Search

Real-time execution of machine learning (ML) pipelines on radiology images is difficult due to limited computing resources in clinical environments, whereas running them in research clusters requires efficient data transfer capabilities. We developed Niffler, an open-source Digital Imaging and Communications in Medicine (DICOM) framework that enables ML and processing pipelines in research clusters by efficiently retrieving images from the hospitals’ PACS and extracting the metadata from the images. We deployed Niffler at our institution (Emory Healthcare, the largest healthcare network in the state of Georgia) and retrieved data from 715 scanners spanning 12 sites, up to 350 GB/day continuously in real-time as a DICOM data stream over the past 2 years. We also used Niffler to retrieve images bulk on-demand based on user-provided filters to facilitate several research projects. This paper presents the architecture and three such use cases of Niffler. First, we executed an IVC filter detection and segmentation pipeline on abdominal radiographs in real-time, which was able to classify 989 test images with an accuracy of 96.0%. Second, we applied the Niffler Metadata Extractor to understand the operational efficiency of individual MRI systems based on calculated metrics. We benchmarked the accuracy of the calculated exam time windows by comparing Niffler against the Clinical Data Warehouse (CDW). Niffler accurately identified the scanners’ examination timeframes and idling times, whereas CDW falsely depicted several exam overlaps due to human errors. Third, with metadata extracted from the images by Niffler, we identified scanners with misconfigured time and reconfigured five scanners. Our evaluations highlight how Niffler enables real-time ML and processing pipelines in a research cluster.

Network slicing enables communication service providers to partition physical infrastructure into logically independent networks. Network slices must be provisioned to meet the service-level objectives (SLOs) of disparate offerings, such as enhanced mobile broadband, ultrareliable low-latency communications, and massive machine-type communications. Network orchestrators must customize service placement and scaling to achieve the SLO of each network slice. In this article, we discuss the challenges encountered by network orchestrators in allocating resources to disparate 5G network slices, and propose the use of artificial intelligence to make core placement and scaling decisions that meet the requirements of network slices deployed on shared infrastructure. We explore how artificial intelligence-driven scaling algorithms, coupled with functionality-aware placement, can enable providers to design closed-loop solutions to meet the disparate SLOs of future network slices.

Demands for wide-area connectivity between enterprise site-edge networks and central office core networks/cloud data centers have grown rapidly. Various software defined wide area network (SD-WAN) solutions have been developed with the primary aim of improving WAN link utilization. However, mechanisms used by existing SD-WAN solutions fail to provide high reliability and performance required by today’s edge to cloud applications. In this article, we present WAN-aware MPTCP which seamlessly aggregates multiple WAN links into a “big pipe” for better WAN resilience thus minimizing application performance degradation under WAN link failures. We leverage the congestion control of MPTCP to balance traffic across multiple WAN links. The key innovation is to combine LAN virtualization at end systems with WAN virtualization at SD-WAN gateways. Through evaluation in both emulated testbeds and real-world deployment, we demonstrate the performance gain of WAN-aware MPTCP in terms of resilience and throughput over existing SD-WAN solutions.

As the Internet of Things (IoT) is estimated to grow to 25 billion by 2021, there is a need for an effective and efficient Intrusion Detection System (IDS) for IoT devices. Traditional network-based IDSs are unable to efficiently detect IoT malware and new evolving forms of attacks like file-less attacks. In this article, we present a system level Device-Edge split IDS for IoT devices. Our IDS profiles IoT devices according to their “behavior” using system-level information like running process parameters and their system calls in an autonomous, efficient, and scalable manner and then detects anomalous behavior indicative of intrusions. The modular design of our IDS along with a unique device-edge split architecture allows for effective attack detection with minimal overhead on the IoT devices. We have extensively evaluated our system using a dataset of 3,973 traditional IoT malware samples and 8 types of sophisticated file-less attacks recently observed against IoT devices in our testbed. We report the evaluation results in terms of detection efficiency and computational.

Nowadays, mobile cloud applications have attracted millions of smartphone users due to the proliferation of cyber technologies and a wide range of applications. Mobile cloud forensic investigation methodologies need tremendous growth due to the increasing crime incidents. The forensic readiness model plays a significant role in the forensic investigation framework by ensuring the ease of investigation to the forensic investigator. The existing forensic readiness supports either the mobile device or cloud environment, which lacks to prepare the information for the investigation. This article presents a mobile cloud forensic readiness process model to identify the factors and prepare the information that effectively supports forensic investigations. The proposed model involves requirements for the mobile cloud forensics from multiple perspectives with the aim of developing the forensic-ready system. As a result, the proposed forensic readiness model enables the mobile cloud forensic to improve the accuracy of investigation as well as reduces the investigation time significantly.

Eyes are among the first features of the face to be noticed. Loss of an eye due to congenital, traumatic, or pathologic etiologies causes disfigurement and loss of sensory feedback. It leaves a psychological impact on the patient; thus, prosthesis should be provided at the earliest to raise the spirits of the afflicted. Iris positioning is one of the important steps in fabricating customized ocular prosthesis. In facial asymmetry cases, comparison of both irises together can be a major disadvantage. This case report illustrates the use of a regular eyeglass in one technique and the use of a bioart facebow in the second technique for centering of iris in a customized ocular prosthesis to overcome the disadvantages of earlier used standard techniques (McArthur’s ocular locator and fixed calliper, pupillometer, visual judgement, and graph grid) of iris centering.

