Concept

Ajustement de loi de probabilité

Publications associées (31)

Collaborative Learning in the Jungle (Decentralized, Byzantine, Heterogeneous, Asynchronous and Nonconvex Learning)

Rachid Guerraoui, El Mahdi El Mhamdi, Sébastien Louis Alexandre Rouault, Le Nguyen Hoang, Arsany Hany Abdelmessih Guirguis, Sadegh Farhadkhani

We study Byzantine collaborative learning, where n nodes seek to collectively learn from each others' local data. The data distribution may vary from one node to another. No node is trusted, and f < n nodes can behave arbitrarily. We prove that collaborati ...

2021

Chaining Meets Chain Rule: Multilevel Entropic Regularization and Training of Neural Networks

Emmanuel Abbé

We derive generalization and excess risk bounds for neural networks using a family of complexity measures based on a multilevel relative entropy. The bounds are obtained by introducing the notion of generated hierarchical coverings of neural networks and b ...

Microtome Publishing2020

Fast ion transport phenomena in turbulent toroidal plasmas

Fabian Manke

In the field of plasma physics, suprathermal ions are encountered e.g. in astrophysical jets, the solar wind, as well as fusion plasmas, where they originate from neutral beam injection or fusion reactions. One aspect of shared interest with astrophysical ...

EPFL2020

Distributional Robustness in Mechanism Design

Cagil Kocyigit

Mechanism design theory examines the design of allocation mechanisms or incentive systems involving multiple rational but self-interested agents and plays a central role in many societally important problems in economics. In mechanism design problems, agen ...

EPFL2020

An Architecture for Load Balance in Computer Cluster Applications

Laurent Bindschaedler

Amid a data revolution that is transforming industries around the globe, computing systems have undergone a paradigm shift where many applications are scaled out to run on multiple computers in a computing cluster. As the storage and processing capabilitie ...

EPFL2020

Collaborative Sampling in Generative Adversarial Networks

Alexandre Massoud Alahi, Yuejiang Liu, Parth Ashit Kothari

The standard practice in Generative Adversarial Networks (GANs) discards the discriminator during sampling. However, this sampling method loses valuable information learned by the discriminator regarding the data distribution. In this work, we propose a co ...

2020

Bayesian Differential Privacy for Machine Learning

Boi Faltings, Aleksei Triastcyn

%0 Conference Paper %T Bayesian Differential Privacy for Machine Learning %A Aleksei Triastcyn %A Boi Faltings %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E A ...

2020

Data-Aware Privacy-Preserving Machine Learning

Aleksei Triastcyn

In this thesis, we focus on the problem of achieving practical privacy guarantees in machine learning (ML), where the classic differential privacy (DP) fails to maintain a good trade-off between user privacy and data utility. Differential privacy guarantee ...

EPFL2020

Design guidelines for Al-12%Si latent heat storage encapsulations to optimize performance and mitigate degradation

Sophia Haussener, Selmar Rudolf Binder

We provide detailed insight into the growth kinetics of intermetallic layers formed at the interface of molten Al-12%Si with cylindrical encapsulations of 316L stainless steel. This knowledge is relevant for the design and operation of high-temperature lat ...

ELSEVIER2020

Local Vulnerabilities and Global Robustness of Coupled Dynamical Systems on Complex Networks

Melvyn Sandy Tyloo

Coupled dynamical systems are omnipresent in everyday life. In general, interactions between individual elements composing the system are captured by complex networks. The latter greatly impact the way coupled systems are functioning and evolving in time. ...

EPFL2020

On Vacuous and Non-Vacuous Generalization Bounds for Deep Neural Networks

Konstantinos Pitas

Deep neural networks have been empirically successful in a variety of tasks, however their theoretical understanding is still poor. In particular, modern deep neural networks have many more parameters than training data. Thus, in principle they should over ...

EPFL2020