Search

Electrooculographical (EOG) artifacts are problematic to electroencephalographical (EEG) signal analysis and degrade performance of brain–computer interfaces. A novel, robust deep wavelet sparse autoencoder (DWSAE) method is presented and validated for fully automated EOG artifact removal. DWSAE takes advantage of wavelet transform and sparse autoencoder to become a universal EOG artifact corrector. After being trained without supervision, the sparse autoencoder performs EOG correction on time–frequency coefficients collected after brain wave signal wavelet decomposition. Corrected coefficients are then used for wavelet reconstruction of uncontaminated EEG signals. DWSAE is compared with five other methods: second-order blind identification, information maximization, joint approximation diagonalization of eigen-matrices, wavelet neural network (WNN) and wavelet thresholding (WT). Experimental results on a visual attention task dataset, a mental state recognition dataset and a semi-simulated contaminated EEG dataset show that DWSAE is capable of suppressing EOG artifacts effectively, while preserving the nature of background EEG signals. The mean square error of signals before and after correction by DWSAE on a semi-simulated contaminated EEG segment of 30 s is the lowest (65.62) when compared to the results produced by WNN and WT. DWSAE addresses limitations posed by these methods in three ways. First, DWSAE can be performed automatically and online in a single channel of EEG data; this has advantages over independent component analysis-based methods. Second, its results are robust and stable in comparison with those of other wavelet-based methods. Third, as an unsupervised learning scheme, DWSAE does not require the off-line training that is necessary for WNN and other supervised learning machine learning-based methods.

The problem of swarm navigation recently becomes a significant topic because of its suitability for various applications like search and rescue with autonomous systems. However, when it comes to a swarm of learning agents, one has to face the challenges from huge state spaces and the lack of scalability when the size of the swarm increases. Reinforcement learning (RL) approaches, which allow agents to interact not only with each other but also with their operational environment to obtain optimal policies, are considered as promising techniques for swarm navigation problems. Different RL algorithms have been used to solve these problems but most of them are limited to discrete state spaces and/or do not scale well with an increase of number of learning agents in the swarm. In this paper, we propose a Swarm Hierarchical Deep Deterministic Policy Gradients (SH-DDPGs) framework to address both drawbacks above in the context of leader-follower swarm navigation. By decomposing the navigation task of the swarm into two primitive sub-tasks: leader-following and collision avoidance, we can guarantee the convergence of the training processes of these sub-tasks in a continuous environment before combining output actions produced from those trained models to complete the entire task. Moreover, our method represents scalability as it is independent to the size of the swarm. Firstly, when training a follower, we only use information of its neighbors within its local view and the leader. Secondly, the trained model of one follower can be reapplied for the remaining followers. Training results show that the proposed SH-DDPGs algorithm is able to converge quickly and allow followers agent to learn an optimal policy for the whole group to navigate through the environment without colliding with each other and flexibly optimize their formation so that the distances among agents are minimized.

Recent years have been witnessing a huge step of artificial intelligence towards being applied in autonomous robots. To build intelligent robots navigating in indoor environment, many research focus on deep reinforcement learning which help robot learn and plan by themselves. Different network architectures are proposed for training agents to navigate and find targeted objects in both real and simulated environments. Despite promising results, one key challenge remaining is that the agent has to perform well in unseen environments and objects. To solve this generalization problem, this work proposes a method using prior knowledge graph capturing relationships between target objects. Experiments on simulated environments show that not only the proposed method enhances the learning process but also significantly improves agents generalization. When compared to similar methods, proposed method has a competitive and even better performance while bringing computational advantages.

Relay attacks are nowadays well known and most designers of secure authentication protocols are aware of them. At present, the main methods to prevent these attacks are based on the so-called distance bounding technique which consists in measuring the round-trip time of the exchanged authentication messages between the prover and the verifier to estimate an upper bound on the distance between these entities. Based on this bound, the verifier checks if the prover is sufficiently close by to rule out an unauthorized entity. Recently, a new work has proposed an authentication protocol that surprisingly uses the side-channel leakage to prevent relay attacks. In this paper, we exhibit some practical and security issues of this protocol and provide a new one that fixes all of them. Then, we argue the resistance of our proposal against both side-channel and relay attacks under some realistic assumptions. Our experimental results show the efficiency of our protocol in terms of false acceptance and false rejection rates.

In this work, we address a novel, but potentially emerging, problem of discriminating the natural human voices and those played back by any kind of audio devices in the context of interactions with in-house voice user interface. The tackled problem may find relevant applications in (1) the far-field voice interactions of vocal interfaces such as Amazon Echo, Google Home, Facebook Portal, etc, and (2) the replay spoofing attack detection. The detection of loudspeaker emitted speech will help avoid false wake-ups or unintended interactions with the devices in the first application, while eliminating attacks involve the replay of recordings collected from enrolled speakers in the second one. At first we collect a real-world dataset under well-controlled conditions containing two classes: recorded speeches directly spoken by numerous people (considered as the natural speech), and recorded speeches played back from various loudspeakers (considered as the loudspeaker emitted speech). Then from this dataset, we build prediction models based on Deep Neural Network (DNN) for which different combination of audio features have been considered. Experiment results confirm the feasibility of the task where the combination of audio embeddings extracted from SoundNet and VGGish network yields the classification accuracy up to about 90%., Comment: 5 pages, 1 figure

Multi-view video plus depth (MVD) has been widely used owing to its effectiveness in three-dimensional data representation. Using MVD, color videos...