In this work, we examine the challenges that service providers encounter in managing complex service function graphs, while controlling service delivery latency. Based on the lessons we learn, we outline the design of a new system, Invenio, that empowers providers to effectively place microservices without prior knowledge of service functionality. Invenio correlates user actions with the messages they trigger seen in network traces, and computes procedural affinity for communication among microservices for each user action. The procedural affinity values can then be used to make placement decisions to meet latency constraints of individual user actions. Preliminary experiments with the Clearwater IP Multimedia Subsystem demonstrate that even a single high-latency link can result in significant performance degradation, and placement with Invenio can increase user quality of experience.

The cloud-native architecture and container-based technologies are revolutionizing how online services and applications are designed, developed, and managed by offering better elasticity and flexibility to developers and operators. However, the increasing adoption of microservice and serverless designs makes application workload more decomposed and transient at a larger scale. Most existing container orchestration systems still manage application workload based on simple system-level resource usage and policies manually created by operators, leading to ineffective application-agnostic scheduling and extra management burden for operators. In this work, we focus on workload placement for containerized applications and services and argue for the integration of application-level telemetry for profiling application status and co-locating application workload. To avoid extra performance overhead and modifications to existing applications, we propose to use the telemetry collected by service mesh to model the application communication patterns with a graph-based representation. By applying a graph partitioning algorithm, we create co-location groups for application workload that minimize cross-group communication traffic to improve the overall application performance, i.e., response time. Our preliminary experiments with a realistic online e-commerce application show that our solution can reduce the average response time by up to 58% compared to the default Kubernetes scheduler.

Freeze-drying is a deceptively complex operation requiring sophisticated design of a robust and efficient process that includes understanding and planning for heterogeneity across the batch and shifts in parameters due to vial or lyophilizer changes. A software tool has been designed to assist in process development and scale-up based on a model that includes consideration of the process heterogeneity. Two drug formulations were used to test the ability of the new tool to develop a freeze-drying cycle and correctly predict product temperatures and drying times. Model inputs were determined experimentally, and the primary drying heterogeneous freeze-drying model was used to design drying cycles that provided data to verify the accuracy of model-predicted product temperature and primary drying time. When model inputs were accurate, model-predicted primary drying times were within 0.1 to 15.9% of experimentally measured values, and product temperature accuracy was between 0.2 and 1.2°C for three vial locations, center, inner edge, and outer edge. However, for some drying cycles, differences in vial heat transfer coefficients due to changes in shelf and product temperature as well as altered product resistance due to product temperature-dependent microcollapse increased inaccuracy (up to 28.6% difference in primary drying time and 5.1°C difference in product temperature). This highlights the need for careful determination of experimental conditions used to calculate model inputs. In future efforts, full characterization of location- and shelf temperature-dependentKv as well as location- and product temperature-dependentRp will enhance the accuracy of the predictions by the model within the user-friendly software.

This work describes the lyophilization process validation and consists of two parts. Part one (Part I: Process Design and Modeling) focuses on the process design and is described in the previous paper, while the current paper is devoted to process qualification and continued process verification. The goal of the study is to show the cutting edge of lyophilization validation based on the integrated community-based opinion and the industrial perspective. This study presents best practices for batch size determination and includes the effect of batch size on drying time, process parameters selection strategies, and batch size overage to compensate for losses during production. It also includes sampling strategies to demonstrate batch uniformity as well as the use of statistical models to ensure adequate sampling. Based on the LyoHUB member organizations survey, the best practices in determining the number of PPQ runs are developed including the bracketing approach with minimum and maximum loads. Standard practice around CQA and CPP selection is outlined and shows the advantages of using control charts and run charts for process trending and quality control. The case studies demonstrating the validation strategy for monoclonal antibody and the impact of the loading process on the lyophilization cycle and product quality as well as the special case of lyophilization for dual-chamber cartridge system are chosen to illustrate the process validation. The standard practices in the validation of the lyophilization process, special lyophilization processes, and their impact on the validation strategy are discussed. Graphical Abstract

This work presents a novel approach to detecting stress differences between experts and novices in Situation Awareness (SA) tasks during maritime navigation using one type of wearable sensor, Empatica E4 Wristband. We propose that for a given workload state, the values of biosignal data collected from wearable sensor vary in experts and novices. We describe methods to conduct a designed SA task experiment, and collected the biosignal data on subjects sailing on a 240° view simulator. The biosignal data were analysed by using a machine learning algorithm, a Convolutional Neural Network. The proposed algorithm showed that the biosingal data associated with the experts can be categorized as different from that of the novices, which is in line with the results of NASA Task Load Index (NASA-TLX) rating scores. This study can contribute to the development of a self-training system in maritime navigation in further studies.

Simultaneous segmentation of multiple organs from different medical imaging modalities is a crucial task as it can be utilized for computer-aided diagnosis, computer-assisted surgery, and therapy planning. Thanks to the recent advances in deep learning, several deep neural networks for medical image segmentation have been introduced successfully for this purpose. In this paper, we focus on learning a deep multi-organ segmentation network that labels voxels. In particular, we examine the critical choice of a loss function in order to handle the notorious imbalance problem that plagues both the input and output of a learning model. The input imbalance refers to the class-imbalance in the input training samples (i.e., small foreground objects embedded in an abundance of background voxels, as well as organs of varying sizes). The output imbalance refers to the imbalance between the false positives and false negatives of the inference model. In order to tackle both types of imbalance during training and inference, we introduce a new curriculum learning based loss function. Specifically, we leverage Dice similarity coefficient to deter model parameters from being held at bad local minima and at the same time gradually learn better model parameters by penalizing for false positives/negatives using a cross entropy term. We evaluated the proposed loss function on three datasets: whole body positron emission tomography (PET) scans with 5 target organs, magnetic resonance imaging (MRI) prostate scans, and ultrasound echocardigraphy images with a single target organ i.e., left ventricular. We show that a simple network architecture with the proposed integrative loss function can outperform state-of-the-art methods and results of the competing methods can be improved when our proposed loss is used.

