Uncertainty representation in tree structured data
| školitel: | doc. Ing. Václav Šmídl, Ph.D. |
| e-mail: | zobrazit e-mail |
| typ práce: | dizertační práce |
| zaměření: | MI_MM, MI_AMSM |
| klíčová slova: | uncertainty representation, ensemble methods, Monte Carlo, |
| popis: | Classical methods of machine learning assume that data are available in the form of a vector of a known dimension. However, this is rarely the case in practice where data are available in structured forms such as incomplete database entries or json files. Mapping of such data into the assumed vector structure (i.e. feature extraction) is typically done by a human, which is a laborious task that is susceptible to errors. Recent methods aim to avoid this step by directly training models on the full data space using the paradigm of deep sets. Discriminative learning is nowadays a common task, however, the uncertainty of trained models is not well known. The aim of the thesis is to propose a method for uncertainty representation of these models. The proposed models will be continually validated on real data from network security domain. |
| literatura: | [1] Zaheer, M., Kottur, S., Ravanbakhsh, S., Poczos, B., Salakhutdinov, R. R., & Smola, A. J. (2017). Deep sets. In Advances in neural information processing systems (pp. 3391-3401). [2] Bishop, C. M. (2006). Pattern recognition and machine learning. springer. [3] Pevný, T., & Somol, P. (2017, June). Using neural network formalism to solve multiple-instance problems. In International Symposium on Neural Networks (pp. 135-142). Springer, Cham. [4] Létal, V., Pevný, T., Šmidl, V., & Somol, P. (2015). Finding New Malicious Domains Using Variational Bayes on Large-Scale Computer Network Data. In NIPS Workshop: Advances in Approximate Bayesian Inference (pp. 1-10). [5] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press. |
| naposledy změněno: | 18.06.2020 10:02:36 |
za obsah této stránky zodpovídá:
Ľubomíra Dvořáková | naposledy změněno: 12.9.2011