Search

We propose a novel statistical approach to evaluate the statistical significance (reliability) of the results from discriminative sub-trajectory mining, which we call Statistically Discriminative Sub-trajectory Mining (Stat-DSM). Given two groups of trajectories, the goal of Stat-DSM is to extract moving patterns in the form of sub-trajectories that occur statistically significantly more often in one group than in the other. An advantage of the proposed method is that the statistical significance of the extracted sub-trajectories are properly controlled in the sense that the probability of finding a false discriminative sub-trajectory is smaller than a specified significance threshold α(e.g. 0.05), which is crucial when the method is used in scientific or social science studies under noisy environments. Finding such statistically discriminative sub-trajectories from a massive trajectory dataset is both computationally and statistically challenging. In the Stat-DSM method, we address these difficulties by introducing a tree representation of sub-trajectories, and applying a permutation-based statistical inference method to the tree. To the best of our knowledge, Stat-DSM is the first method that provides a statistical approach to quantify the reliability of discriminative sub-trajectory mining results. We illustrate the effectiveness and scalability of the Stat-DSM method by applying it to a real-world dataset containing 1,000,000 trajectories.

Soft robotic hands and grippers are increasingly attracting attention as robotic end effectors. Compared with their rigid counterparts, they are safer for human-robot and environment-robot interactions, easier to control, and more compliant, and they cost and weigh less. Design studies of soft robotic hands have focused mostly on the soft fingers and bending actuators. However, the palm is also an essential part in grasping. In this work, we propose a novel design for an inexpensive soft humanoid hand with pneumatic soft fingers and a soft palm. The configuration of the soft palm is based on a modular design that can be applied to actuate different kinds of existing soft fingers. The splaying of the fingers, bending of the whole palm, and abduction and adduction of the thumb are implemented by the palm. Moreover, we present a new design of soft finger, called the hybrid-bending soft finger (HBSF), that can both bend in the grasping axis and deflect in the side-to-side axis, achieving human-like motion (Figure 1). The functions of the HBSF and soft palm are evaluated both in simulation, using the Simulation Open Framework Architecture (SOFA) framework, and experimentally. Six finger designs with 1-11 longitudinal segments are analyzed. The versatility of the soft hand is evaluated and demonstrated experimentally by its grasping of objects according to Feix's taxonomy. The results demonstrate a great diversity of grasps, with 31 of the 33 grasp types of the taxonomy performed successfully with the proposed design, showing promise for grasping a large variety of objects with different shapes and weights.

In this paper, we examine the properly twice epi-differentiability and compute the second order epi-subderivative of the indicator function to a class of sets including the finite union of parabolically derivable and parabolically regular sets. In this way, we provide no-gap second order optimality conditions for a disjunctive constrained problem. Moreover, we derive applications of our results to some types of disjunctive programs.

This Letter proposed a maximum-a-posteriori-based, pilot-aided algorithm for iterative estimation of carrier frequency offset (CFO) and time-varying multiple-input multiple-output (MIMO) multipath channel responses in FD orthogonal frequency division multiplexing (OFDM) systems. Empirical results and related benchmark limits show that the proposed algorithm offers reliable system performance having (i) estimation's mean-squared errors close to related Bayesian Cramer Rao bounds, (ii) data-detection's bit-error rates close to the ideal values under the assumption of perfect channel estimation and synchronisation and (iii) high robustness against the time variation of wireless channels in a network with moving nodes.

We have developed a 5-electrode recording system that combines an implantable electromyography (EMG) device package with transcutaneous inductive power transmission, near-infrared (NIR) transcutaneous data telemetry and 3 Mbps Wi-Fi data acquisition for chronic EMG recording in vivo . This system comprises a hermetically-sealed single-chip, 5-electrode Implantable EMG Acquisition Device (IEAD), a custom external powering and Implant Telemetry Module (ITM), and a custom Wi-Fi-based Raspberry Pi-based Data Acquisition (RaspDAQ) and relay device. The external unit (ITM and RaspDAQ) is powered entirely by a single battery to achieve the objective of untethered EMG recording, for the convenience of clinicians and animal researchers. The IEAD acquires intramuscular EMG signals at 17.85 ksps/electrode while being powered transcutaneously by the ITM using 22 MHz near-field inductive coupling. The acquired EMG data is transmitted transcutaneously via NIR telemetry to the ITM, which in turn, transfers the data to the RaspDAQ for relaying to a laptop computer for display and storage. We have also validated the complete system by acquiring EMG signals from rodents for up to two months. Following the explantation of the devices, we have also conducted failure and histological analysis on the devices and the surrounding tissue, respectively.

