Home | english  | Impressum | Datenschutz | Sitemap | KIT

Recent Research Topics in Workflow Analysis, Privacy and Machine Learning

Recent Research Topics in Workflow Analysis, Privacy and Machine Learning
Typ: Seminar (S)
Semester: SS 2019

Termin der Vorbesprechung und Themenvergabe:

Dienstag 30.4.2019, 13:00 - 14:00 Uhr

SR 348 (3. Stock)
50.34 Informatik, Kollegiengebäude am Fasanengarten     


Prof. Dr.-Ing. Klemens Böhm
Jutta Mülle
Christine Tex
Jakob Bach

LVNr.: 2400109

Anmeldung per E-Mail an sekretariat.boehm@ipd.kit.edu oder durch Eintragen in die Listen im Sekretariat Prof. Böhm.

Current topics of the DBIS Research Group of Prof. Klemens Böhm are Databases, Information Systems and Data Science. In this seminar, we will focus on the subtopics Workflow Management Systems (WfMS), Privacy of Database Systems and Meta-Learning.

- WfMS are important to model and automate processes in a variety of application domains, such as energy markets and auctions. Current trends in research concern all phases of the lifecycle of workflows. They are driven by innovations like cloud infrastructures, improved flexibility and adaptability of processes, enhanced influence of data on workflows, and process mining and verification techniques to achieve higher quality of business processes.

- Privacy of database queries: Nowadays, it is common to store and query databases that contain private data. Important examples are location or energy consumption databases that contain information on, e.g., the daily habits of the data owners. In this seminar, we focus on querying such databases in a privacy-preserving way. To this end, we study the popular concept of differential privacy.

- In machine learning (ML), the field of meta-learning has received growing attention in the last few years. Meta-learning tries to predict how well a ML model will perform on a particular data set, without training the ML model. It uses characteristics of data sets as meta-features and ML performance measures like accuracy as meta-target. Thus, one can apply standard ML approaches like classification, regression, or instance-based methods on such meta-data sets.

In this seminar, topics are recent research on workflow modeling and verification, privacy of querying databases and meta-learning.

In the seminar the language is English. We will organize the presentations in 3 blocks.