RRAM-based Deep Neural Network (DNN) accelerators have gained widespread usage in edge devices. However, the security vulnerabilities of RRAM-based accelerators hinder their real application. Current research on safeguarding RRAM-based accelerators predominantly relies on a single-level protection approach. This has resulted in the restriction of its protection scope, the rigidity and lack of generality in the protection method, or has had an impact on the computational efficiency of the system. As a result, it encounters substantial challenges in attaining comprehensive optimization across multiple dimensions, such as universality, the scope of protection, and security-related overheads. In this paper, we develop specific analyses on accelerators and attacks and build graph-based representations. Based on these, we partition the RRAM-based accelerators’ security into two levels: on-chip security and off-chip security. Furthermore, we propose a two-level protection framework for RRAM-based accelerators, which is calledOmniGuard. At the on-chip security level,OmniGuardproposes a bit-grained shuffle to achieve protection while using lightweight Benes Networks to maintain the CIM capability. At the off-chip level,OmniGuardproposes an RRAM-based AES engine to introduce the AES algorithm into the accelerator with significant acceleration and minimal overhead. Evaluation results demonstrate thatOmniGuardprovides powerful and flexible protection while achieving 1.33×∼4.38× speedup and 1.26×∼2.65× power savings, with only 5% energy overhead and 3% area overhead.

Due to the complex geometric features of facial expressions, achieving realistic facial expression transfer is a challenging task. This paper proposes a two-stage facial expression transfer method based on 3D Morphable Model (3DMM) and diffusion models. In the first stage, 3DMM-based DECA model is employed to map the source identity face and the target expression face 2D image into 3D facial parameters. Subsequently, the expression parameters of the source identity face are substituted with the expression parameters obtained from the target expression face. This procedure reconstructs a 3D flame model of the face with the transferred expression. Finally, the 3D flame model, in conjunction with the source identity face rendering-related parameters, is fed into a differentiable renderer to generate a 2D image, completing a rough expression transfer. However, due to the inability of the 3DMM to reconstruct the hair and teeth in the face, the rough expression transfer 2D images obtained in the first stage have obvious artificial traces. To address the above issues, this paper uses the diffusion model in the second stage to repair artificial traces in 2D images, aiming to achieve a close resemblance to the source identity face after expression transfer.

There has been an encouraging progress in the affective states recognition models based on the single-modality signals as electroencephalogram (EEG) signals or peripheral physiological signals in recent years. However, multimodal physiological signals-based affective states recognition methods have not been thoroughly exploited yet. Here we propose Multiscale Convolutional Neural Networks (Multiscale CNNs) and a biologically inspired decision fusion model for multimodal affective states recognition. First, the raw signals are pre-processed with baseline signals. Then, the High Scale CNN and Low Scale CNN in Multiscale CNNs are utilized to predict the probability of affective states output for EEG and each peripheral physiological signal respectively. Finally, the fusion model calculates the reliability of each single-modality signals by the euclidean distance between various class labels and the classification probability from Multiscale CNNs, and the decision is made by the more reliable modality information while other modalities information is retained. We use this model to classify four affective states from the arousal valence plane in the DEAP and AMIGOS dataset. The results show that the fusion model improves the accuracy of affective states recognition significantly compared with the result on single-modality signals, and the recognition accuracy of the fusion result achieve 98.52 and 99.89 percent in the DEAP and AMIGOS dataset respectively.

In this paper, we consider a three-layer distributed multi-access edge computing (MEC) network where multiple clouds, MEC servers, and edge devices (EDs) are deployed at the top layer, middle layer, and bottom layer, respectively. Each cloud center (CC) is associated with an independent service provider and publishes an application-driven computing task. To deliver the tasks, CCs rely on EDs to generate the raw data and offload part of the computing tasks to both EDs and MEC servers such that their computing and transmission resources can be fully utilized to reduce the system latency. However, in such a three-layer network, the distributed deployment of tasks leads to inevitable resource competition among CCs. To address this issue, we propose a distributed scheme based on multi-agent reinforcement learning, where each CC jointly determines the task offloading and resource allocation strategy based on its inference of other CCs' decisions. Simulation results indicate that a lower system latency is achieved via our proposed scheme compared with the existing schemes. In addition, the influence of the number of CCs, MEC servers, and EDs on latency performance is also discussed.