Medical students with English as an additional language often face difficulties in acquiring English medical terminology derived from Greek and Latin morphemes. To address this problem, this study used a corpus-based approach to identify the most commonly occurring medical morphemes in four sources: Stedman's list of medical morphemes; the Cengage list of general English morphemes; the Center for Development in Learning list of general English morphemes; and the Medical Web Corpus—a web-based corpus of current medical texts available through Sketch Engine text analysis software. Three medical dictionaries were used to validate the findings, leading to a final list of 136 specialized medical morphemes which cover 8.5% of the lexical items in the Medical Web Corpus. The findings provide a reliable and useful resource to help medical students with English as an additional language enhance their English medical vocabulary.

This paper proposes a new calibration method for enhancing robot positional accuracy of the industrial manipulators. By combining the joint deflection model with the conventional kinematic model of a manipulator, the geometric errors and joint deflection errors can be considered together to increase its positional accuracy. Then, a neural network is designed to additionally compensate the unmodeled errors, specially, non-geometric errors. The teaching-learning-based optimization method is employed to optimize weights and bias of the neural network. In order to demonstrate the effectiveness of the proposed method, real experimental studies are carried out on HH 800 manipulator. The enhanced position accuracy of the manipulator after the calibration confirms the feasibility and more positional accuracy over the other calibration methods.

37. International Cosmic Ray Conference, ICRC 2021, Berlin, Germany, 15 Jul 2021 - 22 Jul 2021; Proceedings of Science 395, 1056 (2022). doi:10.22323/1.395.1056 special issue: "37th International Cosmic Ray Conference : ICRC2021 : 12-23 July 2021 : Berlin, Germany - Online / Editorial Board: Alexander Kappes, Institut für Kernphysik, Westfälische Wilhelms-Universität Münster, D-48149 Münster, Germany ; Bianca Keilhauer, Karlsruhe Institute of Technology, Institute for Astroparticle Physics (IAP)", Published by SISSA, Trieste

Enhancing performance of downlink MU systems is an attractive and important research for future wireless systems. The non-orthogonal multiple access (NOMA) method was proposed to improve the performance of MU systems. In order to further improve the outage probability (OP) and ergodic capacity (EC) of downlink NOMA MU systems, we propose the combination of precoding and NOMA methods, and then the OP and EC of MU systems with our novel method are derived in scenarios of perfect and imperfect successive interference cancellation (SIC) scheme. Moreover, the closed-form expression of OP and EC for both scenarios is theoretically derived and compared with Monte Carlo simulations. The results show that, the analysis method is accurate, and the proposed combining precoding and NOMA can further enhance the performance of MU systems in comparing with the original orthogonal multiple access method.

Robot kinematic calibration used to be carried out with the partial pose measurements (position only) of dimension 3 in industry, while full pose measurements (orientation and position) of dimension 6 sometimes could be considered to improve the calibration performance. This paper investigates the effects of measurement dimensions on robot calibration accuracy. It compares the resulting robot accuracies in both partial pose and full pose cases after calibrating three structural types of robot manipulators such as a serial manipulator (Hyundai HA-06 robot), a single closed-chain manipulator (Hyundai HX-165 robot), and a multiple closed-chain manipulator (Hyundai HP-160 robot). These comparative studies show when the full-pose based calibration need to be considered and how much it contributes the improvement of robot accuracy to the different structural type of robot manipulators.

Data nowadays are extremely valuable resource. However, data are created and stored in different places with various formats and types. As a result, it is not easy and efficient for data analysis and data mining which can make profits for every aspect of social applications. To overcome this problem, data conversion is a crucial step that we have to build for linking and merging different data resources to a unified data store. In this paper, based on the intermediate data conversion model, we proposed an elastic data conversion framework for data integration system. Besides, we also performed an experiment to evaluate our model using MySQL and MongoDB.

