3IA PhD/Postdoc Seminar #5

Published on June 24, 2021 Updated on October 24, 2022

on the July 2, 2021

from 10:30am to 12:00pm




10:30 - 11:00
Ayse Unsal (Eurecom)

A Statistical Threshold for Adversarial Classification in Laplace Mechanisms
Abstract: This paper studies the statistical characterization of detecting an adversary who wants to harm some computation such as machine learning models or aggregation by altering the output of a differentially private mechanism in addition to discovering some information about the underlying dataset. An adversary who is able to modify the published information from a differentially private mechanism aims to maximize the possible damage to the system while remaining undetected. We present a trade-off between the privacy parameter of the system, the sensitivity and the attacker’s advantage (the bias) through determining the threshold for the best critical region of the hypothesis testing problem for deciding whether or not the adversary’s attack is detected. Such trade-offs are provided for Laplace mechanisms using one-sided and two-sided hypothesis tests. Corresponding error probabilities are analytically derived and ROC curves are presented for various levels of the sensitivity, the absolute mean of the attack and the privacy parameter. Subsequently, we provide an interval for the bias induced by the adversary so that the defender detects the attack. Finally, we adapt the Kullback-Leibler differential privacy to adversarial classification.

11:00 - 11:30
Ashwin James (CNRS)

Inference of choice granularity via learning model selection in a cognitive task 
Abstract: In cognitive neuroscience, it is common to study a single cognitive decision process by modeling the learning of a single behavior (e.g., turning left when seeing a red light). However, more complex tasks sometimes require a sequence of decision processes, not just a single one. Learning the task, then requires assigning credits of rewards to the actions that led to the reward. Here we try to analyze the learning process of a rat based on a free-moving T-maze experiment with rats, where individual rats may have different models of the space of the T-maze. Consequently, the action choices of each rat in the T-maze may be different. Identifying the individual maze models of the rats provides a better understanding of the whole learning process occurring in brains based on a better emulation of the corresponding cognitive processes.

11:30 - 12:00

Open discussion on the two contributions