Typical depth quality metrics require the ground truth depth image or stereoscopic color image pair, which are not always available in many practical applications. In this paper, we propose a new depth image quality metric which demands only a single pair of color and depth images. Our observations reveal that the depth distortion is strongly related to the local image characteristics, which in turn leads us to formulate a new distortion assessment method for the edge and non-edge pixels in the depth image. The local depth distortion is adaptively weighted using the Gabor filtered color image and added up to the global depth image quality metric. The experimental results show that the proposed metric closely approximates the depth quality metrics that use the ground truth depth or stereo color image pair.

In the context of video background&ndash, foreground separation, we propose a compressive online Robust Principal Component Analysis (RPCA) with optical flow that separates recursively a sequence of video frames into foreground (sparse) and background (low-rank) components. This separation method operates on a small set of measurements taken per frame, in contrast to conventional batch-based RPCA, which processes the full data. The proposed method also leverages multiple prior information by incorporating previously separated background and foreground frames in an n-ℓ1 minimization problem. Moreover, optical flow is utilized to estimate motions between the previous foreground frames and then compensate the motions to achieve higher quality prior foregrounds for improving the separation. Our method is tested on several video sequences in different scenarios for online background&ndash, foreground separation given compressive measurements. The visual and quantitative results show that the proposed method outperforms other existing methods.

Pavement crack detection is an important problem in road maintenance. There are many processing methods, including traditional and modern methods, solving this issue. Traditional methods use edge detection or some other digital image processing for crack detection, but these approaches are sensitive to many types of noise and unwanted objects on the road. For the purpose of increasing accuracy, image pre-processing methods are required for many of these techniques. Recently, some techniques that utilize deep learning to detect cracks in images have achieved high accuracy, without pre-processing. However, some of them are very complicated, some make use of manually collected data and some methods still need some form of pre-processing. In this paper, we propose a method that applies a convolutional neural networks to detect cracks in pavement images. Our research uses two data sets, one public data set and the other collected by ourselves. We also experimentally compare our method with some exiting methods and the experiments show that the proposed approach achieves high accuracy and generates stable models.

Motion detection in pixel domain of videos often requires highly capable resource to perform. This is due to computational demands for decoding the videos to obtain visual image. The problem can be solved by performing motion detection directly on bit streams of the compressed video data. This paper presents a new method for moving object detection in compressed domain of high resolution surveillance videos. In this work, we use video coding parameters (eg. motion vectors) extracted directly from bit stream of the compressed video to separate the moving blocks out of video background. We then propose new techniques for the segmentation and refinement of the foreground blocks to detect the moving objects. Experiments have been conducted on both public and self-recorded video data sets. Experimental results show the outperformance of our proposed method in comparing with conventional one.

In the context of online Robust Principle Component Analysis (RPCA) for the video foreground-background separation, we propose a compressive online RPCA with optical flow that separates recursively a sequence of frames into sparse (foreground) and low-rank (background) components. Our method considers a small set of measurements taken per data vector (frame), which is different from conventional batch RPCA, processing all the data directly. The proposed method also incorporates multiple prior information, namely previous foreground and background frames, to improve the separation and then updates the prior information for the next frame. Moreover, the foreground prior frames are improved by estimating motions between the previous foreground frames using optical flow and compensating the motions to achieve higher quality foreground prior. The proposed method is applied to online video foreground and background separation from compressive measurements. The visual and quantitative results show that our method outperforms the existing methods., Comment: preprint accepted

Fall detection is important for safety for old people or patients living alone. This paper proposes a framework for indoor fall detection using single camera system. Falls are detected based on the analysis of motion orientation, motion magnitude, and human shape changes. With a deep analysis of characteristics of fall events, we propose improvements for motion orientation estimation, large motion detection and human shape detection using motion histogram images (MHI). Fall detection is then determined by analyzing the speed of changing in motion magnitude, motion orientation and human shape before, during and after the fall. Experiments have been conducted on public datasets Li2e having 221 videos of different living environments with various daily activities. The experimental results show high detection accuracies and very fast processing capability.

International audience; Faults injection attacks have become a hot topic in the domain of smartcards. This work exposes a source code-base simulation approach designed to evaluate the robustness of high-level secured implementations against single and multiple fault injections. In addition to an unprotected CRT-RSA implementation, we successfully attacked two countermeasures with the high-level simulation under the data fault model. We define a filtering criterion that operates on found attacks and we refine our simulation analysis accordingly. We introduce a broader fault model that consists in skipping C lines of code and exhibit benefits of such high-level fault model in term of simulation performance and attack coverage.

International audience; Side channel attacks exploit physical information leaked during the operation of a cryptographic device (e.g., a smart card). The confidential data, which can be leaked from side channels, are timing of operations, power consumption, and electromagnetic emanation. In this paper, we propose a preprocessing method based on the fourth-order cumulant, which aims to improve the performance of side channel attacks. It takes advantages of the Gaussian and nonGaussian properties, that respectively characterize the noise and the signal, to remove the effects due to Gaussian noise coupled into side channel signals. The proposed method is then applied to analyze the electromagnetic signals of a synthesized application-specific integrated circuit during a data encryption standard operation. The theoretical and experimental results show that our method significantly reduces the number of side channel signals needed to detect the encryption key.

Cryptographic devices are vulnerable to the nowadays well known side channel leakage analysis. Secret data can be revealed by power analysis attacks such as Simple Power Analysis (SPA), Differential Power Analysis (DPA) and Correlation Power Analysis (CPA). First, we give an overview of DPA in mono-bit and multi-bit cases. Next, the existing multi-bit DPA methods are generalized into the proposed Partitioning Power Analysis (PPA) method. Finally, we focus on the CPA technique, showing that this attack is a case of PPA with special coefficients and a normalization factor. We also propose a method that allows us to improve the performance of CPA by restricting the normalization factor.