Search

Vulnerability to adversarial attacks is one of the principal hurdles to the adoption of deep learning in safety-critical applications. Despite significant efforts, both practical and theoretical, the problem remains open. In this paper, we analyse the geometry of adversarial attacks in the large-data, overparametrized limit for Bayesian Neural Networks (BNNs). We show that, in the limit, vulnerability to gradient-based attacks arises as a result of degeneracy in the data distribution, i.e., when the data lies on a lower-dimensional submanifold of the ambient space. As a direct consequence, we demonstrate that in the limit BNN posteriors are robust to gradient-based adversarial attacks. Experimental results on the MNIST and Fashion MNIST datasets, representing the finite data regime, with BNNs trained with Hamiltonian Monte Carlo and Variational Inference support this line of argument, showing that BNNs can display both high accuracy and robustness to gradient based adversarial attacks., This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 722022

We introduce a novel approach to automatically detect ineffective breathing efforts in patients in intensive care subject to assisted ventilation. The method is based on synthesising from data temporal logic formulae which are able to discriminate between normal and ineffective breaths. The learning procedure consists in first constructing statistical models of normal and abnormal breath signals, and then in looking for an optimally discriminating formula. The space of formula structures, and the space of parameters of each formula, are searched with an evolutionary algorithm and with a Bayesian optimisation scheme, respectively. We present here our preliminary results and we discuss our future research directions.\&nbsp

Neural state classification (NSC) is a recently proposed method for runtime predictive monitoring of hybrid automata (HA) using deep neural networks (DNNs). NSC trains a DNN as an approximate reachability predictor that labels an HA state x as positive if an unsafe state is reachable from x within a given time bound, and labels x as negative otherwise. NSC predictors have very high accuracy, yet are prone to prediction errors that can negatively impact reliability. To overcome this limitation, we present neural predictive monitoring (NPM), a technique that complements NSC predictions with estimates of the predictive uncertainty. These measures yield principled criteria for the rejection of predictions likely to be incorrect, without knowing the true reachability values. We also present an active learning method that significantly reduces the NSC predictor’s error rate and the percentage of rejected predictions. We develop two versions of NPM based, respectively, on the use of frequentist and Bayesian techniques to learn the predictor and the rejection rule. Both versions are highly efficient, with computation times on the order of milliseconds, and effective, managing in our experimental evaluation to successfully reject almost all incorrect predictions. In our experiments on a benchmark suite of six hybrid systems, we found that the frequentist approach consistently outperforms the Bayesian one. We also observed that the Bayesian approach is less practical, requiring a careful and problem-specific choice of hyperparameters.

We study the applicability of deep learning (DL) methods to generate acoustic synthetic data from 1D models of the subsurface. We designed and implemented a Neural Network (NN) and we trained it to generate synthetic seismograms (common shot gathers) from 1-D velocity models on two different datasets: one obtained from published results and the other generated by Finite Differences (FD) numerical simulation. We furthermore compared the results from the proposed model with the published one. Moreover, we tried to to add more flexibility to this methodology by allowing change of wavelet and the acquisition geometry. We obtained good results in terms of both computation efficiency and quality of prediction. The main potentialities of the work are the low computational cost, a high prediction speed and the possibility to solve complex non-linear problems without knowing the physical law behind the phenomenon, which could led great advantages in the solution also of the inverse problem . DL to generate 1-D acoustic synthetic seismograms without solving wave equation Solution to the 1-D problem through custom Recurrent Neural Network Retraining strategy to improve flexibility and applicability Computational complexity analysis.