Don’t miss out on our next seminar of the Logic Lunch series! Next Thursday (April 22nd), Arianna Novaro (ILLC Amsterdam) will talk about Unravelling multi-agent ranked delegations, starting at 12:30. Save the date and join us on Zoom! Please find more information below, and follow us on Twitter and Instagram for more events and activities.
Title: Unravelling multi-agent ranked delegations
Abstract: In this talk, I will present a framework for collective decision-making where the agents have to vote on a given issue, but they can also choose to delegate their vote (if, for instance, they did not have the time or expertise to take a stance on the issue at stake). The agents can express complex delegations, i.e., they can specify a set of trusted delegates and a function–being it a classical voting rule or a propositional formula–to decide their vote, and they can also provide a ranking of preferred delegations. Given these complex delegation ballots, I will present four algorithms that unravel the ballots to get a profile of direct votes, on which the final decision can be taken by using some standard voting rule. In particular, I will discuss both the algorithmic properties and the computational complexity of the four algorithms, for different restrictions on the language of the delegation ballots. This is joint work with Rachael Colley and Umberto Grandi.
We are happy to announce that on April 8th, Mark Law (Imperial College London) will give the fourth talk of our Logic Lunch seminar series, starting at 12:30. Join us on Zoom at this link! And don’t forget to follow us on Twitter and Instagram to keep up to date with our news and events. Please find more information below:
Title: Logic-based Learning of Answer Set Programs
Abstract: In recent years, non-monotonic Inductive Logic Programming (ILP) has received growing interest. Specifically, several new learning frameworks and algorithms have been introduced for learning under the answer set semantics, allowing the learning of common-sense knowledge involving defaults and exceptions, which are essential aspects of human reasoning.
The first part of this seminar will present recent advances which have extended the theory of ILP and yielded a new collection of algorithms, called ILASP (Inductive Learning of Answer Set Programs), which are able to learn ASP programs consisting of normal rules, choice rules and both hard and weak constraints. Learning such programs allows ILASP to be applied in settings which had previously been outside the scope of ILP. In particular, weak constraints represent preference orderings, and so learning weak constraints allows ILASP to be used for preference learning.
The second part of the talk will present more recent work on a less general but much more scalable approach to learning ASP, called FastLAS. FastLAS is able to solve tasks with hypothesis spaces that are many orders of magnitude larger than those tolerated by ILASP, meaning that it can be applied to a greater range of real-world problems.