Although having been the subject of intense research over the years, cardiac function quantification from MRI is still not a fully automatic process in the clinical practice. This is partly due to the shortage of training data covering all relevant cardiovascular disease phenotypes. We propose to synthetically generate short axis CINE MRI using a generative adversarial model to expand the available data sets that consist of predominantly healthy subjects to include more cases with reduced ejection fraction. We introduce a deep learning convolutional neural network (CNN) to predict the end-diastolic volume, end-systolic volume, and implicitly the ejection fraction from cardiac MRI without explicit segmentation. The left ventricle volume predictions were compared to the ground truth values, showing superior accuracy compared to state-of-the-art segmentation methods. We show that using synthetic data generated for pre-training a CNN significantly improves the prediction compared to only using the limited amount of available data, when the training set is imbalanced.

Intent-Based Networking (IBN) has been increasingly deployed in production enterprise networks. Automated network configuration in IBN lets operators focus on intents- i.e., the end to end business objectives-rather than spelling out details of the configurations that implement these objectives. Automation brings its own concerns as the administrators cannot rely on traditional network troubleshooting tools. This situation is further exacerbated in the case of stateful Network Functions (NFs) whose packet processing behavior depends on previously observed traffic patterns. To ensure that the network configuration and state derived from network automation matches the administrator’s specified intent, we propose, Epinoia, a network intent checker for stateful networks. Epinoia relies on a unified model for NFs by leveraging the causal precedence relationships that exist between NF packet I/Os and states. Scalability of Epinoia is achieved by decomposing intents into sub-checking tasks and maintaining a causality graph between checked invariants. Epinoia checks for network-wide intent violations incrementally to reduce overhead in the event of network changes. Our evaluation results using real-world network topologies show that Epinoia can perform comprehensive checking within a few seconds per network with intent updates.

Network Policy management is a tedious and laborious task because of scale and dynamic changes in the network. The advent of Softwarized Networks has led to a renewed interest in intent-based network policy management. Intent-based Networking provides a structured way of specifying the intent of policies which are automatically translated and compiled into network device configuration. While this top-down approach of policy intent to policy configuration has worked well for cloud-native infrastructures such as data centers, it has not seen much adoption in legacy networks. We believe one of the primary reasons for this is the semantic gap between policy intents and policy configurations. The problem is further exacerbated by the heterogeneity, scale-on-the-fly, fragmentation, and lack of structure in non-intent native networks. We introduce Policy Intent Inference (PII) System to bridge the semantic gap with its advanced inference layer that extracts the policy intents from policy configurations fragmented over disparate network devices. We adopt a bottom-up approach to extract all policies within network devices, abstract them into a structured data model, and with the use of clustering and information retrieval techniques, build an optimal solution to extract network-wide policy intents from the underlying network that eases policy management especially policy troubleshooting, reducing the configuration clutter and reducing the time taken to compile and resolve conflicts in policies.

Enterprises are containerizing their business applications and extending those applications from cloud to edge to achieve better flexibility, agility, and performance for their business workload. Unlike data centers, edge sites including infrastructure and orchestration systems are often heterogeneous and highly customized depending on the resource availability, business requirements of the use case, and technical requirements of the application. However, in many business use cases, the lack of IT professionals with proper domain expertise makes it very challenging to create, manage, and support heterogeneous containerized edge sites at a large scale. In this work, we present the eCaaS framework that provides automated lifecycle management of containerized edge sites and applications. With eCaaS, users can create customized edge sites with only high-level business intents which are analyzed and translated to deployment templates with low-level specifications. The edge site deployment templates are then automatically executed to build, deploy, and configure the containerized edge sites and applications. To support more customization options in the future, eCaaS decouples user intents, deployment rules, and deployment specifications and formulates deployment template generation as an SMT problem to achieve better scalability and extensibility. For creating an edge site with five nodes, eCaaS takes less than one second to generate the deployment template and less than ten minutes to complete the entire deployment.

Abstract This work describes lyophilization process validation and consists of two parts. Part I focuses on the process design and is described in the current paper, while part II is devoted to process qualification and continued process verification. The intent of these articles is to provide readers with recent updates on lyophilization validation in the light of community-based combined opinion on the process and reflect the industrial prospective. In this paper, the design space approach for process design is described in details, and examples from practice are provided. The approach shows the relationship between the process inputs; it is based on first principles and gives a thorough scientific understanding of process and product. The lyophilization process modeling and scale-up are also presented showing the impact of facility, equipment, and vial heat transfer coefficient. The case studies demonstrating the effect of batch sizes, fill volume, and dose strength to show the importance of modeling as well as the effect of controlled nucleation on product resistance are discussed. Graphical abstract

The basic importance of classifying locations within a city is to find the best suited place for an individual or an organization. It makes our work easier to find similar areas as per our requirement be it personal or professional. For the experimental setup, the author has selected the geographical location of Lucknow city and has been applied machine learning algorithm to predict the best location as per person’s need. For this experimental setup, author has applied a Python programming language. The experimental results have classified locations into different categories as per the input value of number of categories which found most satisfactory over result from the literature. The author also came to the conclusion that this model helps to analyze business purpose too.

