I gave a chat in the workshop on how the synthesis of logic and machine learning, Particularly spots which include statistical relational Finding out, can allow interpretability.
Last 7 days, I gave a chat on the pint of science on automated devices as well as their impact, referring to the topics of fairness and blameworthiness.
The paper tackles unsupervised program induction about mixed discrete-steady details, and is also accepted at ILP.
Should you be attending NeurIPS this calendar year, chances are you'll be interested in looking at our papers that contact on morality, causality, and interpretability. Preprints can be found about the workshop page.
Our paper (joint with Amelie Levray) on Understanding credal sum-product or service networks has been accepted to AKBC. This sort of networks, as well as other sorts of probabilistic circuits, are interesting since they ensure that specified forms of likelihood estimation queries is often computed in time linear in the scale with the community.
I gave a talk on our the latest NeurIPS paper in Glasgow even though also covering other strategies with the intersection of logic, Understanding and tractability. Thanks to Oana for that invitation.
Now we have a new paper accepted on Understanding ideal linear programming aims. We acquire an “implicit“ speculation development tactic that yields pleasant theoretical bounds. Congrats to Gini and Alex on having this paper recognized. Preprint in this article.
I gave a seminar on extending the expressiveness of probabilistic relational products with initially-purchase attributes, such as common quantification in excess of infinite domains.
Not too long ago, he has consulted with big banking companies on explainable AI and its influence in fiscal establishments.
In the paper, we exploit the XADD data structure to carry out probabilistic inference in mixed discrete-continuous spaces efficiently.
Paulius' work on https://vaishakbelle.com/ algorithmic methods for randomly generating logic courses and probabilistic logic courses has actually been approved towards the principles and practise of constraint programming (CP2020).
The framework is relevant to a sizable course of formalisms, including probabilistic relational types. The paper also studies the synthesis dilemma in that context. Preprint in this article.
I gave an invited tutorial the Bathtub CDT Artwork-AI. I included current traits and future trends on explainable equipment Discovering.
Conference connection Our work on symbolically interpreting variational autoencoders, in addition to a new learnability for SMT (satisfiability modulo principle) formulas obtained accepted at ECAI.