Although showing impressive therapeutic potential, treatments of leukemias with T-cells expressing chimeric antigen receptors (CARs) is limited by their risk of several severe side effects. To overcome these problems, a switchable CAR platform has been developed termed UniCAR. Unlike conventional CAR, which is directed against tumor-associated antigens, UniCAR treatment involves an intermediate target module (TM), which can cross-link UniCAR T cells with tumor cells and lead to destruction. The development of these novel TMs against different tumor targets requires numerous repetitive tests on different synthesizing trials, which is usually limited in quantity and time-consuming. Meanwhile, nano-biosensors are lately known as analytical tools, which are highly sensitive, label-free, rapid and reagent-saving. Among them, silicon nanowire (SiNW) sensors have been extensively investigated by researchers over the past decades thanks to their compatibility with CMOS technology, enabling mass production. In this work, we demonstrated the application of a previously published SiNW biosensor on the detection of the binding of UniCAR and a part of different TMs. The results underline the advantages of the SiNW sensor over the ELISA method in terms of ease of preparation, speed and sensitivity. The method is able to evaluate the binding affinity of UniCAR to different TMs and open a potential to quantify the number of active UniCAR T-cells in an in vivo sample at a later stage. In the end, the application of a nanosensor may speed up the R&D process of the UniCAR concept and later play an important role in clinical monitoring of immunotherapy, especially in the era of precision medicine.

Paraphrase identification plays an important role with various applications in natural language processing tasks such as machine translation, bilingual information retrieval, plagiarism detection, etc. With the development of information technology and the Internet, the requirement of textual comparing is not only in the same language but also in many different language pairs. Especially in Vietnamese, detecting paraphrase in the English–Vietnamese pair of sentences is a high demand because English is one of the most popular foreign languages in Vietnam. However, the in-depth studies on cross- language paraphrase identification tasks between English and Vietnamese are still limited. Therefore, in this paper, we propose a method to identify the English–Vietnamese cross-language paraphrase cases, using hybrid feature classes. These classes are calculated by using the fuzzy-based method as well as the siamese recurrent model, and then combined to get the final result with a mathematical formula. The experimental results show that our model achieves 87.4% F-measure accuracy.

Accurately modeling local surface properties of objects is crucial to many robotic applications, from grasping to material recognition. Surface properties like friction are however difficult to estimate, as visual observation of the object does not convey enough information over these properties. In contrast, haptic exploration is time consuming as it only provides information relevant to the explored parts of the object. In this work, we propose a joint visuo-haptic object model that enables the estimation of surface friction coefficient over an entire object by exploiting the correlation of visual and haptic information, together with a limited haptic exploration by a robotic arm. We demonstrate the validity of the proposed method by showing its ability to estimate varying friction coefficients on a range of real multi-material objects. Furthermore, we illustrate how the estimated friction coefficients can improve grasping success rate by guiding a grasp planner toward high friction areas., Published in RA-L and ICRA2021. 8 pages, 8 figures. Tin IEEE Robotics and Automation Letters, 2021

Exosomes contain cargoes of proteins, lipids, micro-ribonucleic acids, and functional messenger RNAs, and they play a key role in cell-to-cell communication and hold valuable information about biological processes such as disease pathology. To harvest their potentials in disease diagnostics, prognostics, and therapeutics, exosome isolation is a crucial first step in providing pure and intact samples for both research and clinical purposes. Unfortunately, conventional methods for exosome separation suffer from low purity, low capture efficiency, long processing time, large sample volume requirement, the need for dedicated equipment and trained personnel, and high cost. In the last decade, microfluidic devices, especially those that incorporate nanostructures, have emerged as superior alternatives for exosome isolation and detection. In this review, we examine microfluidic platforms, dividing them into six categories based on their capture mechanisms: passive-structure-based affinity, immunomagnetic-based affinity, filtration, acoustofluidics, electrokinetics, and optofluidics. Here, we start out exploring the research and clinical needs that translate into important performance parameters for new exosome isolation designs. Then, we briefly introduce the conventional methods and discuss how their failure to meet those performance standards sparks an intense interest in microfluidic device innovations. The essence of this review is to lead an in-depth discussion on not only the technicality of those microfluidic platforms, but also their strengths and weaknesses with regards to the performance parameters set forth. To close the conversation, we call for the inclusion of exosome confirmation and contamination evaluation as part of future device development and performance assessment process, so that collectively, efforts towards microfluidics and nanotechnology for exosome isolation and analysis may soon see the light of real-world applications.