The first idea that arises in mind after having read the title of this paper would be the heading of this section. This is most certainly because until now we have only heard about how amazing the blockchain technology has turned out to be in terms of cyber security. Almost all the previous methods of implementing cyber security are directly or indirectly implemented in the blockchain technology and it can, therefore, be treated like an epitome of cyber security implementations. This, with the fact that all the variations possible in those individual sub-technologies also become a variation in blockchain, gives us a wide array of possible configurations to suit the requirements of different shapes and sizes of organizations. Moreover, the blockchain is still in the emerging phase, meaning that it still has many undiscovered use cases in the current and future world. In such a situation, tweaking of the implementational mechanisms as per the requirements of the organization have led to the branching in the way in which blockchains are created. This is because different implementations cost differently. This can be considered as a step further in the implementation where all the tools that we are using have been replaced with optimized and efficient use of those tools. This paper tries to satiate the emerging hunger for utilizing the blockchain technology into more and more spheres of life than just currency.

India is a developing country and needs to occupy low-cost plus effective measures in controlling traffic. With the increase in Indian urbanization, more people are moving from rural areas to cities leading to excess population, which can be easily seen on roads. Due to increase in number of citizens in particular cities, automobiles can also be seen in enormous number. With the increment of vehicles, traffic system needs to be in order. Traffic system in most of the Indian cities is based on fixed time control mode or manual over-ride mode. But with the modernization with time, we need to implement a new and efficient traffic control system. In this research work, the authors have proposed an adaptive traffic light control system, which is helpful to avoid congestion of vehicles and time consumption. This adaptive traffic control system uses camera as an input sensor, which takes snapshots of roads from different angles and directions to provide input data as form of images. It will provide real-time traffic data. In this work OpenCV framework is used for identification and determination of objects in real-time traffic images captured by camera. For ground detection and background subtraction of images we used GM algorithm (part of OpenCV), which is also used to calculate the number of vehicles present in each line of road. Intersection traffic lights will change according to the traffic condition that will be detected through image and video feeds.

The proposed research aims at developing an application which uses artificial neural network to identify rash driving by analyzing the data coming from the rider’s smartphone sensors. The neural network classifies the driving data between normal driving and aggressive driving behavior using real-time smartphone sensor values. This way, we will be able to identify the aggressive style of a particular rider and can provide alert signals so that appropriate steps can be taken. To develop the neural network, TensorFlow is used so that the network can be integrated in the Android application and other platforms. To integrate the neural network in an Android environment TensorFlow Lite is used. The sensors required for the application to work are found in every modern smartphone in today’s world; no extra hardware is required by the application. The dataset used in the proposed paper was specifically prepared to train the neural network. The accuracy obtained during the training procedure was 93.67% and the evaluation of the model using testing dataset gave the accuracy of 96.11%. The proposed research is implemented to develop an Android application that identifies aggressive driving style of the smartphone owner.

Many science researchers believe that a healthy thinking depends on the quality of human sleep. Measuring sleep quality through the systems currently available is not only complicated but also largely expensive. Accurate assessment of the quality of sleep depends on many physical and environmental factors such as dimension of bed, light effects, room temperature, any type of disease and oxygen level. Accurate measurement of these factors and establishing their relationship with sleep quality is a complex process. In this work, to summarize the systems available at present in measuring sleep quality of narcolepsy patient, an algorithm is proposed based on EEG energy-level patterns and time slot. EEG data has proved to reflect the activities of brain over all the section with respect to human activities. They are useful in both cases that is in awakening states and in sleep stage. Most of the brain disease is due to the deterioration of brain cells, and another part of algorithm is to identify optimized time slot for narcolepsy patient in order to summarize and to save patient from radiation of sleep quality measuring equipment.

With the ingress of exponential advancement of Internet technologies & social media platforms, there is a potential increase that can be seen in the development of online commercial websites. With time, people started to buy goods from these websites. So, there is also a great increase in selling goods on the Internet. These sites also facilitate their customers to leave their reviews and share their experiences with other users also. These customer reviews help others to make their decision before buying that product. In other words, these reviews help to show the quality of the product. Hence for this process, mining and understanding of the reviews are very important. In this research work, authors aimed to tackle one of the natural language processing (NLP) problems, i.e., sentiment polarity classification. The authors have performed a study to compare the baseline and statistical method (machine learning) for polarity classification. This work is also intended to compare the baseline method and machine learning method, to understand which method is better and more appropriate toward sentiment classification problems with the help of Python programming. The experimental results found to be satisfactory and compared with the existing literature.

