Emotion recognition plays an important role in the field of human–computer interaction (HCI). An electroencephalogram (EEG) is widely used to estimate human emotion owing to its convenience and mobility. Deep neural network (DNN) approaches using an EEG for emotion recognition have recently shown remarkable improvement in terms of their recognition accuracy. However, most studies in this field still require a separate process for extracting handcrafted features despite the ability of a DNN to extract meaningful features by itself. In this paper, we propose a novel method for recognizing an emotion based on the use of three-dimensional convolutional neural networks (3D CNNs), with an efficient representation of the spatio-temporal representations of EEG signals. First, we spatially reconstruct raw EEG signals represented as stacks of one-dimensional (1D) time series data to two-dimensional (2D) EEG frames according to the original electrode position. We then represent a 3D EEG stream by concatenating the 2D EEG frames to the time axis. These 3D reconstructions of the raw EEG signals can be efficiently combined with 3D CNNs, which have shown a remarkable feature representation from spatio-temporal data. Herein, we demonstrate the accuracy of the emotional classification of the proposed method through extensive experiments on the DEAP (a Dataset for Emotion Analysis using EEG, Physiological, and video signals) dataset. Experimental results show that the proposed method achieves a classification accuracy of 99.11%, 99.74%, and 99.73% in the binary classification of valence and arousal, and, in four-class classification, respectively. We investigate the spatio-temporal effectiveness of the proposed method by comparing it to several types of input methods with 2D/3D CNN. We then verify the best performing shape of both the kernel and input data experimentally. We verify that an efficient representation of an EEG and a network that fully takes advantage of the data characteristics can outperform methods that apply handcrafted features.

The detection and segmentation of thrombi are essential for monitoring the disease progression of abdominal aortic aneurysms (AAAs) and for patient care and management. As they have inherent capabilities to learn complex features, deep convolutional neural networks (CNNs) have been recently introduced to improve thrombus detection and segmentation. However, investigations into the use of CNN methods is in the early stages and most of the existing methods are heavily concerned with the segmentation of thrombi, which only works after they have been detected. In this work, we propose a fully automated method for the whole process of the detection and segmentation of thrombi, which is based on a well-established mask region-based convolutional neural network (Mask R-CNN) framework that we improve with optimized loss functions. The combined use of complete intersection over union (CIoU) and smooth L1 loss was designed for accurate thrombus detection and then thrombus segmentation was improved with a modified focal loss. We evaluated our method against 60 clinically approved patient studies (i.e., computed tomography angiography (CTA) image volume data) by conducting 4-fold cross-validation. The results of comparisons to multiple other state-of-the-art methods suggested the superior performance of our method, which achieved the highest F1 score for thrombus detection (0.9197) and outperformed most metrics for thrombus segmentation.

In this paper, we utilize the advantages of an event camera to tackle the visual place recognition (VPR) problem. The event camera’s high measurement rate, low latency, and high dynamic range make it well-suited to overcome the limitations of conventional vision sensors. However, to apply the existing convolutional neural networks (CNNs) based algorithms such as NetVLAD, the asynchronous event stream should be converted to a synchronous image frame, which causes a loss in temporal information. To address this problem, this paper proposes a method that employs the asynchronous characteristic of spiking neural networks (SNNs) to leverage the temporal nature of event streams. The event stream is converted to event images and tensors in our preprocessing module. The SNNs-based reconstruction networks, which are converted from CNNs, reconstruct edge images from event tensors regardless of external environment changes. Visual place recognition is conducted by matching features of the database and those from NetVLAD, which we used as a feature extraction network in this study. To evaluate the performance of VPR by comparing the previous methods for DDD17 and the Brisbane-Event-VPR dataset. Experimental results demonstrate that the matching accuracy of the proposed method is better than previous methods, especially for datasets with adverse weather conditions. We also verify that the performance and energy efficiency are improved with SNNs over CNNs. Our code is available for download on https://github.com/AIRLABkhu/EvReconNet.

Recent advances in autostereoscopic three-dimensional (3D) display systems have led to innovations in consumer electronics and vehicle systems (e.g., head-up displays). However, medical images with stereoscopic depth provided by 3D displays have yet to be developed sufficiently for widespread adoption in diagnostics. Indeed, many stereoscopic 3D displays necessitate special 3D glasses that are unsuitable for clinical environments. This paper proposes a novel glasses-free 3D autostereoscopic display system based on an eye tracking algorithm and explores its viability as a 3D navigator for cardiac computed tomography (CT) images. The proposed method uses a slit-barrier with a backlight unit, which is combined with an eye tracking method that exploits multiple machine learning techniques to display 3D images. To obtain high-quality 3D images with minimal crosstalk, the light field 3D directional subpixel rendering method combined with the eye tracking module is applied using a user’s 3D eye positions. Three-dimensional coronary CT angiography images were volume rendered to investigate the performance of the autostereoscopic 3D display systems. The proposed system was trialed by expert readers, who identified key artery structures faster than with a conventional two-dimensional display without reporting any discomfort or 3D fatigue. With the proposed autostereoscopic 3D display systems, the 3D medical image navigator system has the potential to facilitate faster diagnoses with improved accuracy.

Deep neural networks (DNNs) have successfully classified EEG-based brain-computer interface (BCI) systems. However, recent studies have found that well-designed input samples, known as adversarial examples, can easily fool well-performed deep neural networks model with minor perturbations undetectable by a human. This paper proposes an efficient generative model named generative perturbation network (GPN), which can generate universal adversarial examples with the same architecture for non-targeted and targeted attacks. Furthermore, the proposed model can be efficiently extended to conditionally or simultaneously generate perturbations for various targets and victim models. Our experimental evaluation demonstrates that perturbations generated by the proposed model outperform previous approaches for crafting signal-agnostic perturbations. We demonstrate that the extended network for signal-specific methods also significantly reduces generation time while performing similarly. The transferability across classification networks of the proposed method is superior to the other methods, which shows our perturbations' high level of generality. Our code is available for download on https://github.com/AIRLABkhu/Generative-Perturbation-Networks .