With the development of technology, data is becoming an extremely valuable resource. Data is being created, analyzed and used in a massive scale in every modern system. As a result, data analysis and data mining are very essential in each aspect of social applications. The value of data will be more useful if it can be linked and merged with other different data resources, especially for solving current social problems. In order to make a reality of this big challenge, data transformation is a crucial step that we have to overcome. Among data conversion methods, the intermediate method is to be a preferable one due to its expandable ability. However, for this type of model, finding the intermediate data type for the system is the crucial problem that needs to be solved firstly. In this paper, we will make comparisons among some popular data formats to find the appropriate data type for this elastic data conversion system.

This paper proposes a continuous fuzzy terminal sliding mode control (C-F-TSMC) for model-free second-order nonlinear systems. This controller inherited the advantages of three techniques: fuzzy logic system, terminal sliding mode and super-twisting algorithm (STA). First, a T-S fuzzy logic system is used to model the unknown second-order nonlinear systems. Then, a terminal sliding mode controller based on STA is implemented to achieve the highest tracking performance. The proposed controller provides the superior control properties such as robustness, high tracking accuracy, finite-time convergence, and less chattering phenomenon. Moreover, it is designed without the system model. The theory analysis is accomplished to prove the stability and finite-time convergence of the system. The proposed C-F-TSMC is applied to a serial rigid link robot manipulator. Simulation results are presented to validate the analysis.

Biomedical Named Entities are complex, so approximate matching has been used to improve entity coverage. However, the usual approximate matching approach fetches only one matching result, which is often noisy. In this work, we propose a method for biomedical NER that fetches multiple approximate matches for a given phrase to leverage their variations to estimate entity-likeness. The model uses pooling to discard the unnecessary information from the noisy matching results, and learn the entity-likeness of the phrase with multiple approximate matches. Experimental results on three benchmark datasets from the biomedical domain, BC2GM, NCBI-disease, and BC4CHEMD, demonstrate the effectiveness. Our model improves the average by up to +0.21 points compared to a BioBERT-based NER.

Recently, Structural Health Monitoring (SHM) has emerged to be one of the most effective tools for the diagnosis of damages in structures. Early identification and localization of damage not only help to reduce the maintenance cost but also extend the life cycle of the structures. In this paper, a novel approach using Artificial Neural Networks (ANNs) combined with Genetic Algorithms (GA) is proposed to increase the capacity of damage detection in SHM system. ANNs can make use of different algorithms such as recognition algorithms and regression algorithms to classify, detect, localize and evaluate the severity of the damage. Meanwhile, GA can be applied to identify training parameters as well as to solve the local minima problems from ANNs. To demonstrate the method, an analysis of a bridge is performed. Finite Element (FE) model of the bridge is created using measured vibration data and it is employed as training data for the combined ANN-GA method in the model updating process. The updated model will then be used as a baseline model for damage identification. The result shows that the proposed ANN-GA algorithm provides a high level of accuracy and efficiency in detecting damage in the considered structure.

Trabajo presentado en el International Conference on Robotics and Automation (ICRA), celebrado de forma híbrida (virtual y presencial), desde Xian (China), del 30 de mayo al 5 de junio de 2021, While there exists many methods for manipulating rigid objects with parallel-jaw grippers, grasping with multi- finger robotic hands remains a quite unexplored research topic. Reasoning and planning collision-free trajectories on the additional degrees of freedom of several fingers represents an important challenge that, so far, involves computationally costly and slow processes. In this work, we present Multi-FinGAN, a fast generative multi-finger grasp sampling method that synthesizes high quality grasps directly from RGB-D images in about a second. We achieve this by training in an end-to-end fashion a coarse-to-fine model composed of a classification network that distinguishes grasp types according to a specific taxonomy and a refinement network that produces refined grasp poses and joint angles. We experimentally validate and benchmark our method against a standard grasp-sampling method on 790 grasps in simulation and 20 grasps on a real Franka Emika Panda. All experimental results using our method show consistent improvements both in terms of grasp quality metrics and grasp success rate. Remarkably, our approach is up to 20-30 times faster than the baseline, a significant improvement that opens the door to feedback-based grasp re-planning and task informative grasping. Code is available at https://irobotics.aalto.fi/multi-fingan/., This work was supported in part by the Academy of Finland Strategic Research Council grant 314180, CHIST-ERA project IPALM, Finland 326304, Spain PCI2019-103386, and by the Spanish government with projects HuMoUR TIN2017-90086-R and Maria de Maeztu Seal of Excellence MDM-2016-0656

