Topics in Algorithmic Data Analysis SS'18


News

more ▾

Course Information

Type Advanced Lecture (6 ECTS)
Lecturer Dr. Jilles Vreeken
Email tada-staff (at) mpi-inf.mpg.de
Lectures Thursdays, 14–16 o'clock in Room E1.7 0.01.
(and, a few times, Tuesdays, 14–16 o'clock in Room E2.1 0.01)
Summary In this advanced course we'll be investigating hot topics in data mining that the lecturer thinks are cool. This course is for those of you who are interested in Data Mining, Machine Learning, Data Science, Big Data Analytics – or, as the lecturer prefers to call it – Algorithmic Data Analysis. We'll be looking into how to discover significant and useful patterns from data, efficiently measure non-linear correlations and determine causal relations, as well as how to analyse structured data such as time series and graphs.

Schedule

Month Day Topic Slides Assignment Req. Reading Opt.
Reading
April 17* (Tue) Introduction, Practicalities PDF 1st assignment out
26 Useful Patterns PDF [1,2] [9,10]
May 3 Jilles travelling – no class deadline 1st, 2nd out
8* (Tue) Actionable Patterns PDF [3] [11,12,13]
17 Significance PDF [4] Sec 4.4, 5 [14,15,16,17]
24 Dependence PDF [5] [18,19,20]
29* (Tue) Causality PDF [21] Ch 1, 6.1, 6.3 [21] Ch 6.5, 6.7
31 yay holiday – no class deadline 2nd, 3rd out
June 7 Jilles travelling – no class
14 Causal Inference PDF [21] Ch 2.1, 4 [22,23]
19* (Tue) Rumours in Graphs PDF [6] [24,25,26,27]
28 Graph Summarization PDF [7] [28,29,30,31]
July 5 One, Two,...Tensor! PDF deadline 3rd, 4th out [8] [32,33]
12 It's not Fair! PDF [34]
19 Wrap-Up PDF
26 oral exams deadline 4th
October 8 re-exams

* Tuesday at 14:00 till 16:00 in E2.1 Room 0.01

All report deadlines are on Thursdays at 14:00. In doubt, the assignment web page determines the exact date and time.

Materials

All required and optional reading will be made available. You will need a username and password to access the papers outside the MMCI/MPI network. Contact the lecturer if you don't know the username or password.

In case you do not have a strong enough background in data mining, machine learning, or statistics, these books may help to get you on your way [35,21,36] The university library kindly keeps hard copies of these books available in a so-called Semesteraparat.

Required Reading

[1] Vreeken, J., van Leeuwen, M. & Siebes, A. Krimp: Mining Itemsets that Compress. Data Mining and Knowledge Discovery, 23(1):169-214, Springer, 2011.
[2] van Leeuwen, M. & Vreeken, J. Mining and Using Sets of Patterns through Compression. In Frequent Pattern Mining, Aggarwal, C. & Han, J., pages 165-198, Springer, 2014.
[3] Atzmueller, M. Subgroup Discovery. WIRE's Data Mining and Knowledge Discovery, 5:35-49, Wiley, 2015.
[4] Vreeken, J. & Tatti, N. Interesting Patterns. In Frequent Pattern Mining, Aggarwal, C. & Han, J., pages 105-134, Springer, 2014.
[5] Hoefer, T., Przyrembel, H. & Verleger, S. New Evidence for the Theory of the Stork. Paediatric and Perinatal Epidemiology, 18(1):88-92, 2004.
[6] Prakash, B.A., Vreeken, J. & Faloutsos, C. Spotting Culprits in Epidemics: How many and Which ones?. In Proceedings of the 12th IEEE International Conference on Data Mining (ICDM), Brussels, Belgium, IEEE, 2012.
[7] Koutra, D., Kang, U., Vreeken, J. & Faloutsos, C. VoG: Summarizing Graphs using Rich Vocabularies. In Proceedings of the SIAM International Conference on Data Mining (SDM), Philadelphia, PA, pages 91-99, SIAM, 2014.
[8] Kolda, T.G. & Bader, B.W. Tensor decompositions and applications. SIAM Rev., 51(3):455-500, 2009.

