High Computational Power Research Articles

Driver Identification Using Optimized Deep Learning Model in Smart Transportation

The Intelligent Transportation System (ITS) is said to revolutionize the travel experience by making it safe, secure, and comfortable for the people. Although vehicles have been automated up to a certain extent, it still has critical security issues that require thorough study and advanced solutions. The security vulnerabilities of ITS allows the attacker to steal the vehicle. Therefore, the identification of drivers is required in order to develop a safe and secure system so that the vehicles can be protected from theft. There are two ways in which a driver can be identified: 1) face recognition of the driver, and 2) based on driving behavior. Face recognition includes image processing of 2-D images and learning of the features, which require high computational power. Drivers are known to have unique driving styles, whose data can be captured by the sensors. Therefore, the second method identifies drivers based on the analysis of the sensor data and it requires comparatively lesser computational power. In this paper, an optimized deep learning model is trained on the sensor data to correctly identify the drivers. The Long Short-Term Memory (LSTM) deep learning model is optimized for better performance. The novelty of the approach in this work is the inclusion of hyperparameter tuning using a nature-inspired optimization algorithm, which is an important and essential step in discovering the optimal hyperparameters for training the model which in turn increases the accuracy. The CAN-BUS dataset is used for experimentation and evaluation of the training model. Evaluation parameters such as accuracy, precision score, F1 score, and ROC AUC curve are considered to evaluate the performance of the model.

Compact Image-Style Transfer: Channel Pruning on the Single Training of a Network.

Recent image-style transfer methods use the structure of a VGG feature network to encode and decode the feature map of the image. Since the network is designed for the general image-classification task, it has a number of channels and, accordingly, requires a huge amount of memory and high computational power, which is not mandatory for such a relatively simple task as image-style transfer. In this paper, we propose a new technique to size down the previously used style transfer network for eliminating the redundancy of the VGG feature network in memory consumption and computational cost. Our method automatically finds a number of consistently inactive convolution channels during the network training phase by using two new losses, i.e., channel loss and xor loss. The former maximizes the number of inactive channels and the latter fixes the positions of these inactive channels to be the same for the image. Our method improves the image generation speed to be up to 49% faster and reduces the number of parameters by 20% while maintaining style transferring performance. Additionally, our losses are also effective in pruning the VGG16 classifier network, i.e., parameter reduction by 26% and top-1 accuracy improvement by 0.16% on CIFAR-10.

Spatial filtering in a 6D hybrid-Vlasov scheme to alleviate adaptive mesh refinement artifacts: a case study with Vlasiator (versions 5.0, 5.1, and 5.2.1)

Abstract. Numerical simulation models that are used to investigate the near-Earth space plasma environment require sophisticated methods and algorithms as well as high computational power. Vlasiator 5.0 is a hybrid-Vlasov plasma simulation code that is able to perform 6D (3D in ordinary space and 3D in velocity space) simulations using adaptive mesh refinement (AMR). In this work, we describe a side effect of using AMR in Vlasiator 5.0: the heterologous grid approach creates discontinuities due to the different grid resolution levels. These discontinuities cause spurious oscillations in the electromagnetic fields that alter the global results. We present and test a spatial filtering operator for alleviating this artifact without significantly increasing the computational overhead. We demonstrate the operator's use case in large 6D AMR simulations and evaluate its performance with different implementations.

Cyber collaborative warehouse with dual-cycle operations design

ABSTRACT Warehouse operations have been significantly improved because of the rapid advancement of cyber-physical system technologies, preparing for Work-of-the-Future. The challenge, however, is how to design a collaborative system to deliver optimal performance by multiple agents who are highly distributed but interconnected and operate with technologies that provide massive amounts of real-time data. To address the challenge, in this article, the Cyber Collaborative Protocol for Dual-Cycle Task in Future Warehouse is developed to minimise total operation cost and time. The problem is addressed in two phases; the Global and Local phases. The global phase has higher computational power, maintaining a mathematical model, while the local phase has limited computational power and time, utilising heuristics to deliver the outcome. Computer experiments are utilised for validating the designed protocol compared with other alternatives. The results show that, in all given scenarios, the newly designed protocol outperforms alternatives with statistical significance. The original contribution of this research is the design and control of Cyber Collaborative Warehouse operations with a new focus on collaborative multi-agent interactions. In addition, a major implication is that future warehouses can benefit competitively by operating with Task Administration Protocols such as the new Cyber Collaborative Protocol for Dual-Cycle Task in Future Warehouse.

