Articles by Valentin Dardilhac

Complexity of neural network training and complexity proofs bypassing frontier issues

-- Valentin Dardilhac (LISN, Galac)

summary: We study the complexity of the neural network training decision
problem in two different contexts. First, in the general context, this
problem has been shown to be in extensions of the class ∃R. We have
been able to show that whenever the activation functions are Lipschitz
functions and the ...

Classification of truth revealing social choice algorithms

-- Valentin Dardilhac (LISN, Galac)

summary: The talk will be on the field of social choices. A group of players
want to choose a subset of a set of objects respecting some properties
(maximal weight of the subset, maximal amount of objects in the subset, ...). To do so, they vote and use a social choice ...