Allgemein

Stereotypes in Artificial Intelligence

Today, Artificial Intelligence systems are used in many areas and support among other things decision-making processes (e.g. in checking creditworthiness or assessing the probability of a criminal returning to crime). It is precisely these systems that one wishes to be free of prejudice. Unfortunately, however, this does not always correspond to reality. But to what extent do AI systems show prejudice in general and gender bias in particular? The talk focuses on the area of Commonsense Reasoning, which finds an application in the field of automatic assistants such as Siri and Alexa, and considers the question of how prejudice can be measured in this area.

Dr. Claudia Schon is a Klara-Marie-Faßbinder-Gastprofessor at TH Bingen.

The talk will take place on the 31.01.20 at 14:00 in room 05-514.

Our paper "Modeling Multi-Label Recurrence in Data Streams" won the best paper award at ICBK 2019!

Our paper "Identifying drug side effects from social media using active learning and crowd sourcing", which was a joint work of Sophie Burkhardt, Julia Siekiera, Josua Glodde, Miguel A. Andrade-Navarro, and Stefan Kramer was accepted at the Pacific Symposium for Biocomputing (PSB) conference as an oral presentation.

Our paper "Decoupling Sparsity and Smoothness in the DirichletVariational Autoencoder Topic Model" was accepted at the journal for machine learning research (JMLR) pdf code

With the increasing amount of available medical data, computing power and network speed, modern medical
imaging is facing an unprecedented amount of data to analyze and interpret. Phenomena such as Big Data-omics stemming from several diagnostic procedures and novel multi-parametric imaging modalities tend to produce almost unmanageable quantities of data. The paper addresses the aforementioned context by assuming that a novel paradigm in massive data processing and automation becomes necessary in order to improve diagnostics and facilitate personalized and precision medicine for each patient. Traditional machine learning concepts have demonstrated many shortcomings when it comes to correctly diagnose fatal diseases. At the same time static graph networks are unable to capture the fluctuations in brain processing and monitor disease evolution. Therefore, artificial intelligence and deep learning are increasingly applied in oncologic medical imaging because they excel at providing quantitative assessments of biomedical imaging characteristics. On the other hand, novel concepts borrowed from modern control have paved the path for a dynamic graph theory that can predict neurodegenerative disease evolution and replace longitudinal studies. We chose two important topics, brain data processing and oncologic imaging to show the relevance of these concepts. We believe that these novel paradigms will impact multiple facets of radiology but are convinced that it is unlikely that they will replace radiologists any time in the near future since there are still many challenges in the clinical implementation.

Our paper "Pairwise Learning to Rank by Neural Networks Revisited: Reconstruction, Theoretical Analysis and Practical Performance" was accepted to ECML. Congratulations to Lukas!

We welcome a new member to our group: Julia Siekiera has just started as a research assistant!

Sophie Burkhardt won the faculties dissertation award for her PhD thesis "Online Multi-label Classification using Topic Models"