LotuS2: an ultrafast and highly accurate tool for amplicon sequencing analysis

BackgroundAmplicon sequencing is an established and cost-efficient method for profiling microbiomes. However, many available tools to process this data require both bioinformatics skills and high computational power to process big datasets. Furthermore, there are only few tools that allow for long read amplicon data analysis. To bridge this gap, we developed the LotuS2 (less OTU scripts 2) pipeline, enabling user-friendly, resource friendly, and versatile analysis of raw amplicon sequences.ResultsIn LotuS2, six different sequence clustering algorithms as well as extensive pre- and post-processing options allow for flexible data analysis by both experts, where parameters can be fully adjusted, and novices, where defaults are provided for different scenarios.We benchmarked three independent gut and soil datasets, where LotuS2 was on average 29 times faster compared to other pipelines, yet could better reproduce the alpha- and beta-diversity of technical replicate samples. Further benchmarking a mock community with known taxon composition showed that, compared to the other pipelines, LotuS2 recovered a higher fraction of correctly identified taxa and a higher fraction of reads assigned to true taxa (48% and 57% at species; 83% and 98% at genus level, respectively). At ASV/OTU level, precision and F-score were highest for LotuS2, as was the fraction of correctly reported 16S sequences.ConclusionLotuS2 is a lightweight and user-friendly pipeline that is fast, precise, and streamlined, using extensive pre- and post-ASV/OTU clustering steps to further increase data quality. High data usage rates and reliability enable high-throughput microbiome analysis in minutes.AvailabilityLotuS2 is available from GitHub, conda, or via a Galaxy web interface, documented at http://lotus2.earlham.ac.uk/.6hTcPpJ8wvT9bQx45qQZLRVideo

Design Space Exploration of Clustered Sparsely Connected MPSoC Platforms.

Heterogeneous multiprocessor platforms are the foundation of systems that require high computational power combined with low energy consumption, like the IoT and mobile robotics. In this paper, we present five new algorithms for the design space exploration of platforms with elements grouped in clusters with very few connections in between, while these platforms have favorable electric properties and lower production costs, the limited interconnectivity and inability of heterogeneous platform elements to execute all types of tasks, significantly decrease the chance of finding a feasible mapping of application to the platform. We base the new algorithms on the Non-dominated Sorting Genetic Algorithm II (NSGA-II) meta-heuristic and the previously published SDSE mapping algorithm designed for fully interconnected multiprocessor platforms. With the aim to improve the chance of finding feasible solutions for sparsely connected platforms, we have modified the parts of the search process concerning the penalization of infeasible solutions, chromosome decoding, and mapping strategy. Due to the lack of adequate existing benchmarks, we propose our own synthetic benchmark with multiple application and platform models, which we believe can be easily extended and reused by other researchers for further studying this type of platform. The experiments show that four proposed algorithms can find feasible solutions in 100% of test cases for platforms with dedicated clusters. In the case of tile-like platforms, the same four algorithms show an average success rate of 60%, with one algorithm going up to 84%.

A CNN Hardware Accelerator Using Triangle-based Convolution

Convolutional neural networks (CNNs) have gained a massive impression in the fields of computer vision and especially in the embedded applications because of their high accuracy and performance. However, high computational complexity and power consumption due to convolution operations causes a high demand for low-power accelerators. A 3D geometric optimization strategy is proposed to alleviate the area and power requirements of Multiply Accumulate operations prevalent in all spatial CNNs. The proposed technique is generic and may be easily scaled for accelerators performing spatial 2D convolution.

