Marcos L. P. Bueno

Postdoctoral researcher
Department of Mathematics and Computer Science
Eindhoven University of Technology, the Netherlands

Guest at
Institute for Computing and Information Sciences
Radboud University Nijmegen, the Netherlands



Bio
Publications
Software
Service
Contact



Bio

Currently, I am a postdoctoral researcher at the mathematics and computer science department of TU Eindhoven, the Netherlands.

My current research is on artificial intelligence and machine learning, and includes topics as: 

• Automated machine learning (autoML)

• Probabilistic graphical models (e.g., Bayesian networks)

• Temporal data, data streams

• Biomedical and process mining applications

Prior to that, I have been a tenured assistant professor at the computer science department of the Federal University of Uberlândia, Brazil. I taught theoretical and practical Information Systems Bachelor courses and supervised Bachelor thesis students.

I hold a PhD degree in computer science from the Leiden University, the Netherlands. In my thesis I proposed new probabilistic graphical models founded on Bayesian networks and hidden Markov models, enabling a better understanding of temporal problems in medical and business process domains, among others. I also investigated the discovery of subgroups of temporal data that are substantially different from the whole data, by combining dynamic Bayesian networks with the exceptional model mining framework.
You can download my PhD thesis here (a print copy is available upon request).

I obtained my MSc degree in computer science from the Federal University of Uberlândia, Brazil.

[Home]


Publications

Please see DBLP or my Google scholar profile

[Home]


Software

• Exceptional model mining using dynamic Bayesian networks

  Source code in R for learning exceptional dynamic Bayesian networks from temporal data. You can use your own data, which should be sequences of repeated measurements (e.g. in patient data, each patient has one or more symptoms measured over multiple time points). It is also possible to simulate data from synthetic models then relearn the models (e.g. for comparing the ground-truth and empirical models). For more details, please see the paper:

  Temporal Exceptional Model Mining using Dynamic Bayesian Networks. AALTD workshop of ECML/PKDD 2020 (in press, earlier version here).

  Download the data and code here.

• Partitioned dynamic Bayesian networks (simulated data)

  Source code in R for simulating data from specified PDBNs (partitioned dynamic Bayesian networks). The R code for learning PDBNs from data is available upon request.

  PDBNs are probabilistic graphical models that extend dynamic Bayesian networks (DBNs) for modelling multiple distribution regimes over time (i.e. non-homogeneous distributions). For more details, please refer to the paper:

  Understanding disease processes by partitioned dynamic Bayesian networks.
  Journal of Biomedical Informatics, Volume 61, June 2016, Pages 283-297. [pdf]

  Download the code here.

[Home]


Service

• Program committee: IJCAI 2020, AIME 2020, ECML/PKDD 2020, SGAI 2017-20
• (Sub)reviewer: NIPS 2020, BRACIS 2019, ECML/PKDD 2017-19, PGM 2016-18, AAAI 2017, UAI 2017, TPAMI, j. biomedical informatics, j. AI in medicine, int. j. approximate reasoning, j. heuristics