STEM education is the modern educational method. The knowledge of programming is very important for studying in STEM education, especially in information technology. In this paper, an intelligent search engine for the knowledge of the course about Introduction to Programming is constructed. The knowledge model is organized based on the improved ontology of Rela-model, which represents concepts of the course, relations between those concepts and some inference rules to query the knowledge of programming. Based on this ontology, the problems for querying the knowledge of the course are proposed based on their semantic. The search system has been tested in the real- world by students studying the course. The experimental results are positive for the effectiveness of the system.

Visual Question Answering (VQA), the novel task among the intersection between Computer Vision (CV) and Natural Language Processing (NLP), extracts answers from features of both questions and images. The current approaches in VQA rely on the combination between convolution and recurrent networks, which leads to the huge number of parameters for learning phase. With the success of employing pre-trained models, we integrate BERT [1] for embedding text and two models: ResNets [2] and VGG [3] for embedding image. In addition, we also propose to take advantages of fine-tuning techniques and stacked attention mechanism to combine textual and visual features in a novel learning phase considered its ability to reduce the size of models. To demonstrate our model’s performance, we conduct experiments in the VizWiz VQA Challenge 2020. According to the experimental results, the proposed approach outperforms existing methods for Yes-No questions on VizWiz VQA dataset

We present in this paper a combined technique of long short-term memory and hidden Markov model to prediction problems of crack propagation in engineering. The primary advantage of the hidden Markov model is that the ability to learn with less information, in other words, its future states do not depend on past ones, based only on the present state. We use long short-term memory to train data, and output consequences improved by adding predicted different changes that are computed by hidden Markov model. Applying this combined method to numerical examples of forecasting crack propagation of singled-edge-notched beam forced by 4-point shear, crack-height growth in Marcellus shale under the hydraulic fracturing and deformations of dam structures made from fiber reinforced concrete material is addressed. The tests were carried out with many different sizes of experimental data. It was found that a combined long short-term memory - hidden Markov model results in more accurate solution than only using long short-term memory, especially in the case of the dataset that is lack of information.

Physical layer security is a suitable replacement or complement to the traditional cryptography method for securing a wireless network. Physical layer security exploits the random characteristics of a wireless channel. Deploying multiple relays and a jammer can secure the system performance, however, extra power is required. RF energy harvesting is a promising solution to a wireless network’s energy scarcity problem. RF energy is widely available in the environment due to billions of devices broadcasting it. Therefore, a secure, green system model is proposed with a source, destination, energy harvested relaying nodes, energy harvested jammer, power beacon and an eavesdropper. The theoretical analysis and simulations are provided to evaluate the system performance. The results show that when the number of relays increases, the secrecy outage probability of the considered system decreases. In addition, the system with a jammer outperforms that without a jammer, especially in high SNR scenarios. Finally, the energy harvesting time fraction is an important parameter and needs to be designed carefully in a system.

Coresets can be described as a compact subset such that models trained on coresets will also provide a good fit with models trained on full data set. Using coresets, we can scale down a big data to a tiny one to reduce the computational cost of a machine learning problem. In recent years, data scientists have investigated various methods to create coresets with many techniques and approaches, especially for solving the problem of clustering large datasets. In this paper, we make comparisons among four state-of-the-art algorithms: ProTraS by Ros and Guillaume with improvements, Lightweight Coreset by Bachem et al. Adaptive Sampling Coreset by Feldman et al. and a native Farthest-First-Traversal-based coreset construction. We briefly introduce these four algorithms and compare their performances to find out the benefits and drawbacks of each one.

The farthest-first-traversal (fft) algorithm originally was used by Rosenkrantz et al. in an analysis of heuristics for the traveling salesman problem. This algorithm has been extensively studied for several sampling techniques. In this work, we present a modification of ProTraS algorithm given by Ros and Guillaume, which is also based on the fft algorithm, for sampling datasets for both k-means and k-median clustering algorithms. Unlike ProTraS, proposed algorithm takes the size of samples as an input. The algorithm is implemented in the Spark platform and tested for benchmark datasets. We also estimate the algorithm by comparing with the adaptive sampling and lightweight coreset algorithms, using the adjust Rand index.