Bayesian models for spatially explicit interactions between neighbouring plants

Abstract Interactions between neighbouring plants drive population and community dynamics in terrestrial ecosystems. Understanding these interactions is critical for both fundamental and applied ecology. Spatial approaches to model neighbour interactions are necessary, as interaction strength depends on the distance between neighbouring plants. Recent Bayesian advancements, including the Hamiltonian Monte Carlo algorithm, offer the flexibility and speed to fit models of spatially explicit neighbour interactions. We present a guide for parameterizing these models in the Stan programming language and demonstrate how Bayesian computation can assist ecological inference on plant–plant interactions. Modelling plant neighbour interactions presents several challenges for ecological modelling. First, nonlinear models for distance decay can be prone to identifiability problems, resulting in lack of model convergence. Second, the pairwise data structure of plant–plant interaction matrices often leads to large matrices that demand high computational power. Third, hierarchical structure in plant–plant interaction data is ubiquitous, including repeated measurements within field plots, species and individuals. Hierarchical terms (e.g. ‘random effects’) can result in model convergence problems caused by correlations between coefficients. We explore modelling solutions for these challenges with examples representing spatial data on plant demographic rates: growth, survival and recruitment. We show that ragged matrices reduce computational challenges inherent to pairwise matrices, resulting in higher efficiency across data types. We also demonstrate how metrics for model convergence, including divergent transitions and effective sample size, can help diagnose problems that result from complex nonlinear structures. Finally, we explore when to use different model structures for hierarchical terms, including centred and non‐centred parameterizations. We provide reproducible examples written in Stan to enable ecologists to fit and troubleshoot a broad range of neighbourhood interaction models. Spatially explicit models are increasingly central to many ecological questions. Our work illustrates how novel Bayesian tools can provide flexibility, speed and diagnostic capacity for fitting plant neighbour models to large, complex datasets. The methods we demonstrate are applicable to any dataset that includes a response variable and locations of observations, from forest inventory plots to remotely sensed imagery. Further developments in statistical models for neighbour interactions are likely to improve our understanding of plant population and community ecology across systems and scales.

Rock Classification with Features Based on Higher Order Riesz Transform

Most modern algorithms use convolutional neural networks to classify image data of different kinds. While this approach is a good method to differentiate between natural images of objects, big datasets are needed for the training process. Another drawback is the demand for high computational power. We introduce a new approach which involves classic feature vectors with structural information based on higher order Riesz transform. Following this way we create a framework specialized for texture data like images of rock cross-sections. The key advantages are faster computations and more versatile choices of the underlying machine learning tools while maintaining a comparable accuracy in comparison with state-of-the-art algorithms.

Underground Metal Pipeline Localization Using Low-Cost Wireless Magnetic Sensors Mounted on an Excavator

To minimize the excavation damage of underground pipelines, in this article, we propose to provide real-time feedback to operators by monitoring the distance information from the underground pipeline. Despite multiple efforts to locate underground ferromagnetic pipelines using the magnetic anomaly detection, it has not been implemented in real-time excavation operations due to low signal-to-noise ratios, unknown external magnetic interference (EMI), and high computational power. To address these limitations, we propose an approach to locating a custom-designed wireless sensor system on the excavator bucket to increase accuracy of real-time distance estimation with efficient EMI cancelation. We present the system using a 14:1 lab-scaled excavator and six sizes of ferromagnetic pipes. To create a distance estimation model of each pipe, we measure the magnetic anomaly caused by ferromagnetic pipes from eight distances (from 5 to 75 mm, spaced 10 mm apart). The wireless system with the localization algorithm successfully estimates the distance between the sensor and the pipe with a less than 4.38 ± 1.62 mm error overall. This pilot study shows that the proposed system with the localization algorithm is able to accurately locate pipes. This will lay the foundation for providing accurate spatial information, and ultimately prevent critical excavation/drilling damage at construction sites.

