• Dresden


Florian Jug, Peter Steinbach, others (DAIS/CIBI)

09.09.2019 – 13.09.2019

SLUB, Klemperer-Saal
Zellescher Weg 18
01069 Dresden

Registration website:

The Dresden Deep Learning Hackathon ( #d3hack2019 ) is meant to bring together machine learning experts and scientific practitioners. Teams of 2-4 scientists can apply for the hackathon given a scientific problem they want to solve with machine learning. Upon approval, they will be assisted by one or two machine learning experts for 5 days consecutively! This effort is meant to give your team a head-start and potentially create an end-to-end machine learning solution for your science. The teams are motivated to publish a scientific paper about the hackathon efforts at dedicated conferences or in established journals - at best jointly with their mentors - after the hackathon. A win-win situation for all parties involved. 

The scope of scientific domains that can apply is not limited. For sure, our mentors have a given background mostly with regard to 2D or 3D images. So we will try to match that as close as possible. However, we are still in the process of fixing mentors (we have expressions of interest of about 5 more than listed below). We will also consider a limited amount of applications using standard machine learning (MLP, SVM, RandomForests,...).
If you are unclear whether your topic fits the hackathon, please reach out to us.

Most importantly, any team without a readily available data set for training will be discarded from the candidate list. In other words, if you are interested in applying machine learning to your data, you shouldn't use the hackathon to annotate your data.

The workshop admission fee amounts to € 300 per participant to cover room rent and catering. We are still looking for sponsors, so there is a non-negligible probability that the admission fee will be reduced in the future.

The call for applications closes on June 30, 2019, at 12pm AoE! After this date, a review board of mentors and organizers will judge the applications and send out confirmations to the applications until mid July the latest. The registration mechanism of participants will be circulated then.

For members of non-academic institutions: We cannot allow applications from non-academic institutions or industry to our hackathon. If you want to participate with a project as a company, this project needs to be embedded in a scientific group and the majority of team members need to be employed by a scientific institution.  On top, the results of the hackathon are expected to be published. So be prepared to undisclose your results and (at best) the data and code which produced these results.

Learning goals:
- Fundamentals of deep learning with CNNs.
- Keras API with the Tensorflow backend.
- How to define your deep net.
- How to train it.

Bring your own laptop. Keras and Tensorflow backend should already be installed. (We will have GPU nodes you can use if your laptop does not offer a fast GPU.) Solid understanding of the fundamentals of linear algebra. Programming skills (never programmed… that will not work out, sorry!)

DeepLearning, Keras, Tensorflow, Python

Keras, Python, Tensorflow

Florian Jug (This email address is being protected from spambots. You need JavaScript enabled to view it.) and Peter Steinbach (This email address is being protected from spambots. You need JavaScript enabled to view it.)