Cancer is termed as a fatal genetic disease that can change the behavior and growth of human cells. Breast cancer is the most prevalent cancer occurring among women in today’s world. According to an Indian survey, the death rate caused by cancer has increased by approximately 25% from the past few years. Breast cancer proves out to be deadly in majority of the cases and is the major reason for death among women. Tumor is a term used for the abnormal growth of cell tissues in the human body. Benign and Malignant are the two types of cancer tumors. The main objective of this research work is to prepare a report on the percentage of people suffering with cancer tumors using machine learning algorithms. The authors have taken advantage of the most efficient machine learning algorithms to develop models for prediction which will detect breast cancer occurring rate. Authors have used a breast cancer data set which consists of 567 rows of 30 different attributes of cancer characteristics out of which benign and malignant data has been taken as the target attribute. A prediction has been done to calculate the percentage of people affected with the benign and malignant tumors and pictorial results have been displayed as results. In this proposed approach, the authors have used supervised learning algorithms like K-Nearest-Neighbors, Naive Bayes, Decision tree, and support vector machine prediction models for an unbiased estimate. The results indicated that K-Nearest-Neighbors is the best predictor with 91.6% accuracy. This model will be capable of providing a higher degree of accuracy compared to already existing literature work. The authors have compared their experimental results with the existing approaches and found satisfactory.

Ambiguous surveillance and analysis of the theft activity are one of the main reasons for inadequate security management of a place. The place can be our home, office, shops, etc. In this research work, an automated system has been proposed, which uses contour displacement points resulting in eternal surveillance for detection of the theft activity through CCTV camera. The algorithm developed in this work, utilises object detection technique in presenting real-time detection characteristics of the stolen object and identity of the theft in the form of camera frames. Through massive literature work, it is found that ideas adopted in developing various intelligent surveillance systems are majorly theoretically driven and lacks with required strong output modules. This research work is intended to demonstrate the implications of a convolutional neural network towards the prediction of visual objects from a previously trained data set and then form a suitable model for identifying any theft activity. For the experimental purpose, authors have used OpenCV library [1], which tends to be a very effective component to accomplish the entire work objectively. At the last, results have been analysed and all results and developed conditional algorithm.

The sinking of the ship named Titanic is one of the most historic shipwrecks in the world. It was held on April 14, 1912. Thousands of people died in this accident. Out of 3000 passenger, almost 1500 cause death in this accident. The reason behind this accident is due to less lifeboat because they never thought that this ship would ever sink because it is one of the largest ships in history at that time. So in this paper, an analytical approach has been proposed by the authors in order to predict the survival rate of people on the Titanic ship. For the experimental study, authors have selected Titanic dataset and applied suitable classifiers with the help of Python programming. For study purpose, spot check algorithm has been applied to predict what kind of people was survived. The experimental results have shown the model prediction value around 86.29% through spot check algorithm which found most satisfactory over results found in the literature varied from 72 to 82% only.

Coarctation of Aorta (CoA) is a congenital disease consisting of a narrowing that obstructs the systemic blood flow. This proof-of-concept study aimed to develop a framework for automatically and robustly personalizing aortic hemodynamic computations for the assessment of pre- and post-intervention CoA patients from 3D rotational angiography (3DRA) data. We propose a framework that combines hemodynamic modelling and machine learning (ML) based techniques, and rely on 3DRA data for non-invasive pressure computation in CoA patients. The key features of our framework are a parameter estimation method for calibrating inlet and outlet boundary conditions, and regional mechanical wall properties, to ensure that the computational results match the patient-specific measurements, and an improved ML based pressure drop model capable of predicting the instantaneous pressure drop for a wide range of flow conditions and anatomical CoA variations. We evaluated the framework by investigating 6 patient datasets, under pre- and post-operative setting, and, since all calibration procedures converged successfully, the proposed approach is deemed robust. We compared the peak-to-peak and the cycle-averaged pressure drop computed using the reduced-order hemodynamic model with the catheter based measurements, before and after virtual and actual stenting. The mean absolute error for the peak-to-peak pressure drop, which is the most relevant measure for clinical decision making, was 2.98 mmHg for the pre- and 2.11 mmHg for the post-operative setting. Moreover, the proposed method is computationally efficient: the average execution time was of only $$2.1 \pm 0.8$$ minutes on a standard hardware configuration. The use of 3DRA for hemodynamic modelling could allow for a complete hemodynamic assessment, as well as virtual interventions or surgeries and predictive modeling. However, before such an approach can be used routinely, significant advancements are required for automating the workflow.

This paper presents an efficient topology and route management approach in Software-Defined Wide Area Networks (SD-WAN). Traditional WANs suffer from low utilization and lack of global view of the network. Therefore, during failures, topology/service/traffic changes, or new policy requirements, the system does not always converge to the global optimal state. Using Software Defined Networking architectures in WANs provides the opportunity to design WANs with higher fault tolerance, scalability, and manageability. We exploit the correlation matrix derived from monitoring system between the virtual links to infer the underlying route topology and propose a route update approach that minimizes the total route update cost on all flows. We formulate the problem as an integer linear programming optimization problem and provide a centralized control approach that minimizes the total cost while satisfying the quality of service (QoS) on all flows. Experimental results on real network topologies demonstrate the effectiveness of the proposed approach in terms of disruption cost and average disrupted flows.

In this paper, we propose an intelligent framework for Traffic Matrix (TM) inference in Software Defined Networks (SDN) where the Ternary Content Addressable Memory (TCAM) entries of switches are partitioned into two parts to: 1) effectively aggregate part of incoming flows for aggregate measurements, and 2) de-aggregate and directly measure the most informative flows for per-flow measurements. These measurements are then processed to effectively estimate the size of network flows. Under hard resource constraints of limited TCAM sizes, we show how to design the optimal and efficient-compressed flow aggregation matrices. We propose an optimal Multi-Armed Bandit (MAB) based algorithm to adaptively measure the most rewarding flows. We evaluate the performance of our framework using real traffic traces from different network environments and by considering two main applications: TM estimation and Heavy Hitter (HH) detection. Moreover, we have implemented a prototype of our framework in Mininet to demonstrate its effectiveness.