Evaluation study of face detection by Viola-Jones algorithm

Face detection technology is an essential step in almost all face related analysis applications such as face feature extraction, face alignment, face verification, face identification, face parsing, face recognition, age recognition, and gender classification. Numerous algorithms were introduced for face detection, one of which is the Viola-Jones algorithm being introduced in 2001. This algorithm is still widely used due to its simplicity and ability of detection in real-time with relatively high accuracy and low computational power requirements compared to other recent algorithms such as deep learning based algorithms. In this paper, Viola-Jones algorithm is implemented and evaluated through different tests. And its strengths, limitations, and affecting factors are provided according to the obtained results. This paper concentrates on the algorithm limitations and the reasons of these limitations, and suggests some solutions if possible. This can help in enhancing the algorithm performance by increasing the detection accuracy or reducing the time taken for detection or training…etc.

Development and assessment of a reactor system prognosis model with physics-guided machine learning

Autonomous control systems provide recommendations to help operators in decision-making during plant operations ranging from normal operation to accident management. An important step of autonomous control is prognosis. In nuclear engineering domain, prognosis is the process of predicting future conditions of a system or equipment based on present signs and symptoms of a fault. The prognosis model allows predicting future reactor states for possible candidate control strategies so that the outcomes can be evaluated to determine the best control strategy. The prognosis model requires representing direct relationships between the symptoms and the predictions. In nuclear engineering, computational simulations are approximate representations of the operation of the real system. However, prognosis with computational simulations requires high computation power and time due to possible large number of scenarios. Necessary computation resources can be reduced with machine learning (ML) approach for fast predictions by building a surrogate function using the simulation data. A critical issue is, ML models are ignorant of physical knowledge, and these models approximate statistical relationships between the system variables. This ignorance can produce results that are inconsistent with physical laws, even if an optimal result is achieved from a mathematical point of view. Physics-guided machine learning (PGML) is an approach to tackle this issue. This work formulates and illustrates a framework to guide development and assessment of the ML-based prognosis model for autonomous control systems. The development of the prognosis model considers the training of a ML model which consists of optimizing many aspects of the ML approach. The assessment of the prognosis model considers training data limitations and uncertainties of the ML approach. Prognosis models with standalone ML and PGML are developed and assessed on the loss-of-flow scenario of Experimental Breeder Reactor II. The results indicate that PGML based prognosis model has the best performance compared to other prognosis models.

LR-AKAP: A Lightweight and Robust Security Protocol for Smart Home Environments.

For the betterment of human life, smart Internet of Things (IoT)-based systems are needed for the new era. IoT is evolving swiftly for its applications in the smart environment, including smart airports, smart buildings, smart manufacturing, smart homes, etc. A smart home environment includes resource-constrained devices that are interlinked, monitored, controlled, and analyzed with the help of the Internet. In a distributed smart environment, devices with low and high computational power work together and require authenticity. Therefore, a computationally efficient and secure protocol is needed. The authentication protocol is employed to ensure that authorized smart devices communicate with the smart environment and are accessible by authorized personnel only. We have designed a novel, lightweight secure protocol for a smart home environment. The introduced novel protocol can withstand well-known attacks and is effective with respect to computation and communication complexities. Comparative, formal, and informal analyses were conducted to draw the comparison between the introduced protocol and previous state-of-the-art protocols.

Neuro‐fuzzy Elman wavelet control for nonlinear uncertain systems with fuzzy input and unknown fuzzy disturbances: Application to robotics

SummaryKeeping into view the irregular existence of uncertainties and disturbances in nonlinear systems, a novel adaptive control scheme based on neuro‐fuzzy Elman wavelet neural network (NFEWNN) has been proposed in this work which is capable of accurately modeling system uncertainties. In addition to this, an improvement in an evolutionary algorithm named particle swarm optimization has also been introduced to optimize the initial parameters of the controller. Uncertain non‐linear discrete‐time systems in the non‐strict feedback form with unknown disturbances are taken into consideration. This work aims to design an efficient adaptive control scheme for the proposed systems in the case of fuzzy input and fuzzy external disturbance which makes the control of the system more complex and difficult. The proposed NFEWNN control scheme is capable of tracking the desired trajectory successfully along with accurately modeling system uncertainties with high computational power, low computational burden, and rapid convergence rate. The stability of the controller has been proved using Lyapunov theory and guaranteed convergence of the controller is proved. Finally, real‐world applications of the work have been found in robotics. The controller has been implemented on the robotic manipulator and the dynamic model of aerial robotic vehicles to prove its effectiveness and efficiency.

