Research interests

My work lies in the fields of Artificial Intelligence and Machine Learning/Data Mining and is motivated by real world problems and societal challenges. Indicative topics of interest:

model extraction
model comparison
change detection and evolution monitoring
adaptive learning
model management and meta-mining
...

Indicative application domains:

social streams (topic detection, change detection, sentiment analysis, prediction)
sensor streams (stream clustering))
trajectory data (similarity queries, data warehouses, traffic mining)
user ratings (user recommendations, group recommendations)
user reviews (multicriteria recommendations, data redundancy issues)
bioarchaelogical data (isotopic fingerprint, model stability)
...

Research projects

Below is a list of the research projects I am/ have been involved in ...

* NoBIAS (Funding: EU -->), Role: Network coordinator & Project Investigator

AI-based systems are widely employed nowadays to make decisions that have far-reaching impacts on individuals and society. Their decisions might affect everyone, everywhere and anytime entailing risks, such as being denied a credit, a job, a medical treatment, or specific news. Bias may arise at all stages of AI-based decision making processes: (i) when data is collected, (ii) when algorithms turn data into decision making capacity, or (iii) when results of decision making are used in applications. Therefore, it is necessary to move beyond traditional AI algorithms optimized for predictive performance and embed ethical and legal principles in the training, design and deployment of AI algorithms to ensure social good while still benefiting from the potential of AI.

NoBIAS will develop novel methods for AI-based decision making without bias by taking into account ethical and legal considerations in the design of technical solutions. The core objectives of NoBIAS are to understand legal, social and technical challenges of bias in AI-decision making, to counter them by developing fairness-aware algorithms, to automatically explain AI results, and to document the overall process for data provenance and transparency.

* BIAS (Funding: Volkswagen Stiftung -->), Role: Project Investigator

AI techniques based on big data and algorithmic processing are increasingly used to guide decisions in important societal spheres, including hiring decisions, university admissions, loan granting, and crime prediction. However, there are growing concerns with regard to the epistemic and normative quality of AI evaluations and predictions. Our shared research question is: How can standards of unbiased attitudes and non-discriminatory practices be met in big data analysis and algorithm-based decision-making?

In approaching this question, we will provide philosophical analyses of the relevant concepts and principles in the context of AI (bias, discrimination, fairness), investigate their adequate reception in pertinent legal frameworks (data protection, consumer, competition, anti-discrimination law), and develop concrete technical solutions (debiasing strategies, discrimination detection procedures etc.).

* OSCAR (Funding: DFG -->), Role: Project Investigator

Many data accumulating in the Web reflect opinions on diverse subjects - products, institutions, events (e.g., elections) or topics (e.g., earth warming). Opinionated documents constitute a continuous stream; polarity learning on them delivers insights on the attitude of people towards each subject. Polarity learning algorithms must cope with classic Big Data characteristics: high volume and velocity of the arriving data, and volatility of the learned concepts, since subjects and attitudes of people toward certain subjects change over time. In OSCAR, we will develop classifiers that operate on an evolving feature space, adapt to changes in both vocabulary and data and operate with limited class labels.

* Transalpine mobility and knowledge transfer (Funding: DFG FOR 1670) , Role: Postdoc researcher

The project aims at the establishment of an isotopic fingerprint for bioarchaeological finds, especially cremations, and its application to archaeological and cultural-historical problems of the Late Bronze Age until Roman Times. From a computer science persective, our focus is on the development of innovative methods that allow complete scientific analysis of project related data despite their complexity. We focus on data management and automated data analysis (similarity search, cluster analysis, outlier recognition) for the establishment of small-scaled isotopic fingerprints.

* GeoPKDD (Geographic Privacy-aware Knowledge Discovery and Delivery) (FP6/IST project, 2005-09), Role: PhD researcher

GeoPKDD aims at developing theory, techniques and systems for knowledge discovery and delivery, based on new automated privacy-preserving methods for extracting user-consumable forms of knowledge from large amounts of raw data referenced in both space and time dimensions.

* Knowledge Discovery and Pattern Management - the PBMS approach (Funding: EPEAEK II / Heracletos Programme, 2003-2005), Role: PhD researcher

The goal of this project is the efficient management of data mining patterns extracted from large databases, with emphasis on the pattern similarity assesment problem.

* PANDA (Patterns for Next Generation Database Systems) (Funding: IST project 2001-2004), Role: PhD researcher

PANDA working group studies current state-of-the-art in pattern management and explores novel theoretical and practical aspects of a Pattern Base Management System (so-called, PBMS). PANDA's goal is the efficient and effective management of patterns; just as raw data are managed by traditional DBMS.

Prototypes

Below is a list of some prototypes I have been involved in ...

* MONIC - Modeling and Monitoring Cluster Transitions & FINGERPRINT - Summarizing Cluster Evolution in Dynamic Environments

MONIC is a framework for monitoring the evolution of a population in terms of the underlying cluster models. Since evolution is a permanent characteristic of the data, we also proposed FINGERPRINT for summarizing the evolution into few interesting clusters over time.