Notice
Information Structures for Privacy and Fairness
- document 1 document 2 document 3
- niveau 1 niveau 2 niveau 3
Descriptif
The increasingly pervasive use of big data and machine learning is raising various ethical issues, in particular privacy and fairness.
In this talk, I will discuss some frameworks to understand and mitigate the issues, focusing on iterative methods coming from information theory and statistics.
In the area of privacy protection, differential privacy (DP) and its variants are the most successful approaches to date. One of the fundamental issues of DP is how to reconcile the loss of information that it implies with the need to preserve the utility of the data. In this regard, a useful tool to recover utility is the Iterative Bayesian Update (IBU), an instance of the famous Expectation-Maximization method from Statistics. I will show that the IBU, combined with the metric version of DP, outperforms the state-of-the art, which is based on algebraic methods combined with the Randomized Response mechanism, widely adopted by the Big Tech industry (Google, Apple, Amazon, …). Furthermore I will discuss a surprising duality between the IBU and one of the methods used to enhance metric DP, that is the Blahut-Arimoto algorithm from Rate-Distortion Theory. Finally, I will discuss the issue of biased decisions in machine learning, and will show that the IBU can be applied also in this domain to ensure a fairer treatment of disadvantaged groups.
Sur le même thème
-
Des systèmes de numération pour le calcul modulaire
BajardJean-ClaudeLe calcul modulaire est utilisé dans de nombreuses applications des mathématiques, telles que la cryptographie. La réduction modulaire dans un contexte très général est coûteuse, car elle n
-
Soirée de présentation de l'ouvrage "Le parfait fasciste"
De GraziaVictoriaLebrunJeanPrésentation de l'ouvrage "Le parfait fasciste", de Victoria de Grazia, présenté par Jean Lebrun
-
Adoprivacy - Aggravation des vulnérabilités, inégalités sociales, genre par les activités numériques
CaseauAnne-CécileJehelSophieBoubéeNicoleDymytrovaValentynaArfaouiMehdiDéageMargotLes enjeux de la protection de la vie privée sur les plateformes numériques sont complexes, ils supposeraient des connaissances juridiques et des habiletés sociales qui sont inégalement partagées. Les
-
2024 Memory, Light, Spin Glasses, and Deep Classification
LeonettiMarcoMarco Leonetti livre une présentation de haut niveau, à la fois riche et inspirante à l'interface de l'apprentissage machine, de la physique statistique et de l'optique ondulatoire en milieu diffusant
-
Inauguration de l'exposition - Vanessa Vitse : Nombres de Sophie Germain et codes secrets
VitseVanessaExposé de Vanessa Vitse (Institut Fourier) : Nombres de Sophie Germain et codes secrets
-
Integrating Simulation, Machine Learning, and High-performance Computing to Support Public Health D…
OzikJonathanIntegrating Simulation, Machine Learning, and High-performance Computing to Support Public Health Decision Making
-
Research in Knowledge-Enhanced Machine Learning at the Center for Artificial Intelligence (C4AI)
CozmanFabio GagliardiResearch in Knowledge-Enhanced Machine Learning at the Center for Artificial Intelligence (C4AI)
-
Quelles villes intelligentes pour demain ?
RivanoHervéJongwaneJoannaQuelles villes intelligentes pour demain ?
-
AI and Human Decision-Making: An Interdisciplinary Perspective
CastellucciaClaudeThis seminar will talk about some of the privacy risks of these systems and will describe some recent attacks. It will also discuss why they sometimes fail to deliver. Finally, we will also show that
-
5.1. Code-Based Digital Signatures
Marquez-CorbellaIreneSendrierNicolasFiniaszMatthieuWelcome to the last week of this MOOC on code-based cryptography. This week, we will be discussing other cryptographic constructions relying on coding theory. We have seen how to do public key
-
4.5. Error-Correcting Pairs
Marquez-CorbellaIreneSendrierNicolasFiniaszMatthieuWe present in this session a general decoding method for linear codes. And we will see it in an example. Let C be a generalized Reed-Solomon code of dimension k associated to the pair (c, d). Then,
-
5.4. Parallel-CFS
Marquez-CorbellaIreneSendrierNicolasFiniaszMatthieuIn this session, I will present a variant of the CFS signature scheme called parallel-CFS. We start from a simple question: what happens if you try to use two different hash functions and compute