A multi-Kalman filter-based approach for decoding arm kinematics from EMG recordings

BackgroundRemarkable work has been recently introduced to enhance the usage of Electromyography (EMG) signals in operating prosthetic arms. Despite the rapid advancements in this field, providing a reliable, naturalistic myoelectric prosthesis remains a significant challenge. Other challenges include the limited number of allowed movements, lack of simultaneous, continuous control and the high computational power that could be needed for accurate decoding. In this study, we propose an EMG-based multi-Kalman filter approach to decode arm kinematics; specifically, the elbow angle (θ), wrist joint horizontal (X) and vertical (Y) positions in a continuous and simultaneous manner.ResultsTen subjects were examined from which we recorded arm kinematics and EMG signals of the biceps, triceps, lateral and anterior deltoid muscles corresponding to a randomized set of movements. The performance of the proposed decoder is assessed using the correlation coefficient (CC) and the normalized root-mean-square error (NRMSE) computed between the actual and the decoded kinematic. Results demonstrate that when training and testing the decoder using same-subject data, an average CC of 0.68 ± 0.1, 0.67 ± 0.12 and 0.64 ± 0.11, and average NRMSE of 0.21 ± 0.06, 0.18 ± 0.03 and 0.24 ± 0.07 were achieved for θ, X, and Y, respectively. When training the decoder using the data of one subject and decoding the data of other subjects, an average CC of 0.61 ± 0.19, 0.61 ± 0.16 and 0.48 ± 0.17, and an average NRMSE of 0.23 ± 0.07, 0.2 ± 0.05 and 0.38 ± 0.15 were achieved for θ, X, and Y, respectively.ConclusionsThese results suggest the efficacy of the proposed approach and indicates the possibility of obtaining a subject-independent decoder.

Planet Optimization with Deep Convolutional Neural Network for Lightweight Intrusion Detection in Resource-Constrained IoT Networks

Cyber security is becoming a challenging issue, because of the growth of the Internet of Things (IoT), in which an immense quantity of tiny smart gadgets push trillions of bytes of data over the Internet. Such gadgets have several security flaws, due to a lack of hardware security support and defense mechanisms, thus, making them prone to cyber-attacks. Moreover, IoT gateways present limited security features for identifying such threats, particularly the absence of intrusion detection techniques powered by deep learning (DL). Certainly, DL methods need higher computational power that exceeds the capability of such gateways. This article focuses on the development of Planet Optimization with a deep convolutional neural network for lightweight intrusion detection (PODCNN-LWID) in a resource-constrained IoT environment. The presented PODCNN-LWID technique primarily aims to identify and categorize intrusions. In the presented PODCNN-LWID model, two major processes are involved, namely, classification and parameter tuning. At the primary stage, the PODCNN-LWID technique applies a DCNN model for the intrusion identification process. Next, in the second stage, the PODCNN-LWID model utilizes the PO algorithm as a hyperparameter tuning process. The experimental validation of the PODCNN-LWID model is carried out on a benchmark dataset, and the results are assessed using varying measures. The comparison study reports the enhancements of the PODCNN-LWID model over other approaches.

Real-Time Motion Detection Network Based on Single Linear Bottleneck and Pooling Compensation

Motion (change) detection is a basic preprocessing step in video processing, which has many application scenarios. One challenge is that deep learning-based methods require high computation power to improve their accuracy. In this paper, we introduce a novel semantic segmentation and lightweight-based network for motion detection, called Real-time Motion Detection Network Based on Single Linear Bottleneck and Pooling Compensation (MDNet-LBPC). In the feature extraction stage, the most computationally expensive CNN block is replaced with our single linear bottleneck operator to reduce the computational cost. During the decoder stage, our pooling compensation mechanism can supplement the useful motion detection information. To our best knowledge, this is the first work to use the lightweight operator to solve the motion detection task. We show that the acceleration performance of the single linear bottleneck is 5% higher than that of the linear bottleneck, which is more suitable for improving the efficiency of model inference. On the dataset CDNet2014, MDNet-LBPC increases the frames per second (FPS) metric by 123 compared to the suboptimal method FgSegNet_v2, ranking first in inference speed. Meanwhile, our MDNet-LBPC achieves 95.74% on the accuracy metric, which is comparable to the state-of-the-art methods.

