WiDS REGENSBURG 2021
Virtual Conference (April 13th and 14th 2021)
MATERIAL
Recordings:
https://youtu.be/Q01OTACxqsM (Schmid)
https://youtu.be/N1igWnzZRQ8 (Poblete)
https://youtu.be/FPl2VkKxBIc (Mügschl-Scharf)
https://youtu.be/NWIWgE2uZHo (Hoffmann)
https://youtu.be/kWoVOWs3kfs (Puga)
AGENDA
We are proud to have presented an impressive range of talks! The keynotes were given by renowned faculty from academia, whereas our technical talks by experienced data scientists from both academia and industry had a more applied flavour. The program was completed by short lightning talks from promising students who wanted to share their latest data science projects with the audience.
Day 1 (April 13th 2021)
1:00 pm: Opening Remarks
1:10 pm: Keynote: Dynamic Algorithms for Graphs and Clustering by Prof. Monika Henzinger (Universität Wien)
Clustering is a key technique which is used for example in pattern recognition and data mining. If the data set is large and dynamically changing a data structure is desirable that efficiently maintains a clustering while the data set is modified by an arbitrary sequence of data point insertions and deletions. Such a data structure is called a dynamic clustering algorithm. In some applications the data exhibits a structure that can be represented by a graph and that can be used to determine interesting patterns in the data. As the data changes the corresponding graph representation changes dynamically as well, requiring a dynamic graph algorithm to maintain the patterns of interest. We will give a survey of the state of the art in dynamic clustering and dynamic graph algorithms.
2:10 pm: Break
2:30 pm: Gaussian Processes in a Nutshell by Dr. Barbara Rakitsch (Bosch Center for AI)
3:00 pm: A Comparison of Model-Based Methods for Imputing Incomplete Multivariate Time Series by Noor Jamaludeen (Universität Magdeburg)
3:30 pm: Using Satellite Images to Detect Migrant Boats in Distress by Elisabeth Wittmann (OTH Regensburg)
3:40 pm: Balanced Diet for Fit[ting] Neural Networks by Cindy Döttl (OTH Regensburg)
3:50 pm: Starting Up! From Research to Business by Dr. Monika Mügschl-Scharf (FUTUR – Gründerberatung)
Can your research be the basis for a promising business idea? Who can help you to start a successful business? Where do you get funding? FUTUR – Gründerberatung provides support in answering these questions.
4:00 pm: Break
4:15 pm: A.I. in Semiconductor Manufacturing by Dr. Liana Movsesyan & Dr. Yao Yang (Infineon)
4:45 pm: Automated Data Analyses in Production: an Algorithm Toolkit by Veronika Völkl (OSRAM Opto Semiconductors)
5:15 pm: Break
5:30 pm: Keynote: Deciphering fairness desiderata for machine learning by Prof. Ruth Urner (York University, Toronto)
The expansion of machine learning based predictive tools into numerous segments of people’s lives has led to growing concerns about these tools replicating and exacerbating undesired biases in decision making. These concerns have been met by the machine learning research community with a large (and growing) array of mathematical fairness notions and algorithmic techniques enhancing these. It turns out that there is a variety of intuitively reasonable notions of what should be expected of a fair predictor and, moreover, different such notions may lead to conflicting conclusions. In this talk, I will outline some of the more common fairness notions in the machine learning literature, discuss their rationale and provide some high level guidelines as to which aspects of a given problem we may wish to consider when choosing a suitable notion of fairness.
6:30 pm: Closing Remarks
Day 2 (April 14th 2021)
1:00 pm: Opening Remarks
1:10 pm: Methods & Challenges for the Safety Assurance of Deep Neural Networks in Computer Vision by Gesina Schwalbe (Continental)
1:40 pm: A Deep Learning Approach to Radiation Dose Estimation by Dr. Dr. Theresa Götz (Fraunhofer IIS)
2:10 pm: Break
2:30 pm: Towards Robust Hate Speech Detection with Contextual Embeddings by Julia Hoffmann (Universität Regensburg)
2:40 pm: Triplet-based Learning with the Help of Crowdlabeling on Medical Data by Anne Rother (Universität Magdeburg)
2:50 pm: Data Science applied to Medical Research by Clara Puga (Universität Magdeburg)
3:00 pm: Keynote: Mining Social Networks to Learn about Rumors, Hate Speech, Bias and Polarization by Prof. Barbara Poblete (Universidad de Chile)
Online social networks are a rich resource of unedited user-generated multimedia content. Buried within their day-to-day chatter, we can find breaking news, opinions and valuable insight into human behaviour, including the articulation of emerging social movements. Nevertheless, in recent years social platforms have become fertile ground for diverse information disorders and hate speech expressions. This situation poses an important challenge to the extraction of useful and trustworthy information from social media. In this talk I provide an overview of existing work in the area of social media information credibility, starting with our research in 2011 on rumor propagation during the massive earthquake in Chile in 2010. I discuss, as well, the complex problem of automatic hate speech detection in online social networks. In particular, how our review of the existing literature in the area shows important experimental errors and dataset biases that produce an overestimation of current state-of-the-art techniques. Specifically, these issues become evident at the moment of attempting to apply these models to more diverse scenarios or to transfer this knowledge to languages other than English. As a particular way of dealing with the need to extract reliable information from online social media, I talk about two applications, Twically and Galean. These applications harvest collective signals created from social media text to provide a broad view of natural disasters and real-world news, respectively
4:00 pm: Break
4:15 pm: Leveraging Transfer Learning for Small Datasets by Maya Sekeran (AVL)
4:45 pm: Anomaly – everything but normal?!? by Andrea Spichtinger (Syskron)
5:15 pm: Break
5:30 pm: Keynote: The Third Wave of Artificial Intelligence – From Blackbox Machine Learning to Explanation-Based Cooperation by Prof. Ute Schmid (Universität Bamberg):
Machine learning is considered as an important technology with high potential for many application domains in industry as well as society. Impressive results of deep neural networks, for instance for image classification, promise that complex decision models can be derived from raw data without the need of feature engineering (end-to-end learning). However, there is an increasing awareness of the short-comings of data-intensive black box machine learning approaches: For many application domains it is either impossible or very expensive to provide the amount an quality of data necessary for deep learning. Furthermore, legal or ethical or simply practical considerations often make it necessary that decisions of learned models are transparent and comprehensible to human decision makers. Consequently, AI researchers and practitioners alike proclaim the need for the so-called 3rd Wave of AI to overcome the problems and restrictions of an AI which is focusing on purely data-driven approaches. In the talk, it is shown that machine learning research offers many alternative, often less data-intensive, approaches. Current topics and approaches for explainable and interactive machine learning will be introduced and illustrated with some example applications.
6:30 pm: Closing Remarks