I gave a chat, entitled "Explainability like a company", at the above celebration that mentioned expectations regarding explainable AI And the way can be enabled in programs.
Weighted model counting often assumes that weights are only specified on literals, often necessitating the necessity to introduce auxillary variables. We look at a different tactic based upon psuedo-Boolean features, leading to a far more normal definition. Empirically, we also get SOTA benefits.
The Lab carries out research in synthetic intelligence, by unifying Understanding and logic, with a new emphasis on explainability
I attended the SML workshop during the Black Forest, and discussed the connections between explainable AI and statistical relational Finding out.
Our paper (joint with Amelie Levray) on Finding out credal sum-products networks continues to be acknowledged to AKBC. These kinds of networks, coupled with other types of probabilistic circuits, are appealing simply because they assurance that sure types of probability estimation queries can be computed in time linear in the size from the network.
The report, to appear from the Biochemist, surveys several of the motivations and approaches for building AI interpretable and accountable.
Keen on schooling neural networks with sensible constraints? Now we have a completely new paper that aims towards full gratification of Boolean and linear arithmetic constraints on training at AAAI-2022. Congrats to Nick and Rafael!
The post introduces a general reasonable framework for reasoning about discrete and constant probabilistic versions in dynamical domains.
We review preparing in relational Markov final decision processes involving discrete and constant states and steps, and an unidentified amount of objects (via probabilistic programming).
Within the paper, we exploit the XADD knowledge framework to accomplish probabilistic inference in blended discrete-continuous spaces https://vaishakbelle.com/ effectively.
Paulius' work on algorithmic techniques for randomly creating logic plans and probabilistic logic plans has become acknowledged for the principles and practise of constraint programming (CP2020).
Our MLJ (2017) report on preparing with hybrid MDPs was approved for presentation at the journal observe.
If you are attending AAAI this yr, you could possibly have an interest in trying out our papers that contact on fairness, abstraction and generalized sum-product or service complications.
Our paper on synthesizing programs with loops inside the presence of probabilistic noise, accepted the journal of approximate reasoning, has also been accepted to the ICAPS journal track. Preprint to the entire paper in this article.