In this work, a new robotic calibration method is proposed for reducing the positional errors of the robot manipulator. First, geometric errors of a robot are identified by using a conventional kinematic calibration model of the robot. Then, a radial basis function is constructed for compensating the compliance errors based on the effective torques for further increasing the positional precision of the robot. The enhanced positional accuracy of the robot manipulator in experimental studies that are carried on a YS100 robot illustrates the advantages of the suggested algorithm than the other techniques.

Web Application Firewall (WAF) is a security technology that helps mitigate and prevent common attacks in web applications. Designed as a barrier between the web server and the user (possibly an attacker), the WAF analyzes, detects, and early warns of possible attacks. However, the current techniques for detecting web application attacks are still mainly based on available attack signatures. This method has the advantage of being simple and easy to apply in most cases, a typical implementation of this method is ModSecurity (ModSecurity: open source web application firewall, https://modsecurity.org.) with ModSecurity CRS (ModSecurity CRS: a set of generic attack detection rules for use with ModSecurity, https://coreruleset.org.). The main weakness of this approach is that it is unable to detect new attack vectors and mistakenly detects a normal user as an attacker also. In response to these issues, the goal of this paper is to develop a WAF system based on ModSecurity with ModSecurity CRS, which focuses on reducing the false-positive rate of ModSecurity CRS based on machine learning methods. Specifically, this is a method that combines ModSecurity CRS and machine learning.

Object Tracking (OT) on a Moving Camera so-called Moving Object Tracking (MOT) is extremely vital in Computer Vision. While other conventional tracking methods based on fixed camera can only track the objects in its range, a moving camera can tackle this issue by following the objects. Moreover, single tracker is used widely to track object but it is not effective due to the moving camera because the challenges such as sudden movements, blurring, pose variation. The paper proposes a method inherited by tracking by detection approach. It integrates a single tracker with object detection method. The proposed tracking system can track object efficiency and effectively because object detection method can be used to find the tracked object again if the single tracker loses track. Three main contributions are presented in the paper as follow. First, the proposed Unified Visual based-MOT system can do the tasks such as Localization, 3D Environment Reconstruction and Tracking based on Stereo Camera and Inertial Measurement Unit (IMU). Second, it takes into account camera motion and the moving objects to improve the precision rate in localization and tracking. Third, proposed tracking system based on integration of single tracker as Deep Particle Filter and Object Detection as Yolov3. The overall system is tested on the dataset KITTI 2012, and it has achieved a good accuracy rate in real time.

Optical Character Recognition (OCR) for scanned paper invoices is very challenging due to the variability of 19 invoice layouts, different information fields, large data tables, and low scanning quality. In this case, table structure recognition is a critical task in which all rows, columns, and cells must be accurately positioned and extracted. Existing methods such as DeepDeSRT, TableNet only dealt with high-quality born-digital images (e.g., PDF) with low noise and apparent table structure. This paper proposes an efficient method called CluSTi (Clustering method for recognition of the Structure of Tables in invoice scanned Images). The contributions of CluSTi are three-fold. Firstly, it removes heavy noises in the table images using a clustering algorithm. Secondly, it extracts all text boxes using state-of-the-art text recognition. Thirdly, based on the horizontal and vertical clustering algorithm with optimized parameters, CluSTi groups the text boxes into their correct rows and columns, respectively. The method was evaluated on three datasets: i) 397 public scanned images; ii) 193 PDF document images from ICDAR 2013 competition dataset; and iii) 281 PDF document images from ICDAR 2019’s numeric tables. The evaluation results showed that CluSTi achieved an \(\textit{F}_1\textit{-score}\) of 87.5%, 98.5%, and 94.5%, respectively. Our method also outperformed DeepDeSRT with an \(\textit{F}_1\textit{-score}\) of 91.44% on only 34 images from the ICDAR 2013 competition dataset. To the best of our knowledge, CluSTi is the first method to tackle the table structure recognition problem on scanned images.

We propose a novel statistical approach to evaluate the statistical significance (reliability) of the findings in the discriminative sub-trajectory mining problem, called Statistically Discriminative Sub-trajectory Mining (Stat-DSM). Given two groups of trajectories, the goal is to extract moving patterns in the form of sub-trajectories that occur statistically significantly more often in one group than in the other. An advantage of the Stat-DSM method is that the statistical significance of the extracted sub-trajectories are properly controlled in the sense that the probability of finding a false discriminative sub-trajectory is smaller than a specified significance threshold a (e.g., 0.05). We conduct experiments on real-world datasets to demonstrate the effectiveness of the Stat-DSM method.

