Robustness Verification for Classifier Ensembles

We give a formal verification procedure that decides whether a classifier ensemble is robust against arbitrary randomized attacks. Such attacks consist of a set of deterministic attacks and a distribution over this set. The robustness-checking problem consists of assessing, given a set of classifiers and a labelled data set, whether there exists...

Graphs with given diameter maximizing the algebraic connectivity

We propose a class of graphs GD(n₁,n₂,⋯,n _D+1), containing of a chain of D+1 cliques K n₁,K n ₂,⋯,Kn_D+1, where neighboring cliques are fully-interconnected. The class of graphs has diameter D and size N=∑_1≤i≤D+1n_i. We prove that this class of graphs can...

Synthesis of robust feedback missile control strategies by using LMI techniques

Robust programming, and more specific LMI techniques, have proven to be an extremely effective tool for the synthesis of missile guidance laws. The advantage of this model-based approach is that a large class of uncertainties can be taken into account and a robust guidance law is obtained. In this paper the class of uncertainties applicable for...

Virus spread in networks

The influence of the network characteristics on the virus spread is analyzed in a new-the N-intertwined Markov chain-model, whose only approximation lies in the application of mean field theory. The mean field approximation is quantified in detail. The N-intertwined model has been compared with the exact 2N-state Markov model and with previously...

M/M/∞ transience: Tail asymptotics of congestion periods

The c-congestion period, defined as a time interval in which the number of customers is larger than c all the time, is a key quantity in the design of communication networks. Particularly in the setting of M/M/∞ systems, the analysis of the duration of the congestion period, D_c, has attracted substantial attention; related quantities...

Rate adaptation using acknowledgement feedback: Throughput upper bounds

We consider packet-by-packet rate adaptation to maximize the throughput over a finite-state Markov channel. To limit the amount of feedback data, we use past packet acknowledgements (ACKs) and past rates as channel state information. It is known that the maximum achievable throughput is computationally prohibitive to determine. Thus, in this...