Enabling Intelligent IoTs for Histopathology Image Analysis Using Convolutional Neural Networks.

Medical imaging is an essential data source that has been leveraged worldwide in healthcare systems. In pathology, histopathology images are used for cancer diagnosis, whereas these images are very complex and their analyses by pathologists require large amounts of time and effort. On the other hand, although convolutional neural networks (CNNs) have produced near-human results in image processing tasks, their processing time is becoming longer and they need higher computational power. In this paper, we implement a quantized ResNet model on two histopathology image datasets to optimize the inference power consumption. We analyze classification accuracy, energy estimation, and hardware utilization metrics to evaluate our method. First, the original RGB-colored images are utilized for the training phase, and then compression methods such as channel reduction and sparsity are applied. Our results show an accuracy increase of 6% from RGB on 32-bit (baseline) to the optimized representation of sparsity on RGB with a lower bit-width, i.e., <8:8>. For energy estimation on the used CNN model, we found that the energy used in RGB color mode with 32-bit is considerably higher than the other lower bit-width and compressed color modes. Moreover, we show that lower bit-width implementations yield higher resource utilization and a lower memory bottleneck ratio. This work is suitable for inference on energy-limited devices, which are increasingly being used in the Internet of Things (IoT) systems that facilitate healthcare systems.

An Edge-Based Architecture for Offloading Model Predictive Control for UAVs

Thanks to the development of 5G networks, edge computing has gained popularity in several areas of technology in which the needs for high computational power and low time delays are essential. These requirements are indispensable in the field of robotics, especially when we are thinking in terms of real-time autonomous missions in mobile robots. Edge computing will provide the necessary resources in terms of computation and storage, while 5G technologies will provide minimal latency. High computational capacity is crucial in autonomous missions, especially for cases in which we are using computationally demanding high-level algorithms. In the case of Unmanned Aerial Vehicles (UAVs), the onboard processors usually have limited computational capabilities; therefore, it is necessary to offload some of these tasks to the cloud or edge, depending on the time criticality of the application. Especially in the case of UAVs, the requirement to have large payloads to cover the computational needs conflicts with other payload requirements, reducing the overall flying time and hindering autonomous operations from a regulatory perspective. In this article, we propose an edge-based architecture for autonomous UAV missions in which we offload the high-level control task of the UAV’s trajectory to the edge in order to take advantage of the available resources and push the Model Predictive Controller (MPC) to its limits. Additionally, we use Kubernetes to orchestrate our application, which runs on the edge and presents multiple experimental results that prove the efficacy of the proposed novel scheme.

Change Point Models for Real-Time Cyber Attack Detection in Connected Vehicle Environment

Connected vehicle (CV) systems are subject to potential cyber attacks because of increasing connectivity between its different components, such as vehicles, roadside infrastructure, and traffic management centers. However, it is a challenge to detect security threats in real-time and develop appropriate or effective countermeasures for a CV system because of the dynamic behavior of such attacks, high computational power requirement, and a historical data requirement for training detection models. To address these challenges, statistical models, especially change point models, have potentials for real-time anomaly detection. Thus, the objective of this study is to investigate the efficacy of two change point models; Expectation Maximization (EM) and two forms of Cumulative Summation (CUSUM) algorithms (i.e., typical and adaptive), for real-time vehicle-to-infrastructure (V2I) cyber attack detection in a CV Environment. To prove the efficacy of these models, we evaluated these two models for three different types of cyber attack, denial of service (DOS), impersonation, and false information, using basic safety messages (BSMs) generated from CVs through simulation. Results from numerical analysis revealed that EM, CUSUM, and adaptive CUSUM (aCUSUM) could detect these cyberattacks, such as DOS, impersonation, and false information with low false positives.