eXTraAI: EXplainable and TRAnsparent Artificial Intelligence

What We Do

One of the biggest challenges in the state-of-the-art machine learning is the complexity of predictive models. Recent techniques like deep neural networks, gradient boosting or random forests create models with thousands or even millions of parameters. This makes decisions generated by these black-box models completely opaque. Model obscurity undermines trust in model decisions, hampers model debugging, blocks model auditability, exposes models to problems with concept drift or data drift.
Our group works on next generation of explainers for predictive models. This will be a disruptive change in the way how machine learning models are created, deployed, and maintained.

Grants

,,MLGenSig: Machine Learning Methods for building the Integrated Genetic Signatures''

NCN 2017-2020

,,DALEX: Descriptive mAchine Learning EXplanations''

NCN 2018-2021

Publications

Key works

Recent contributions

Dalex: Explainers for complex predictive models in R (2018) Biecek, P., 2018, Journal of Machine Learning Research http://www.jmlr.org/papers/volume19/18-416/18-416.pdf

survxai: an R package for structure-agnostic explanations of survival models, Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek (2018), Journal of Open Source Software https://www.theoj.org/joss-papers/joss.00961/10.21105.joss.00961.pdf

iBreakDown: Uncertainty of Model Explanations for Non-additive Predictive Models, Alicja Gosiewska, Przemyslaw Biecek 2019, ARXIV: arXiv:1903.11420v1, https://arxiv.org/abs/1903.11420

Extended 3D and 4D cumulative plots for evaluation of unmatched incurred sample reanalysis, Piotr Rudzki, Michał Kaza, Przemysław Biecek (2018), Bioanalysis https://www.future-science.com/doi/abs/10.4155/bio-2017-0210

SMAD7 is a novel independent predictor of survival in patients with cutaneous melanoma, Kaczorowski M, Biecek P, Donizy P, Pieniazek M, Matkowski R, Halon A (2019), Translational Research https://www.ncbi.nlm.nih.gov/pubmed/30342000