ILLC-LIRa September 2024

illc-lira@list.uva.nl

1 participants
4 discussions

LIRa session: Giuliano Rosella (Thursday, 3 October 16:30)
by LIRa Seminar 28 Sep '24

28 Sep '24

Dear all, We will have our next LIRa session on Thursday, 3 October 16:30. To attend online, please use our recurring zoom link: https://uva-live.zoom.us/j/89230639823?pwd=YWJuSnJmTDhXcWhmd1ZkeG5zb0o5UT09 (Meeting ID: 892 3063 9823, Passcode: 421723) You can find the details of the talk below. Speaker: Giuliano Rosella (University of Turin) Date and Time: Thursday, October 3rd 2024, 16:30-18:00 Venue: online Title: The algebra of Lewis’s counterfactuals and their duality theory Abstract. The logico-algebraic study of Lewis’s hierarchy of variably strict conditional logics has been essentially unexplored, hindering our understanding of their mathematical foundations, and the connections with other logical systems. This work aims to fill this gap by providing a comprehensive logico-algebraic analysis of Lewis’s logics. We begin by introducing novel finite axiomatizations for varying strengths of Lewis’s logics, distinguishing between global and local consequence relations on Lewisian sphere models. We then demonstrate that the global consequence relation is strongly algebraizable in terms of a specific class of Boolean algebras with a binary operator representing the counterfactual implication. In contrast, we show that the local consequence relation is generally not algebraizable, although it can be characterized as the degree-preserving logic over the same algebraic models. Further, we delve into the algebraic semantics of Lewis’s logics, developing two dual equivalences with respect to particular topological spaces. In more details, we show a duality with respect to the topological version of Lewis’s sphere models, and also with respect to Stone spaces with a selection function; using the latter, we demonstrate the strong completeness of Lewis’s logics with respect to sphere models. Finally, we draw some considerations concerning the limit assumption over sphere models. Hope to see you there! The LIRa team

1 0

LIRa session: Rebecca Reiffenhäuser (tomorrow, Thursday, 19 September 16:30)
by LIRa Seminar 18 Sep '24

18 Sep '24

Dear all, We will have our next LIRa session tomorrow, on Thursday, 19 September 16:30. This talk will be hybrid, with the speaker on location. To attend online, please use our recurring zoom link: https://uva-live.zoom.us/j/89230639823?pwd=YWJuSnJmTDhXcWhmd1ZkeG5zb0o5UT09 (Meeting ID: 892 3063 9823, Passcode: 421723) You can find the details of the talk below. Speaker: Rebecca Reiffenhäuser (ILLC, University of Amsterdam) Date and Time: Thursday, September 19th 2024, 16:30-18:00 Venue: ILLC Seminar Room F1.15, Science Park 107. Title: How to sell your items online - with almost no prior info. Abstract. The problem of allocating a set of items to a set of buyers in a way that maximizes an objective, like the overall valuation of participants for the items they receive (social welfare), is a central setting with numerous applications. We consider the harder, but very prevalent case where buyers/bids arrive over time in some fixed order, and decisions have to be made immediately (online) on their arrival. It is impossible to get close to the optimal value that would be achievable for a prophet who can see future participants valuations beforehand, and giving all items to some random person is actually the best you can do. Therefore, we additionally assume to get a tiny piece of information on each of the future participants beforehand: For every buyer to arrive, we assume to have a single sample of what his valuation function might be, e.g. from when he participated in an earlier auction. We show that this suffices to achieve an expected, constant-factor approximation to the value of the future-seeing prophet for buyers with XOS valuation functions, and also consider the case that buyers might misreport their values strategically – which so far we can only deal with if we have some more samples. (Based on joint work with Paul Duetting, Thomas Kesselheim, Brendan Lucier and Sahil Singla, to appear in FOCS 2024.) Hope to see you there! The LIRa team

1 0

LIRa session: Rebecca Reiffenhäuser (Thursday, 19 September 16:30)
by LIRa Seminar 14 Sep '24

14 Sep '24

Dear all, We will have our next LIRa session on Thursday, 19 September 16:30. This talk will be hybrid, with the speaker on location. To attend online, please use our recurring zoom link: https://uva-live.zoom.us/j/89230639823?pwd=YWJuSnJmTDhXcWhmd1ZkeG5zb0o5UT09 (Meeting ID: 892 3063 9823, Passcode: 421723) You can find the details of the talk below. Speaker: Rebecca Reiffenhäuser (ILLC, University of Amsterdam) Date and Time: Thursday, September 19th 2024, 16:30-18:00 Venue: ILLC Seminar Room F1.15, Science Park 107. Title: How to sell your items online - with almost no prior info. Abstract. The problem of allocating a set of items to a set of buyers in a way that maximizes an objective, like the overall valuation of participants for the items they receive (social welfare), is a central setting with numerous applications. We consider the harder, but very prevalent case where buyers/bids arrive over time in some fixed order, and decisions have to be made immediately (online) on their arrival. It is impossible to get close to the optimal value that would be achievable for a prophet who can see future participants valuations beforehand, and giving all items to some random person is actually the best you can do. Therefore, we additionally assume to get a tiny piece of information on each of the future participants beforehand: For every buyer to arrive, we assume to have a single sample of what his valuation function might be, e.g. from when he participated in an earlier auction. We show that this suffices to achieve an expected, constant-factor approximation to the value of the future-seeing prophet for buyers with XOS valuation functions, and also consider the case that buyers might misreport their values strategically – which so far we can only deal with if we have some more samples. (Based on joint work with Paul Duetting, Thomas Kesselheim, Brendan Lucier and Sahil Singla, to appear in FOCS 2024.) Hope to see you there! The LIRa team

1 0

LIRa session: Caleb Schultz Kisby (tomorrow, Thursday, 5 September 16:30)
by LIRa Seminar 04 Sep '24

04 Sep '24

Dear all, We will have our next LIRa session tomorrow, on Thursday, 5 September 16:30. To attend online, please use our recurring zoom link: https://uva-live.zoom.us/j/89230639823?pwd=YWJuSnJmTDhXcWhmd1ZkeG5zb0o5UT09 (Meeting ID: 892 3063 9823, Passcode: 421723) You can find the details of the talk below. Speaker: Caleb Schultz Kisby (Indiana University, Bloomington) Date and Time: Thursday, September 5th 2024, 16:30-18:00 Venue: online Title: The Modeling Power of Neural Networks Abstract. Neural networks are very good at learning without human guidance, yet they're also known for making blunders that seem silly from the point of view of logic. (And this situation hasn't changed, despite modern neural network systems like GPT-4). This is a long-standing problem in artificial intelligence: How can we better understand and control neural networks using logic? In response, there have been countless proposals for "neuro-symbolic" systems that incorporate logic into neural networks, or vice versa. In this talk I will present one such proposal that is close to the hearts of modal and epistemic logicians: Treat (binary) neural networks as a class of models in modal logic by (1) adding a valuation of propositions (as sets of neurons), and (2) interpreting ◇φ as the forward propagation (or diffusion) of input φ through the net. We can then do "business as usual," using neural networks as our models. To cement this idea, I will compare the modeling power of neural networks with other classes of models, in particular: relational, plausibility, neighborhood, and social network models. If time permits, I will mention recent work in which we "dynamify" this logic, in the spirit of modeling neural network update and learning. This talk is based on joint work (in progress) with Saúl Blanco and Larry Moss. Our work on the dynamics of neural network update appears in AAAI 2024 and FLAIRS 2022. Hope to see you there! The LIRa team

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

ILLC-LIRa September 2024