Hashing based multimedia retrieval methods have been widely studied because of the advantages of computation efficiency and low storage cost. Generally, there are two ways of doing hashing: Supervised and unsupervised hashing. Supervised hashing was known to be more efficient and popular. However, in many real applications, only unsupervised method work because there are no labels. But unsupervised hashing method still face some challenges such as how to get semantic information from unlabeled data. To address this problem, in this paper, we propose a deep unsupervised hashing method based on semantic segmentation (Seg-HashNet). Specifically, a pre-trained semantic segmentation model is used to mine informative semantic information to guide the learning process and generate highquality hash codes. Extensive experiments on two public datasets have been performed, which showed that Unsupervised Semantic Segmentation hashing is superior to the latest unsupervised hashing method in image retrieval tasks.

As an important field of research in Human-Machine Interactions, emotion recognition based on the electroencephalography (EEG) signals has become common research. The traditional machine learning approaches use well-designed classifiers with hand-crafted features which may be limited to domain knowledge. Motivated by the outstanding performance of deep learning approaches in recognition tasks, we proposed a 3D convolutional neural network model to extract the spatial-temporal features automatically in the EEG signals. By the pre-processing method with baseline signals and the electrode topological structure relocated, the proposed model achieves a high accuracy rate of 96.61%, 96.43% in the Two class classification task (low/high arousal, low/high valence) and 93.53% in the Four class classification task (low arousal and low valence/high arousal and low valence/low arousal and high valence/high arousal and high valence) in the DEAP dataset, and 97.52%, 96.96% in the Two class classification task and 95.86% in the Four class classification task in the AMIGOS dataset.

With monocular Visual-Inertial Odometry (VIO) system, 3D point cloud and camera motion can be estimated simultaneously. Because pure sparse 3D points provide a structureless representation of the environment, generating 3D mesh from sparse points can further model the environment topology and produce dense mapping. To improve the accuracy of 3D mesh generation and localization, we propose a tightly-coupled monocular VIO system, PLP-VIO, which exploits point features and line features as well as plane regularities. The co-planarity constraints are used to leverage additional structure information for the more accurate estimation of 3D points and spatial lines in state estimator. To detect plane and 3D mesh robustly, we combine both the line features with point features in the detection method. The effectiveness of the proposed method is verified on both synthetic data and public datasets and is compared with other state-of-the-art algorithms.

In this work, we consider a heterogeneous multi-cloud mobile edge computing (Het-MEC) network, where multiple independent cloud centers (CCs) publish tasks to edge devices (EDs) and MEC servers, and compete for their computing and transmission resources. To minimize the system latency, we propose a distributed scheme for each CC to determine its data offloading and resource allocation strategy independently. Competition among multiple CCs in this distributed scheme is depicted by our designed multi-agent reinforcement learning (MARL) based algorithm, where each CC is self-motivated to learn the explicit models of other CCs and adjusts their behaviors. Simulation results indicate that multiple clouds are self-organized to take full advantage of computing and transmission resources to minimize their own task latency, while a lower system latency can be achieved compared with the cloud computing and local computing schemes.

Here we study the low-resolution visual recognition problem. Conventional methods are usually trained on images with large ROIs (regions of interest), while the regions and insider images are often small and blur in real-world applications. Therefore, deep neural networks learned on high-resolution images cannot be directly used for recognizing low-resolution objects. To overcome this challenging problem, we propose to use the teacher-student learning paradigm for distilling useful feature information from a pre-trained deep model on high-resolution visual data. In practice, a distillation loss is used to seek the perceptual consistency of low-resolution images and high-resolution images. By simultaneously optimizing the recognition loss and distillation loss, we formulate a novel low-resolution recognition approach. Experiments conducted on benchmarks demonstrate that the proposed method is capable to learn well-performed models for recognizing low-resolution objects, which is superior to the state-of-the-art methods.

Railway transportation plays an important role in both economic and social development. The requirements of the railway traffic increase in recent decades. In order to meet the growing demand, a new generation control system of railway transportation emerges. It consists of collection, transmission, analysis and scheduling module. In such a context, an information transmission system is built to connect trains and scheduling center. However, the infrastructure of the railway system cannot provide enough bandwidth for such amount of data. As a result, the efficiency of data transmission cannot be ensured. In this paper, we focus on the compression algorithm that reduce the amount of transmitted data and improve the system performance. Based on the analysis of the common algorithms, an efficient compression algorithm, named delta-encoding, is proposed. It consists of two steps: preprocessing and compression. Delta-encoding utilizes a class-based difference model, which reduces the data redundancy, to realize a preprocessing algorithm. With the combination of preprocessing algorithm and a regular compression algorithm, delta-encoding has better performance on compression ratio, and becomes a universal hybrid algorithm for structured data in IoT system rather than a specific algorithm in high-speed train system. Finally, several experiments are provided to prove that delta-encoding have advantages in both compression ratio and compression time.