I am a PhD student at the Fraunhofer Institute for Intelligent Analysis and Information Systems in Sankt Augustin. Particularly I am working in the group for Computational Aspects of Mining and Learning within the department for Knowledge Discovery. My PhD topic is "Mining Interesting Patterns - Efficiency and Effectiveness" and it is supervised by Prof. Dr. Stefan Wrobel and Dr. Tamas Horvath. See the research interest section below for more details on my scientific work.

In addition I assist Prof. Wrobel in his machine learning and data mining courses at the University of Bonn. If you are a student there who is looking for a practicum and/or thesis in the area of pattern mining, please contact me.

Publications

Below you find a list of my publications. Please contact me in case you have questions or comments regarding any of them.

• Mario Boley, Tamás Horváth, Axel Poigné, Stefan Wrobel. Listing Closed Sets of Strongly Accessible Set Systems with Applications to Data Mining. Theoretical Computer Science, 411(3):691-700, 2010. [ BibTeX ]
• Mario Boley, Thomas Gärtner. On the Complexity of Constraint-Based Theory Extraction. Discovery Science, 12th International Conference, DS 2009, Porto, Portugal, volume 5808 of LNCS, page 92-106. 2009. Springer. [ BibTeX ]
• Mario Boley, Henrik Grosskreutz. Approximating the number of frequent sets in dense data. Knowledge and Information Systems, 21(1):65-89, oct 2009. [ BibTeX ]
• Mario Boley, Henrik Grosskreutz. Non-redundant Subgroup Discovery Using a Closure System. Machine Learning and Knowledge Discovery in Databases, European Conference, ECML PKDD 2009, Bled, Slovenia, Proceedings, Part I, page 179-194. 2009. Springer. [ pdf | slides | video | BibTeX ]
• Mario Boley, Tamás Horváth, Stefan Wrobel. Efficient Discovery of Interesting Patterns Based on Strong Closedness. Statistical Analysis and Data Mining, 2(5-6):346-360, 2009. [ BibTeX ]
• Mario Boley, Tamas Horváth, Axel Poigne, Stefan Wrobel. New Results on Listing Closed Sets of Strongly Accessible Set Systems. 7th International Workshop on Mining and Learning with Graphs, 2009. [ pdf | BibTeX ]
• Shankar Vembu, Thomas Gärtner, Mario Boley. Probabilistic Structured Predictors. Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence (UAI), 2009. [ pdf | slides | BibTeX ]
• Mario Boley, Tamás Horváth, Stefan Wrobel. Efficient Discovery of Interesting Patterns Based on Strong Closedness. Proceedings of the SIAM International Conference on Data Mining, SDM 2009, April 30 - May 2, 2009, Sparks, Nevada, USA, page 1002-1013. 2009. SIAM. [ implementation | electr.edt. | BibTeX ]
• Mario Boley, Henrik Grosskreutz. A Randomized Approach for Approximating the Number of Frequent Sets. Proceedings of the IEEE International Conference on Data Mining, 2008. [ implementation | BibTeX ]
• Mario Boley, Tamás Horváth, Axel Poigné, Stefan Wrobel. Efficient Closed Pattern Mining in Strongly Accessible Set Systems (Extended Abstract). PKDD, page 382-389. 2007. [ pdf | BibTeX ]
• Mario Boley. On Approximating Minimum Infrequent and Maximum Frequent Sets. Discovery Science, page 68-77. 2007. [ pdf | slides | BibTeX ]
• Mario Boley. Intelligent Pattern Mining via Quick Parameter Evaluation. NSF Symposium on Next Generationof Data Mining and Cyber-Enabled Discovery for Innovation (NGDM), 2007. [ pdf | BibTeX ]

Research Interests

Overview:
My thesis investigates the computational aspects of intelligent and efficient data analysis systems. In particular I am focusing on local pattern mining methods like for instance association rules, subgroups, or graph mining. An intelligent and efficient system includes:

1. intelligent methods for restricting the output to truly interesting patterns
2. intelligent support for setting the corresponding parameters
3. provable efficient algorithms for listing the remaining output
4. monitoring these algorithms throughout their execution
   

Technically this involves analyzing the complexity of as well as designing algorithms for listing different pattern classes and the quick computation of corresponding key quantities describing the expected output. In particular the latter task often requires approximation algorithms employing greedy strategies or Markov chain Monte Carlo methods. Throughout all of my studies I set a high value on both:

• rigorous theoretical investigation of the involved computational problems and the algorithms solving them
• practical implementation and optimization of the algorithms for real-world application

Illustrative List of Detailed Question/Tasks I am Interested in:

• developing a provable efficient algorithm for listing closed error-tolerant frequent patterns
• self-reducibility of computing the number of closed frequent sets
• complexity of listing discriminative patterns
• developing a suitable notion of "subgroup-closedness" of hypotheses
• computing prefixtrees of optimal compactness
• complexity of listing all circuits of a matroid
• development of a computational model for listing problems that does not suffer from the drawbacks of the Turing model for these tasks