© 2017 Elsevier B.V. A key requirement for network management is the accurate and reliable monitoring of relevant network characteristics. In today's large-scale networks, this is a challenging task due to the scarcity of network measurement resources and the hard constraints that this imposes. This paper proposes a new framework, called SNIPER, which leverages the flexibility provided by Software-Defined Networking (SDN) to design the optimal observation or measurement matrix that can lead to the best achievable estimation accuracy using Matrix Completion (MC) techniques. To cope with the complexity of designing large-scale optimal observation matrices, we use the Evolutionary Optimization Algorithms (EOA) which directly target the ultimate estimation accuracy as the optimization objective function. We evaluate the performance of SNIPER using both synthetic and real network measurement traces from different network topologies and by considering two main applications for per-flow size and delay estimations. Our results show that SNIPER can be applied to a variety of network performance measurements under hard resource constraints. For example, by measuring only 8.8% of all per-flow path delays in Harvard network [1], congested paths can be detected with probability of 0.94. To demonstrate the feasibility of our framework, we also have implemented a prototype of SNIPER in Mininet.

In this paper, we introduce a new technique for partitioning a large-scale under-determined linear inverse problem into multiple smaller subproblems that can be efficiently solved independently, and in parallel. When it is impossible or inefficient to solve a large-scale under-determined linear inverse problem, this technique can be used to significantly speed up the computation process without compromising the accuracy of the solution. We present numerical results that show the effectiveness of this approach when applied to network inference problems including traffic matrix estimation and network anomaly detection, both are important for managing large, complex networks and cyber security. Our proposed framework is applicable to other emerging applications in computational intelligence that can be formulated as under-determined linear inverse problems.

Invasive coronary angiography (ICA) is the gold standard in Coronary Artery Disease (CAD) imaging. Detection of the end-diastolic frame (EDF) and, in general, cardiac phase detection on each temporal frame of a coronary angiography acquisition is of significant importance for the anatomical and non-invasive functional assessment of CAD. This task is generally performed via manual frame selection or semi-automated selection based on simultaneously acquired ECG signals – thus introducing the requirement of simultaneous ECG recordings. In this paper, we evaluate the performance of a purely image based workflow relying on deep neural networks for fully automated cardiac phase and EDF detection on coronary angiographies. A first deep neural network (DNN), trained to detect coronary arteries, is employed to preselect a subset of frames in which coronary arteries are well visible. A second DNN predicts cardiac phase labels for each frame. Only in the training and evaluation phases for the second DNN, ECG signals are used to provide ground truth labels for each angiographic frame. The networks were trained on 56,655 coronary angiographies from 6820 patients and evaluated on 20,780 coronary angiographies from 6261 patients. No exclusion criteria related to patient state (stable or acute CAD), previous interventions (PCI or CABG), or pathology were formulated. Cardiac phase detection had an accuracy of 98.8 %, a sensitivity of 99.3 % and a specificity of 97.6 % on the evaluation set. EDF prediction had a precision of 98.4 % and a recall of 97.9 %. Several sub-group analyses were performed, indicating that the cardiac phase detection performance is largely independent from acquisition angles, the heart rate of the patient, and the angiographic view (LCA / RCA). The average execution time of cardiac phase detection for one angiographic series was on average less than five seconds on a standard workstation. We conclude that the proposed image based workflow potentially obviates the need for manual frame selection and ECG acquisition, representing a relevant step towards automated CAD assessment.

In recent years, computational fluid dynamics (CFD) has become a valuable tool for investigating hemodynamics in cerebral aneurysms. CFD provides flow-related quantities, which have been shown to have a potential impact on aneurysm growth and risk of rupture. However, the adoption of CFD tools in clinical settings is currently limited by the high computational cost and the engineering expertise required for employing these tools, e.g., for mesh generation, appropriate choice of spatial and temporal resolution, and of boundary conditions. Herein, we address these challenges by introducing a practical and robust methodology, focusing on computational performance and minimizing user interaction through automated parameter selection. We propose a fully automated pipeline that covers the steps from a patient-specific anatomical model to results, based on a fast, graphics processing unit- (GPU-) accelerated CFD solver and a parameter selection methodology. We use a reduced order model to compute the initial estimates of the spatial and temporal resolutions and an iterative approach that further adjusts the resolution during the simulation without user interaction. The pipeline and the solver are validated based on previously published results, and by comparing the results obtained for 20 cerebral aneurysm cases with those generated by a state-of-the-art commercial solver (Ansys CFX, Canonsburg PA). The automatically selected spatial and temporal resolutions lead to results which closely agree with the state-of-the-art, with an average relative difference of only 2%. Due to the GPU-based parallelization, simulations are computationally efficient, with a median computation time of 40 minutes per simulation.

