Neural-Symbolic Computing Raven Beutner, Bernd Finkbeiner, Christopher Hahn, Niklas Metzger, Frederik Schmitt, Julian Siber

News

09.05.2022

LSF Registration

Please excuse the problems with LSF registration. The department confirmed that registering for the seminar should now be possible for everyone. If any problems continue to exist, please let us know. Note that you need to register until next Monday, May 16.

04.05.2022

First Group Talk on Feedforward and Recurrent Neural Networks

As a reminder, tomorrow at 10:15 am, we will have our first group talk on feedforward and recurrent neural networks. If you would like to attend in-person, please join us in room 1.06 in E1 1. If you prefer to participate remotely, use the same Zoom link as for our... Read more

As a reminder, tomorrow at 10:15 am, we will have our first group talk on feedforward and recurrent neural networks. If you would like to attend in-person, please join us in room 1.06 in E1 1. If you prefer to participate remotely, use the same Zoom link as for our kick-off meeting, which you can also find in the information tab of the CMS.

25.04.2022

Paper Assignment

We have assigned the papers and hope to have met your preferences. You can find your paper on your personal status page. We were able to assign everyone to a paper that they rated at least good and most of you to a paper that they rated very good. Your advisor will... Read more

We have assigned the papers and hope to have met your preferences. You can find your paper on your personal status page. We were able to assign everyone to a paper that they rated at least good and most of you to a paper that they rated very good. Your advisor will reach out to you shortly to introduce you to your group.

21.04.2022

Slides, Paper Preferences, and Timetable

You can now find the slides from today’s kick-off meeting and all papers in the materials section in the CMS. On your personal status page you can enter your paper preferences until Sunday, April 24, 11:59 pm. We use the tutorial assignment mechanism of the CMS to... Read more

You can now find the slides from today’s kick-off meeting and all papers in the materials section in the CMS. On your personal status page you can enter your paper preferences until Sunday, April 24, 11:59 pm. We use the tutorial assignment mechanism of the CMS to assign the papers. The number of the tutorial slot corresponds to the paper number as found in the materials section and as presented in today’s kick-off meeting.

Our next meeting will be the presentation on feedforward and recurrent neural networks on Thursday, May 5, 10:15 am. We added all meetings to the timetable in the CMS. You can import it to your personal calendar.

 

Neural-Symbolic Computing

You can register for the seminar at https://seminars.cs.uni-saarland.de/seminars22 until April 12th 23:59 CET.

The way our brain forms thoughts can be classified into two categories (according to Kahneman in his book “Thinking Fast and Slow”):

System 1: fast, automatic, frequent, stereotypic, unconscious.

  • Is this a cat or a dog?
  • What does this sentence mean in English?

System 2: slow, effortful, logical, conscious.

  • 17*16 = ?
  • If a -> b does b -> a?

The traditional view is that deep learning is limited to System 1 type of reasoning. Mostly because of the perception that deep neural networks are unable to solve complex logical reasoning tasks reliably. Historically, applications of machine learning were thus often restricted to sub-problems within larger logical frameworks, such as resolving heuristics in solvers.
In this seminar, we will explore new research that shows that deep neural networks are, in fact, able to reason on “symbolic systems”, i.e., systems that are built with symbols like programming languages or formal logics.

Example Topics:

  • What are your chances against an AI in a programming competition?
  • Is it possible to teach temporal logics to neural networks?
  • Can neural networks discover unknown connections in mathematics?

 

Requirements

Participants should have strong interest in logical reasoning and/or machine learning. There is, however, no formal prerequisite.

 

Organization

The seminar can be attended both in-person (Room 1.06 in E1 1) and remotely via Zoom.

The structure of the seminar is as follows:

In the first week of the seminar, you will choose a research paper on neural-symbolic computing.

Group Phase: Depending on your choice you will be assigned to a background topic of machine learning such as graph neural networks. In a team of three students, you will then prepare an informal lecture on your topic and the following discussion; presenting the basics to your fellow students. Each topic is assigned to an advisor that will help you with your preparations. This informal lecture will not be graded, so you can see it as a rehearsal. The group phase, thus, gives you the necessary foundations for the following individual phase.

Individual Phase: You will prepare, with the help of your advisor, a research talk on the paper you have chosen presenting the findings of the paper to your fellow students. This talk is weighted most in your final grade.

Project Phase: You will be given a neural-symbolic computing task. Each team is required to solve this problem with the methods explored in this seminar, for example by applying deep neural network architectures of your area. We do not expect own implementations of the discussed methods. You can use available libraries. This project has to be passed and it will not be graded.

 

Dates

Kick-off meeting: Thursday, April 21, at 10:15 am

Weekly meetings: Thursdays at 10:15 am



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