Optional Reading

[9] van Leeuwen, M., Vreeken, J. & Siebes, A. Identifying the Components. Data Mining and Knowledge Discovery, 19(2):173-292, 2009.
[10] Smets, K. & Vreeken, J. The Odd One Out: Identifying and Characterising Anomalies. In Proceedings of the 11th SIAM International Conference on Data Mining (SDM), Mesa, AZ, pages 804-815, Society for Industrial and Applied Mathematics (SIAM), 2011.
[11] Friedman, J.H. & Fisher, N.I. Bump Hunting in High-Dimensional Data. Statistics and Computing, 9:123-143, Springer, 1999.
[12] Boley, M., Goldsmith, B.R., Ghiringhelli, L. & Vreeken, J. Identifying Consistent Statements about Numerical Data with Dispersion-Corrected Subgroup Discovery. Data Mining and Knowledge Discovery, 31(5):1391-1418, Springer, 2017.
[13] Kalofolias, J., Boley, M. & Vreeken, J. Efficiently Discovering Locally Exceptional yet Globally Representative Subgroups. In Proceedings of the 17th IEEE International Conference on Data Mining (ICDM), New Orleans, LA, IEEE, 2017.
[14] Gionis, A., Mannila, H., Mielikäinen, T. & Tsaparas, P. Assessing data mining results via swap randomization. ACM Transactions on Knowledge Discovery from Data, 1(3), ACM, 2007.
[15] Hanhijärvi, S., Ojala, M., Vuokko, N., Puolamäki, K., Tatti, N. & Mannila, H. Tell me something I don't know: randomization strategies for iterative data mining. In Proceedings of the 15th ACM International Conference on Knowledge Discovery and Data Mining (SIGKDD), Paris, France, pages 379-388, ACM, 2009.
[16] Jaynes, E. On the rationale of maximum-entropy methods. Proceedings of the IEEE, 70(9):939-952, IEEE, 1982.
[17] De Bie, T. Maximum entropy models and subjective interestingness: an application to tiles in binary databases. Data Mining and Knowledge Discovery, 23(3):407-446, Springer, 2011.
[18] Reshef, D.N., Reshef, Y.A., Finucane, H.K., Grossman, S.R., McVean, G., Turnbaugh, P.J., Lander, E.S., Mitzenmacher, M. & Sabeti, P.C. Detecting Novel Associations in Large Data Sets. Science, 334(6062):1518-1524, 2011.
[19] Crescenzo, A.D. & Longobardi, M. On cumulative entropies. Journal of Statistical Planning and Inference, 139(2009):4072-4087, 2009.
[20] Nguyen, H.V., Müller, E., Vreeken, J. & Böhm, K. Multivariate Maximal Correlation Analysis. In Proceedings of the 31st International Conference on Machine Learning (ICML), Beijing, China, pages 775-783, JMLR, 2014.
[21] Peters, J., Janzing, D. & Schölkopf, B. Elements of Causal Inference. MIT Press, 2017.
[22] Janzing, D. & Schölkopf, B. Causal Inference Using the Algorithmic Markov Condition. IEEE Transactions on Information Technology, 56(10):5168-5194, 2010.
[23] Janzing, D., Mooij, J., Zhang, K., hang, , Lemeire, J., Zscheischler, J., Daniusis, P., Steudel, B. & Schölkopf, B. Information-geometric Approach to Inferring Causal Directions. , 182-183:1-31, 2012.
[24] Lappas, T., Terzi, E., Gunopulos, D. & Mannila, H. Finding effectors in social networks. In Proceedings of the 16th ACM International Conference on Knowledge Discovery and Data Mining (SIGKDD), Washington, DC, pages 1059-1068, ACM, 2010.
[25] Shah, D. & Zaman, T. Rumors in a Network: Who's the Culprit?. IEEE Transactions on Information Technology, 57(8):5163-5181, 2011.
[26] Sundareisan, S., Vreeken, J. & Prakash, B.A. Hidden Hazards: Finding Missing Nodes in Large Graph Epidemics. In Proceedings of the SIAM International Conference on Data Mining (SDM'15), SIAM, 2015.
[27] Rozenshtein, P., Gionis, A., Prakash, B.A. & Vreeken, J. Reconstructing an Epidemic over Time. In Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD'16), pages 1835-1844, ACM, 2016.
[28] Chakrabarti, D., Papadimitriou, S., Modha, D.S. & Faloutsos, C. Fully automatic cross-associations. In Proceedings of the 10th ACM International Conference on Knowledge Discovery and Data Mining (SIGKDD), Seattle, WA, pages 79-88, 2004.
[29] Kang, U. & Faloutsos, C. Beyond Caveman Communities: Hubs and Spokes for Graph Compression and Mining. In Proceedings of the 11th IEEE International Conference on Data Mining (ICDM), Vancouver, Canada, pages 300-309, IEEE, 2011.
[30] Shah, N., Koutra, D., Zou, T., Gallagher, B. & Faloutsos, C. TimeCrunch: Interpretable Dynamic Graph Summarization. , pages 1055-1064, 2015.
[31] Goeble, S., Tonch, A., Böhm, C. & Plant, C. MeGS: Partitioning Meaningful Subgraph Structures Using Minimum Description Length. In Proceedings of the IEEE International Conference on Data Mining (ICDM), pages 889-894, IEEE, 2016.
[32] Skillicorn, D. Understanding Complex Datasets: Data Mining with Matrix Decompositions. Chapman \& Hall/CRC, 2007.
[33] Acar, E. & Yener, B. Unsupervised Multiway Data Analysis: A Literature Survey. IEEE Trans. Knowl. Data En., 21(1):6-20, 2009.
[34] Dwork, C., Hardt, M., Pitassi, T., Reingold, O. & Zemel, R. Fairness through Awareness. In Proceedings of the 3rd Innovations in Theoretical Computer Science Conference (ITCS), ACM, 2012.
[35] Aggarwal, C.C. Data Mining - The Textbook. Springer, 2015.
[36] Wasserman, L. All of Statistics. Springer, 2005.