This study is concerned with outage performance analysis of a cognitive two-way relay system having a primary network of one transmitter and multiple receivers over Nakagami- m fading channels. In the considered cognitive radio system, the secondary network facilitates information exchange between source nodes with the aid of multiple decode-and-forward two-way relays. In the presence of multiple relays, opportunistic relay selection is employed to obtain a relay with the best channel quality for reliable information recovery. To analyse the performance of the considered cognitive radio system, exact and asymptotic closed-form expressions of outage probability (OP) are derived in this study. The derived theoretical OP expressions are in good agreement with related empirical values which show that using relays can enhance the performance of secondary networks without sacrificing that of the primary network.

This study focuses on solving the vehicle routing problem (VRP) of E-logistics service providers. In our problem, each vehicle must visit some pick up nodes first, for instance, warehouses to pick up the orders then makes deliveries for customers in the list. Each pickup node has its own list of more than one customers requiring delivery. The objective is to minimize the total travelling cost while real-world application constraints, such as heterogeneous vehicles, capacity limits, time window, driver working duration, etc. are still considered. This research firstly proposes a mathematical model for this multiple pickup and multiple delivery vehicle routing problem with time window and heterogeneous fleets (MPMDVRPTWHF). In the next step, the ant colony optimization algorithm is studied to solve the problem in the large-scale.

Cardiovascular diseases (CVDs) are the leading cause of death globally. For an effective treatment of CVDs, automatic CVDs detection based on Electrocardiograph (ECG) monitoring is highly desirable. One major challenge in ECG classification is feature learning. This paper reviews developed techniques for feature extraction and compared their performances in Atrial Fibrillation (AF), Myocardial Infarction (MI) and Pericarditis detection. Feature extraction methods in the literature reviews can be divided into mainly four categories: linear feature, nonlinear feature, wavelet transform, deep learning. Three studies were implemented using database from PhysioNet to evaluate the effectiveness of different feature extraction techniques. The AF detection algorithm use morphology features, statistic features, spectral feature, and wavelet entropy, presented a sensitivity of 96%, specificity of 93% and accuracy of 94.1%. In the case of Pericarditis and MI classification, information thoery-based features including subband energy, permutation entropy, and approximate entropy are the most noteworthy features. The study of detecting MI using machine learning based model of Convolutional Neural Network showed a sensitivity of 92.04% which yield the most promising results.

In this paper, we propose a non-orthogonal multiple access (NOMA) relaying system, where a source node communicates simultaneously with multiple users via the assistance of the best amplify-and-forward (AF) relay. The best relay is selected among N relays which are capable of harvesting the energy from radio frequency (RF) signals. We analyze the performance of the proposed NOMA relaying system in the conditions of imperfect channel state information (CSI) and Rayleigh fading by deriving the exact expressions of the outage probability (OP) and the approximate expression of the ergodic capacities of each user and the whole system. We also determine the optimal energy harvesting duration which minimizes the OP. Numerical results show that, for the same parameter settings, the performance of the proposed NOMA relaying system, especially the ergodic capacity of the whole system, outperforms that of the orthogonal-multiple-access (OMA) relaying system. Monte-Carlo simulations are used to validate the correctness of the analytical results.

In this paper, we analyze the energy efficiency (EE) of a non-orthogonal multiple access channel with successive interference cancellation for the uplink under Gaussian-mixture aggregate interference, which is a realistic noise plus interference model to capture the asynchronism in a heterogeneous cellular network. The considered EE is measured via the minimum energy per bit Eb/N0 min for reliable communication and the wideband slope of the spectral efficiency as a function of energy per bit Eb/N0. For completeness, both Gaussian signaling schemes and practical finite alphabet inputs are examined. To this end, our approach is to calculate Eb/N0 min and the wideband slope region via the Kullback-Leibler divergence. Due to the presence of mixture of multiple Gaussian distributions, the Kullback-Leibler divergence cannot be expressed in closed-form. As an alternative, we exploit upper bounds on the divergence, and we show that the bounds are achievable at the limit points, i.e., when the signal-to-noise-ratio (SNR) approaches zero. As a result, Eb/N0 min of each user can be established in closed-form. Next, we apply Taylor series expressions and exploit the achievability in the previous step to evaluate the second derivative of the rates when SNR goes to zero. It is then shown that the wideband slope region can be found effectively. The proposed method can serve as an important tool for making more accurate throughput and EE evaluations of important wireless underlay networks in future cellular systems.