Background Structured reports have been shown to improve communication between radiologists and providers. However, some radiologists are concerned about resultant decreased workflow efficiency. We tested a machine learning-based algorithm designed to convert unstructured computed tomography pulmonary angiography (CTPA) reports into structured reports. Methods A self-supervised convolutional neural network-based algorithm was trained on a dataset of 475 manually structured CTPA reports. Labels for individual statements included “pulmonary arteries,” “lungs and airways,” “pleura,” “mediastinum and lymph nodes,” “cardiovascular,” “soft tissues and bones,” “upper abdomen,” and “lines/tubes.” The algorithm was applied to a test set of 400 unstructured CTPA reports, generating a predicted label for each statement, which was evaluated by two independent observers. Per-statement accuracy was calculated based on strict criteria (algorithm label counted as correct if the statement unequivocally contained content only related to that particular label) and a modified criteria, accounting for problematic statements, including typographical errors, statements that did not fit well into the classification scheme, statements containing content for multiple labels, etc. Results Of the 4,157 statements, 3,806 (91.6%) and 3,986 (95.9%) were correctly labeled by the algorithm using strict and modified criteria, respectively, while 274 (6.6%) were problematic for the manual observers to label, the majority of which (n = 173) were due to more than one section being included in one statement. Conclusion This algorithm showed high accuracy in converting free-text findings into structured reports, which could improve communication between radiologists and clinicians without loss of productivity and provide more structured data for research/data mining applications.

As the Internet of Things (IoT) grows at a rapid pace, there is a need for an effective and efficient form of security tailored for IoT devices. In this paper, we introduce E-Spion, an anomaly-based system level Intrusion Detection System (IDS) for IoT devices. E-Spion profiles IoT devices according to their 'behavior' using system level information, like running process parameters and their system calls, in an autonomous, efficient, and scalable manner. These profiles are then used to detect anomalous behaviors indicative of intrusions. E-Spion provides three layers of detection with increasing detection efficiency but at the same time higher overhead costs on the devices. We have extensively evaluated E-Spion using a comprehensive dataset of 3973 IoT malware samples in our testbed. We observe a detection efficiency ranging from 78% to 100% depending on the layers of detection employed. We provide an analysis and comparison of the different layers of E-Spion in terms of detection accuracy and overhead costs. We also analyze the behavior of the malware samples in terms of our device logs at each layer.

This study presents a new one-parameter family of the well-known fixed point iteration method for solving nonlinear equations numerically. The proposed family is derived by implementing approximation through a straight line. The presence of an arbitrary parameter in the proposed family improves convergence characteristic of the simple fixed point iteration as it has a wider domain of convergence. Furthermore, we propose many two-step predictor–corrector iterative schemes for finding fixed points, which inherit the advantages of the proposed fixed point iterative schemes. Finally, several examples are given to further illustrate their efficiency.

Transesophageal echocardiography (TEE) is routinely used to provide important qualitative and quantitative information regarding mitral regurgitation. Contemporary planning of surgical mitral valve repair, however, still relies heavily upon subjective predictions based on experience and intuition. While patient-specific mitral valve modeling holds promise, its effectiveness is limited by assumptions that must be made about constitutive material properties. In this paper, we propose and develop a semi-automated framework that combines machine learning image analysis with geometrical and biomechanical models to build a patient-specific mitral valve representation that incorporates image-derived material properties. We use our computational framework, along with 3D TEE images of the open and closed mitral valve, to estimate values for chordae rest lengths and leaflet material properties. These parameters are initialized using generic values and optimized to match the visualized deformation of mitral valve geometry between the open and closed states. Optimization is achieved by minimizing the summed Euclidean distances between the estimated and image-derived closed mitral valve geometry. The spatially varying material parameters of the mitral leaflets are estimated using an extended Kalman filter to take advantage of the temporal information available from TEE. This semi-automated and patient-specific modeling framework was tested on 15 TEE image acquisitions from 14 patients. Simulated mitral valve closures yielded average errors (measured by point-to-point Euclidean distances) of 1.86 ± 1.24 mm. The estimated material parameters suggest that the anterior leaflet is stiffer than the posterior leaflet and that these properties vary between individuals, consistent with experimental observations described in the literature.

Two forms of network inference (or tomography) problems have been studied rigorously: (a) traffic matrix estimation or completion based on link-level traffic measurements, and (b) link-level loss or delay inference based on end-to-end measurements. These problems are often posed as underdetermined linear inverse (UDLI) problems and solved in a centralized manner, where all the measurements are collected at a central node, which then applies a variety of inference techniques to estimate the attributes of interest. This paper proposes a novel framework for decentralizing these large-scale UDLI network inference problems by intelligently partitioning it into smaller sub-problems and solving them independently and in parallel. The resulting estimates, referred to as multiple descriptions, can then be fused together to compute the global estimate. We apply this Multiple Description and Fusion Estimation (MDFE) framework to three classical problems: traffic matrix estimation, traffic matrix completion, and loss inference. Using real topologies and traces, we demonstrate how MDFE can speed up computation time while maintaining (even improving) the estimation accuracy and how it enhances robustness against noise and failures. We also show that our MDFE framework is compatible with a variety of existing inference techniques used to solve the UDLI problems.

In this article, we focus on capturing an expert’s experiences using wearable sensors. For this, first, we outline a set of high-level tasks that facilitate the transfer of experience from an expert to a trainee. Next, we define a mapping strategy to associate each task with one or more low-level functions such as gaze, voice, video, body posture, hand/arm gestures, biosignals, fatigue levels, haptic feedback, and location of the user in the environment. These low-level functions are then decomposed to their associated state-of-the-art sensors. Based on the requirements and constraints associated with the use cases from three different industrial partners, a set of sensors are proposed for the experience-capturing prototype. Finally, we discuss the attributes and features of the proposed prototype, along with its key challenges, constraints, and possible future directions.