Course format

The course has two hours of lectures per week. There are no weekly tutorial group meetings. Instead, you will have to write four essays based on the material covered on the lectures and scientific articles assigned by the lecturer.

Structure and Content

In general terms, the course will consist of

  1. lectures, and
  2. assignments that include critically reading scientific articles
At a high level, the topics we will cover will include
  1. Mining Interesting Patterns
  2. Mining Causal Dependencies
  3. Mining Structured Data
Loosely speaking, students will learn about current hot topics in exploratory data analysis, with an emphasis on statistically well-founded approaches, including those based on information theoretic principles.

Assignments

Students will individually do one assignment per topic – four in total. For every assignment, you will have to read one or more research papers and hand in a report that critically discusses this material and answers the assignment questions. Reports should summarise the key aspects, but more importantly, should include original and critical thought that show you have acquired a meta level understanding of the topic – plain summaries will not suffice. All sources you've drawn from should be referenced. The expected length of a report is 3 pages, but there is no limit.

The deadlines for the reports are at 14:00 Saarbrücken standard-time. You are free to hand in earlier.

Grading and Exam

The assignments will be graded in scale of Fail, Pass, Good, and Excellent. Any assignment not handed in by the deadline is automatically considered failed, and cannot be re-done. You are allowed to re-do one Failed assignment: you have to hand in the improved assignment within two weeks. Two failures mean you are not eligible for the exam, and hence failed the course.

You can earn up to three bonus points by obtaining Excellent or Good grades for the assignments. An Excellent grade gives you one bonus point, as do every two Good grades, up to a maximum of three bonus points. Each bonus point improves your final grade by 1/3 assuming you pass the final exam. For example, if you have two bonus points and you receive 2.0 from the final exam, your final grade will be 1.3. You fail the course if you fail the final exam, irrespective of your possible bonus points. Failed assignments do not reduce your final grade, provided you are eligible to sit the final exam.

The final exams will be oral. The final exam will cover all the material discussed in the lectures and the topics on which you did your assignments. The main exam will be on July 26th. The re-exam will be on October 8th. The exact time slot per student will be announced per email. Inform the lecturer of any potential clashes as soon as you know them.

Prerequisites

Students should have basic working knowledge of data analysis and statistics, e.g. by successfully having taken courses related to data mining, machine learning, and/or statistics, such as Information Retrieval and Data Mining, Machine Learning, Probabilistic Graphical Models, Statistical Learning, etc.