Research Topics of Machine Learning Group

Deep Learning

We develop and evaluate deep architectures for a variety of complex real-world tasks such as automated image classification, natural language processing and human action recognition. The main focus of our work lies on convolutional nets and deep autoencoders.

Read more

Interpretable Machine Learning

Powerful machine learning methods such as deep neural networks perform impressively well, but have a significant disadvantage, the lack of interpretability. We develop techniques which enable one to understand and interpret the reasoning of these complex systems.

Read more

Data-driven Video Analysis

We investigate the usage of data-driven techniques for a variety of complex tasks on videos, including annotation of human actions, compressed-domain analysis and video compression.
Finding optimal representations for video data is one of the goals of this research.

Read more

EEG-based Quality Assessment

We develop EEG-based techniques for extracting brain correlates of user’s perceived quality.
With these quality measures we aim to improve current video compression and 3D visualization standards.

Read more

Big Data Analytics

In the era of Big Data intelligent and efficient analytics techniques take on greater significance. Our research focuses on the development of fast approximation techniques such as local sensitive hashing and novel information fusion approaches.

Read more

Robust Methods

Real-world data is often nonstationary and contains outliers which can heavily bias statistical analysis. Our research focuses on the development of algorithms based on robust divergences which can cope with these challenges.

Read more