Web page:	AI for Automated Algorithm Design

Course type:	Lecture
Time:	Tuesdays, 16:00-17:30 and Thursdays, 08:15-09:45
Location:	Building 051 - 00-031
Organizers:	Martin Riedmiller , Frank Hutter , Manuel Blum
Web page:	Machine Learning
Slides:	Hyperparameter optimization slides: ML.pdf , ML2.pdf , ML3.pdf
HPOlib google group:	click here

Course type:	Seminar
Time:	Thursdays, 14:15-15:45 (first meeting: May 8)
Location:	Building 052 - HS 02-017
Organizers:	Frank Hutter , Marius Lindauer
Web page:	Automated Parameter Tuning and Algorithm Configuration

Course type:	Seminar
Time:	Fridays, 12:15-13:45
Location:	S1-018
Organizer:	Frank Hutter
Web page:	Automated Parameter Tuning and Algorithm Configuration