The Global Navigation Satellite System (GNSS) plays a crucial role in critical infrastructure by delivering precise timing and positional data. Nonetheless, the civilian segment of the GNSS remains susceptible to various spoofing attacks, necessitating robust detection mechanisms. The ability to deter such attacks significantly enhances the reliability and security of systems utilizing GNSS technology. Supervised Machine Learning (ML) techniques have shown promise in spoof detection. However, their effectiveness hinges on training data encompassing all possible attack scenarios, rendering them vulnerable to novel attack vectors. To address this limitation, we explore representation learning-based methods. These methods can be trained with a single data class and subsequently applied to classify test samples as either belonging to the training class or not. In this context, we introduce a GNSS spoof detection model comprising a Variational AutoEncoder (VAE) and a Generative Adversarial Network (GAN). The composite model is designed to efficiently learn the class distribution of the training data. The features used for training are extracted from the radio frequency and tracking modules of a standard GNSS receiver. To train our model, we leverage the Texas Spoofing Test Battery (TEXBAT) datasets. Our trained model yields three distinct detectors capable of effectively identifying spoofed signals. The detection performance across simpler to intermediate datasets for these detectors reaches approximately 99%, demonstrating their robustness. In the case of subtle attack scenario represented by DS-7, our approach achieves an approximate detection rate of 95%. In contrast, under supervised learning, the best detection score for DS-7 remains limited to 44.1%.
28 April 2017 in IEEE Transactions on Medical Imaging
Sequential dictionary learning algorithms have been successfully applied to functional magnetic resonance imaging (fMRI) data analysis. fMRI data sets are, however, structured data matrices with the notions of temporal smoothness in the column direction. This prior information, which can be converted into a constraint of smoothness on the learned dictionary atoms, has seldomly been included in classical dictionary learning algorithms when applied to fMRI data analysis. In this paper, we tackle this problem by proposing two new sequential dictionary learning algorithms dedicated to fMRI data analysis by accounting for this prior information. These algorithms differ from the existing ones in their dictionary update stage. The steps of this stage are derived as a variant of the power method for computing the SVD. The proposed algorithms generate regularized dictionary atoms via the solution of a left regularized rank-one matrix approximation problem where temporal smoothness is enforced via regularization through basis expansion and sparse basis expansion in the dictionary update stage. Applications on synthetic data experiments and real fMRI data sets illustrating the performance of the proposed algorithms are provided.
In this paper, a robust sequential dictionary learning (DL) algorithm is presented. The proposed algorithm is motivated from the maximum likelihood perspective on dictionary learning and its link to the minimization of the Kullback-Leibler divergence. It is obtained by using a robust loss function in the data fidelity term of the DL objective instead of the usual quadratic loss. The proposed robust loss function is derived from the α-divergence as an alternative to the Kullback-Leibler divergence, which leads to a quadratic loss. Compared to other robust approaches, the proposed loss has the advantage of belonging to class of redescending M-estimators, guaranteeing inference stability from large deviations from the Gaussian nominal noise model. The algorithm is obtained by solving a sequence of penalized rank-1 matrix approximation problems, where the ℓ 1 -norm is introduced as a penalty promoting sparsity and then using a block coordinate descent approach to estimate the unknowns. Performance comparison with similar robust DL algorithms on digit recognition, background removal, and gray-scale image denoising is performed highlighting efficacy of the proposed algorithm.
The application of machine learning in healthcare, financial, social media, and other sensitive sectors not only involves high accuracy but privacy as well. Due to the emergence of the Cloud as a computation and one-to-many access paradigm; training and classification/inference tasks have been outsourced to Cloud. However, its usage is limited due to legal and ethical constraints regarding privacy. In this work, we propose a privacy-preserving neural networks-based classification model based on Homomorphic Encryption (HE) where the user can send an encrypted instance to the cloud and receive an encrypted inference from it to preserve the user’s query privacy. In contrast to existing works, we demonstrate the realistic limitations of HE for privacy-preserving machine learning by changing its parameters for enhanced security and accuracy. We showcase scenarios where the choice of HE parameters impedes accurate classification and present an optimized setting for achieving reliable classification. We present several results to demonstrate its effectiveness using MNIST dataset with highly improved inference time for a query as compared to the state of the art.
22 December 2023 in IEEE Transactions on Consumer Electronics
The Global Navigation Satellite System (GNSS) plays critical role in providing precise timing and positioning information for various applications. However, its civilian signals are vulnerable to spoofing attacks, necessitating robust detection methods. While supervised learning has shown promise in spoof detection, it requires distinct features in the training data for effective learning. In this study, we propose a novel training feature set that combines power and Signal Quality Monitoring (SQM) metrics from a single-antenna GNSS receiver. We introduce a Two-Stage Artificial Neural Network (TS-ANN) that leverages this feature set alongside multi-correlator finger values for effective spoof detection. Furthermore, supervised learning models often struggle with previously unseen attacks due to limited overlap with the training data. To address this, we present a zero-day attack detector based on unsupervised representation learning using a Variational Autoencoder (VAE) which is trained exclusively on authentic datasets, enhancing its ability to detect novel attack patterns. Experimental results on publicly available TEXBAT datasets demonstrate that our TS-ANN achieves a detection probability (PD) exceeding 99% for similar test datasets. In more sophisticated attack scenarios, such as DS-7, performance may decrease to 50.68%. Nevertheless, our zero-day detector maintains a PD exceeding 92.5%, highlighting its resilience to previously unseen attacks.