List of publications

For an automatically updated version of this list (in case I forget to add one in the future), check my Google Scholar or this page.

Publication Year
IteROAR: Quantifying the Interpretation of Feature Importance Methods (pre-register @ NeurIPS) 2021
XAI Handbook: Towards a Unified Framework for Explainable AI (ICCVW) 2021
Contextual Classification Using Self-Supervised Auxiliary Models for Deep Neural Networks (ICPR) 2020
P ≈ NP, at least in Visual Question Answering (ICPR) 2020
Revisiting Sequence-To-Sequence Video Object Segmentation with Multi-Task Loss and Skip-Memory (ICPR) 2020
Adversarial Defense based on Structure-to-Signal Autoencoders (WACV) 2020
GTC US: Understanding Deep Networks Through Properties of the Input Space (Invited Talk) 2019
GTC Munich: Towards Understanding Deep Network Behaviour (Invited Talk) 2018
What do Deep Networks like to See (CVPR) 2018
Classless Association using Neural Networks (ICANN) 2017
Symbolic Association using Parallel Multilayer Perceptron (ICANN) 2016
Multi-camera vehicle identification system (Patent: US 9449258 B1) 2015
Application Exploring of Ubiquitous Pressure Sensitive Matrix as Input Resource for Home-Service Robots 2014
Retrieving Objects, People and Places from a Video Collection: TRECVID'12 Instance Search Task. 2012

Some of my work

Here, you can find projects and demos I have been working on over the years.

Parameter-free Clustering

Implementation of DBClass-d algorithm. A non-parametric clustering algorithm.

  • Repository

Sentiment Video Dashboars

In order to have an overview when extracting features with a DNN on video, I created a dashboard to aggregate all the information that can be gathered at once.

  • Live Demo

ML Reading Group

We get together each week and discuss a paper related to Machine Learning. A space to learn and catch up with the latest research.

  • Details

Explainable AI and Neural Networks

Leading a project on explainable AI. Sponsored by the german ministry for education and research.

  • Project Site (DFKI)

Deep Learning Course

I developed a full course on Deep Learning with practical excercises for university and industry.

Areas of expertise

These are some fields and technologies I have learned and feel comfortable working with.

Python

My weapon of choice. I've been programing in Python for well over 10 years. I'm very familiar with NumPy, Matplotlib, Pandas, scikit-learn, Pillow and SciPy. I rather open an iPython console than the calculator.

Keras/TensorFlow/PyTorch

I have developed countless prototypes using one of these three frameworks. Depending on the application, I resort to Keras for quick yet powerful prototypes, PyTorch for versatile ideas or Tensorflow for its efficiency.

Computer Vision

The area that drove me to AI. I'm fascinated by how images can be manipulated programatically (you know, instead of developing an accounting system). It still bewilders me how simple vision task for humans are so complex to model using computers.

Artificial Intelligence/Machine Learning

The extension of my quest in computer vision. It turns out we need to understand more general models of intelligence and other areas of AI may have some answers as well. Text, graph-theory, robotics. Every bit of introspection can be turned into code.

Linux

Happy to stay close to the penguin. I love the freedom and the power that comes with Linux. Bash enthusiast. I just edited this using nano :)

Scientific Research

The pillar of understanding. The foundation of reason. Science is the window that unveils the true beauty of the universe. It is not perfect but it is the best we have. Respect the rigor, the ideas and the evidence. Be open. Learn from it, challenge it and improve upon it.

Thoughts (personal blog)

A few thoughts and articles about computer vision, science or AI in general.

Get in touch

Send me an E-mail to First.Last at dfki.de