Working with large quantities of digital images can often lead to prohibitive computational challenges due to their massive number of pixels and high dimensionality. The extraction of compressed vectorial representations from images is therefore a task of vital importance in the field of computer vision. In this paper, we propose a new architecture for extracting such features from images in an unsupervised manner, which is based on convolutional neural networks. The model is referred to as the Unsupervised Convolutional Siamese Network (UCSN), and is trained to embed a set of images in a vector space, such that local distance structure in the space of images is approximately preserved. We compare the UCSN to several classical methods by using the extracted features as input to a classification system. Our results indicate that the UCSN produces vectorial representations that are suitable for classification purposes.

Skip connections in deep networks have improved both segmentation and classification performance by facilitating the training of deeper network architectures and reducing the risks for vanishing gradients. The skip connections equip encoder-decoder like networks with richer feature representations, but at the cost of higher memory usage, computation, and possibly resulting in transferring non-discriminative feature maps. In this paper, we focus on improving the skip connections used in segmentation networks. We propose light, learnable skip connections which learn to first select the most discriminative channels, and then aggregate the selected ones as single channel attending to the most discriminative regions of input. We evaluate the proposed method on 3 different 2D and volumetric datasets and demonstrate that the proposed skip connections can outperform the traditional heavy skip connections of 4 different models in terms of segmentation accuracy (2% Dice), memory usage (at least 50%), and the number of network parameters (up to 70%).

Network monitoring has been evolving over several years to be able to identify and react to issues at a faster rate to reduce network downtime. With the expansion of cloud-users and the need for higher networking capability, the deployments are vast and complex, constantly making network troubleshooting slower. In this paper, we introduce a novel network monitoring and troubleshooting architecture that is user-defined, allowing users to define their troubleshooting profiles for a network device. We introduce the framework that can be easily integrated with the network device operating system which can work either independently on the device or can be aggregated across multiple devices in a network deployment. We present use-cases where this device-level telemetry abstraction would be very useful and how it can be easily extended. Also, we describe the machine-learning aspect which predicts thresholds based on data patterns, thus automating responses when an anomalous event occurs. Finally, we analyze the footprint the framework would have on a real network device and its overall benefit for the network in terms of latency and reaction time.

Cloudification of edge to core infrastructure has led to new and rich application and service deployment and operational models. These ecosystems have complex relationships between the application vendors, infrastructure operators and application users. Traditional licensing and compliance enforcement methods such as those based on in person audits and dynamic issuing of license keys inhibit the resource provisioning and consumption flexibility offered by cloudified services due to scalability and management overheads. In this work, we argue the need for a trusted framework for application usage rights compliance. This new architecture named "Metered Boot" provides a way to realize trusted, capacity/usage based rights compliance for service deployments that allows decoupling of usage rights governed by application vendors from the resource provisioning by the infrastructure provider. We have built a Metered Boot prototype for a particular usecase of NFV usage rights compliance.

Internet of Things is a new innovative technology topic under a science fiction and imagination or virtual world associated by visualizing physical world. There are conjecture and cybernetic world still possible about it. This is a truly abundant terrain of fantasy and original ideas when the combination of real life and virtual life through IoT. The interconnection of embedded devices, sensors, actuators can communicate the information from one to another (M2M) over Internet and data exchange. The important challenge faced in IoT is security which helps to secure the information through data exchanging, and is known as Web of Things (WoT). Moreover, to perform routing in 6LoWPAN network, RPL uses destination-oriented DAG (DODAG) between the different IPv6 nodes. We can also interface the CoAP and 6LoWPAN through RPL using UDP protocol for various constrained nodes in Contiki OS. In this paper, we focus on the proposed project Constraint Application Protocol (CoAP) that falls under a software component at application layer. The DTLS security based on LESS algorithm is provided under the CoAP protocol and gives the unicast messages because DTLS does not support multicast. The proposed solution to make a multicast message is distributing session key using key distribution centre. This is used to encrypt or decrypt the multicast message and then design and improve the cryptography algorithm. We have implemented this proposed work on the Contiki OS with the framework using Cooja simulator that is an erudite tool to make the work more efficient and optimized it.

Published version, available at: http://dx.doi.org/10.3390/computers8010009Belongs to Special Issue "Augmented and Mixed Reality in Work Context", suppl. 9. With the recent developments in augmented reality (AR) technologies comes an increased interest in the use of smart glasses for hands-on training. Whether this interest is turned into market success depends at the least on whether the interaction with smart AR glasses satisfies users, an aspect of AR use that so far has received little attention. With this contribution, we seek to change this. The objective of the article, therefore, is to investigate user satisfaction in AR applied to three cases of practical use. User satisfaction of AR can be broken down into satisfaction with the interaction and satisfaction with the delivery device. A total of 142 participants from three different industrial sectors contributed to this study, namely, aeronautics, medicine, and astronautics. In our analysis, we investigated the influence of different factors, such as age, gender, level of education, level of Internet knowledge, and the roles of the participants in the different sectors. Even though users were not familiar with the smart glasses, results show that general computer knowledge has a positive effect on user satisfaction. Further analysis using two-factor interactions showed that there is no significant interaction between the different factors and user satisfaction. The results of the study affirm that the questionnaires developed for user satisfaction of smart glasses and the AR application performed well, but leave room for improvement.