Introduced by Vaswani et al., transformer architecture, with the effective use of self-attention mechanism, has shown outstanding performance in translating sequence of text from one language to another. In this paper, we conduct experiments using the self-attention in converting an abstract meaning representation (AMR) graph, a semantic representation, into a natural language sentence, also known as the task of AMR-to-text generation. On the benchmark dataset for this task, we obtain promising results comparing to existing deep learning methods in the literature.

Blockchain has gradually been popularized by its transparency, fairness, and democracy. This technology has opened the door to the development of Ethereum, a blockchain platform with smart contracts that can hold and automatically transfer tokens. Like a legacy computer program, smart contracts are vulnerable to security bugs. In recent years, many successful attacks on Ethereum network have been recorded, cost victims millions of dollars. In this paper, we classify attack vectors of Ethereum smart contracts, then propose some behaviour-based methods to detect them. To realize the ideas, we implement Abbe, a tool that can not only discover known attacks but also detect zero-day vulnerabilities.

Though the kinematic parameters had been well identified, there are still existing some non-negligible non-geometric error sources such as friction, gear backlash, gear transmission, temperature variation etc. They need to be eliminated to further improve the accuracy of the robotic system. In this paper, a new hybrid calibration method for improving the absolute positioning accuracy of robot manipulators is proposed. The geometric errors and joint deflection errors are simultaneously calibrated by robot model identification technique and a radial basis function neural network is applied for compensating the robot positions errors, which are caused by the non-geometric error sources. A real implementation was performed with Hyundai HH800 robot and a laser tracker to demonstrate the effectiveness of the proposed method.

Coresets can be described as a compact subset such that models trained on coresets will also provide a good fit with models trained on full data set. By using coresets, we can scale down a big data to a tiny one in order to reduce the computational cost of a machine learning problem. In recent years, data scientists have investigated various methods to create coresets. The two state-of-the-art algorithms have been proposed in 2018 are ProTraS by Ros & Guillaume and Lightweight Coreset by Bachem et al. In this paper, we briefly introduce these two algorithms and make a comparison between them to find out the benefits and drawbacks of each one.

Safe yet stable grasping requires a robotic hand to apply sufficient force on the object to immobilize it while keeping it from getting damaged. Soft robotic hands have been proposed for safe grasping due to their passive compliance, but even such a hand can crush objects if the applied force is too high. Thus for safe grasping, regulating the grasping force is of uttermost importance even with soft hands. In this work, we present a force controlled soft hand and use it to achieve safe grasping. To this end, resistive force and bend sensors are integrated in a soft hand, and a data-driven calibration method is proposed to estimate contact interaction forces. Given the force readings, the pneumatic pressures are regulated using a proportional-integral controller to achieve desired force. The controller is experimentally evaluated and benchmarked by grasping easily deformable objects such as plastic and paper cups without neither dropping nor deforming them. Together, the results demonstrate that our force controlled soft hand can grasp deformable objects in a safe yet stable manner., Comment: Accepted to 2020 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC 2020)

Gaussian process (GP) is a flexible kernel-based learning method which has found widespread applications in signal processing. In this paper, a supervised approach is proposed to handle single-channel speech separation (SCSS) problem. We focus on modeling a nonlinear mapping between mixed and clean speeches based on GP regression, in which reconstructed audio signal is estimated by the predictive mean of GP model. The nonlinear conjugate gradient method was utilized to perform the hyper-parameter optimization. The experiment on a subset of TIMIT speech dataset is carried out to confirm the validity of the proposed approach.

The development of IoT brings many opportunities but also many challenges. Recently, increasingly more malware has appeared to target IoT devices. Machine learning is one of the typical techniques used in the detection of malware. In this paper, we survey three approaches for IoT malware detection based on the application of convolutional neural networks on different data representations including sequences, images, and assembly code. The comparison was conducted on the task of distinguishing malware from nonmalware. We also analyze the results to assess the pros/cons of each method.