Classic Contributions to Machine Learning Aleksandar Bojchevski, Rebekka Burkholz

News

03.05.2022

Learning about Machine Learning

Dear Students,

As promised here are a few optional additional resources that cover some machine learning basics:

Dear Students,

As promised here are a few optional additional resources that cover some machine learning basics:

There are plenty of other resources online that you can easily find with a quick search. If you come across some resource that you find particularly valuable or helpful don't hesitate to share it with your fellow students.

02.05.2022

Paper Assignment

Dear Students,

The paper assignment is complete and almost all students got their first choice. The first round of presentations will be as follows:

Group 1 - Introduction - 30.05.22

  • Mohamad Altamer      A Few Useful Things to Know About Machine... Read more

Dear Students,

The paper assignment is complete and almost all students got their first choice. The first round of presentations will be as follows:

Group 1 - Introduction - 30.05.22

  • Mohamad Altamer      A Few Useful Things to Know About Machine Learning
  • Yufeng Jiang                Introduction to Convolutional Neural Networks                                             
  • Max Biwersi                A Gentle Introduction to Graph Neural Networks                                            
  • Severin Adler              Dropout: A Simple Way to Prevent Neural Networks from Overfitting                         

Group 2 - Methods and Applications - 13.06.22

  • Leonard Butz              Deep Physical Neural Networks Trained with Backpropagation                                
  • Albert Klimenko          Zero-Shot Text-to-Image Generation                          
  • Xudong Zhang            Machine Learning and Phone Data Can Improve Targeting of Humanitarian Aid                                      
  • Julian Blaes                 Tackling Climate Change with Machine Learning

Group 3 - ML and Society - 20.06.22

  • Thomas Boisvert         Deep Learning: A Critical Appraisal                
  • Emily Ries                   Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification    
  • Zoi Giagtzoglou           Delphi: Towards Machine Ethics and Norms

The next step is to read the paper and prepare the first (practice) presentation. You shouldn't underestimate the workload -- we suggest that you start with the first read through of the paper as soon as possible. Remember you are free to use other resources (blogs, videos, lectures, etc.) to help you better understand the material. We also highly encourage you to consult the resources that we shared in the kick off meeting on how to prepare a good presentation. We will also share a few additional resources on machine learning basics in the next few days.

We ask you to submit your presentation here at least 3 days before your presentation date (i.e. by 27.05 for group 1, by 10.06 for group 2, and by 17.06 for group 3). You are still allowed to fine tune and improve your presentation after this deadline but the submission should not significantly deviate from what you present. You are allowed to use any software for your presentation (PowerPoint, Keynote, Google Docs, etc.) but please submit a PDF export.

Your presentation should be 15 minutes long. We highly recommended that you do at least one practice run to make sure that the story flows nicely and that you are on time!

If you have any questions do not hesitate to get in touch.

- Rebekka and Aleksandar

26.04.2022

Paper choice

Please indicate your preferences for a paper that you are going to present here till May 1st, 2022.

24.04.2022

Kick-off meeting

Hello everyone,

We are excited to meet you at our kick-off meeting on Monday, Apr 25th, 2022, at 16:15 on Zoom. We will share the slides that were presented on the meeting in the Materials section.

 

 

Classic Contributions to Machine Learning

Machine learning has not only prepared the modern breakthroughs in computer vision and natural language processing. It has also a successful history of enabling data driven insights. We will discuss classic concepts and seminal papers in the field that cover different techniques ranging from linear regression to deep learning. We will discover joint underlying principles and challenge some of them by modern observations related to overparameterization.
Are the simplest models always the best choice? And is counting parameters really the best way to measure model complexity?
Join the seminar if you enjoy thinking about these kind of questions.

Course Organization

The goal of the proseminar is to learn giving scientific presentations. You will study a given topic, summarize it, and present it to the other participants.

First, we will provide advice and resources that will help you prepare and deliver a scientific presentation. After you have received a topic, you will give an initial practice presentation and receive detailed feedback from all participants. You will have the chance to incorporate this feedback before giving a final graded presentation. At the end of the semester you also need to submit a short 2-page summary.

In the first part of the semester, we will have practice presentations for 2-3 topics each week. The second part of the semester will consist of the final presentations. Attendance is mandatory and the meetings will be in person or virtual. The exact meeting dates will be determined in the kick-off meeting.

 

 

 



Privacy Policy | Legal Notice
If you encounter technical problems, please contact the administrators