As part of the Research Focus Cognitive Sciences, the Machine Learning in Cognitive Science Lab is dedicated to improving the cognitive abilities of machines and reducing the friction in human-computer interaction. We develop novel signal processing and deep learning algorithms for the analysis of sensory data and investigate new approaches for interacting with machines such as through EEG or eye tracking.


Book cover "Companion Technology"
Image: Springer International Publishing

New Book on "Companion Technology"

A new book on "Companion Technology: A Paradigm Shift in Human-Technology Interaction" by Biundo & Wendemuth (eds), 2017 is now available from Springer International Publishing. It deals with the systematic and interdisciplinary study of cognitive abilities and their implementation in technical systems, and covers aspects such as multimodality, individuality, adaptability, availability, cooperativeness, and trustworthiness.

In collaboration with Michael Kotzyba, Tatiana Gossen and Andreas Nürnberger from the Data and Knowledge Engineering Group at the Otto von Guericke University Magdeburg, Sebastian Stober contributed the chapter  "Model-Based Frameworks for User Adapted Information Exploration: An Overview." This work was done within the Transregional Collaborative Research Centre SFB/TRR 62 "Companion-Technology for Cognitive Technical Systems" funded by the German Research Foundation (DFG).

Book cover "Companion Technology"
Image: Springer International Publishing

Open Position: Lab Infrastructure Manager (Research Assistant)

We are currently looking for a student research assistant with the necessary technical skills and interest to manage our lab infrastructure within the UPracticeML project, which includes several GPU compute servers as well as dedicated machines for storage and virtualization. Suitable candidates should have experience with linux (Ubuntu) server administration and network security. Relevant services comprise, for instance, LDAP authentication, a reverse HTTP-proxy, Jupyterhub, Docker and Mattermost. You will also help to develop and deploy a solution for scheduling and distributing compute jobs across the different servers. There is furthermore the option to get involved in the our lab’s research projects, which span a wide range of deep learning applications.

The position is open for up to 19 hours per week and can be extended for up to two years. For any inquiries, please contact Sebastian Stober. To apply, please send an email with your CV.

November 9: Sebastian Stober received E-Learning UP Award 2017

Sebastian Stober received the E-Learning UP Award 2017 for his lecture "Representation Learning - From a Deep Learning Perspective." The award is intended to acknowledge and highlight the efforts and achievements of lecturers who successfully used the possibilities of the digital media in an exemplary way to facilitate learning.

November 1: Project UPracticeML started

On November 1, 2017, the new project UPracticeML - a cooperation with the Applied Computational Linguistics Discourse Research Lab of Prof. Manfred Stede with a duration of 2 years - has started. The project is funded by the Federal Ministry of Education and Research (BMBF) and aims to extend the machine learning curriculum in the Cognitive Systems Master at the University of Potsdam. From this grant, approximately 200.000 Euro will be invested in dedicated hardware infrastructure to support deep learning research and teaching. Furthermore, two new PhD students will join the MLCog Lab.

October 16: Talk on “Deep Learning for Music Recommendation and Generation” by Sander Dieleman (DeepMind) at the Berlin MIR Meetup

October 16, 2017 | 19:30h | SoundCloud Berlin | registration required

Abstract: The advent of deep learning has made it possible to extract high-level information from perceptual signals without having to specify manually and explicitly how to obtain it; instead, this can be learned from examples. This creates opportunities for automated content analysis of musical audio signals. In this talk, I will discuss how deep learning techniques can be used for audio-based music recommendation. I will also discuss my ongoing work on music generation in the raw waveform domain with WaveNet.

Bio:Sander Dieleman is a Research Scientist at DeepMind in London, UK, where he has worked on the development of AlphaGo and WaveNet. He was previously a PhD student at Ghent University, where he conducted research on feature learning and deep learning techniques for learning hierarchical representations of musical audio signals. During his PhD he also developed the Theano-based deep learning library Lasagne and won solo and team gold medals respectively in Kaggle’s “Galaxy Zoo” competition and the first National Data Science Bowl. In the summer of 2014, he interned at Spotify in New York, where he worked on implementing audio-based music recommendation using deep learning on an industrial scale.

New Deep Learning Courses

Starting with the winter term 2017/2018, the former course "Representation Learning - from a Deep Learning Perspective" is replaced by two new courses: "Introduction to Deep Learning" taught by Sebastian Stober and "Learning Generative Models" taught by Andreas Krug within the Junior Teaching Professionals Programme of the Potsdam Graduate School. Interested students are encourage to apply for the courses. Further information can be found at the course websites.

September 7: Paper "Adaptation of the Event-Related Potential Technique for Analyzing Artificial Neural Nets" presented at CCNeuro 2017

The paper “Adaptation of the Event-Related Potential Technique for Analyzing Artificial Neural Nets” by Andreas Krug and Sebastian Stober was accepted at the inaugural Conference on Cognitive Computational Neuroscience (CCNeuro) in New York City. This new conference aims to establish a forum for discussion among neuroscience, cognitive science, and artificial intelligence researchers who are dedicated to understanding the neural computations that underlie complex behavior. Andreas Krug presented the paper as a poster at the conference on September 7, 2017. Paper (PDF)Poster (PDF)