I am happy to be part of the Technical Program Committee of ISIPTA 2019. That is the 20-year anniversary edition of the world’s main forum on imprecise probabilities. If we had to describe its central theme in one sentence, it would be that “There’s more to uncertainty than probabilities.” Indeed, a wide range of other …

I am co-editing a Special Issue on “Imprecise Probabilities, logic and Rationality” in the International Journal of Approximate Reasoning (IJAR-elsevier). This SI intends to contribute to the state-of-the-art of the interactions and connections between imprecise probabilities and logic, and more generally with formal theories of rationality, the hope being that this cross-disciplinary view will lead …

I have implemented a Python library for modelling, inference and updating with Almost Desirable Gambles (ADG) models. It is both friendly and flexible. It works with continuous, discrete and mixed variables. Here you can find some additional info, setup instructions and 4 examples (notebooks): https://github.com/PyRational/PyRational/blob/master/notebooks/index.ipynb The notebooks (and relative examples) are very simple, their purpose …

Janez Demsar has reimplemented our library about Bayesian hypothesis testing for comparing competing algorithms in ML. It can now be installed directly with pip. Hereafter, a brief description. Baycomp is a library for Bayesian comparison of classifiers. Functions compare two classifiers on one or on multiple data sets. They compute three probabilities: the probability that …

In a previous post we derived the Covariance Inequality from a Bayesian (Imprecise probability) perspective. There is another and more elegant way to derive this inequality: $$Cov(X,Y)^2\leq Var(X)Var(Y)$$ To do that, we introduce again our favorite subject, Alice. Let us summarize the problem again. Assume that there two real variables $X,Y$ and that Alice only …

I thank the organizers of ISIPTA 2017 for having invited me. Here you can find the link to my Keynote talk:

QM is based on four main axioms, which were derived after a long process of trial and error. The motivations for the axioms are not always clear and even to experts the basic axioms of QM often appear counter-intuitive. In a recent paper [1], we have shown that: It is possible to derive quantum mechanics …

The following post discusses how to use a Bayesian hierarchical test (and also the Python module that implements it) to compare classifiers assessed via m-runs k-folds cross-validation. In the Bayesian correlated t-test and also in the frequentist correlated t-test, we can only analyze cross-validation results on a single dataset. In particular, the Bayesian correlated t-test …