Search

The human element has been cited as the main contributory factor for various maritime navigational incidents. Measuring and assessing human performance is key to enhance and maintain maritime navigational safety. In order to address the above problem, there is a need to develop an efficient measurement and assessment tool for navigational competencies for various behaviors such as situational awareness, decision making, teamwork, and communication and influencing skills. Multiple forms of data is collected in a Navigation Bridge Simulator in a Bridge Team Scenario to conduct this research. The topic is especially relevant to improve the safety of maritime navigation in crowded waters such as the Singapore Strait. An effective AI-based competency assessment tool for safe navigation (AICATSAN) is developed to measure and assess navigational competence. Hybrid deep learning methods using eye trackers are employed to develop AICATSAN, which was assessed to produce accurate results with preliminary test data sample. Convolutional neural network (CNN) and long short-term memory (LSTM) networks are used for multi-modal, multi-feature and multisensory for recognition of complex and concurrent activity details. The AICATSAN is able to quickly analyze the human performance of a given navigational scenario. As an indicative assessment system to evaluate behavioral competence through the use of navigation simulator, this AICATSAN system will observe key indicators to analyze human behavioral competencies and facilitate the evaluation of navigational competence. It has the potential to be used for training, recruitment or promotion to the higher ranks.

Improving the performance of the network architectures that mimic brain operation is a research trend. Optimizing the latency on hardware circuits of the artificial neural network continues investigating. In the third generation, the Spiking Neural Networks (SNNs) with biological plausibility and similarity to the functionality of the human brain are emerging. A more comprehensive study is expected to understand the inherent behavior of SNNs, especially under adversarial attacks. This study concatenates the proposed SLIT layer with the convolutional neural networks (CNNs) to degrade the latency of deep neural networks on the hardware platform. The input data modified with the SLIT layer is applied to interrogate the adversarial attack on Spiking Neural Network. We estimate new topology with MNIST and CIFAR-10 datasets. Latency of the inference phase on CNNs for image classification application is assessed on the chip ZC7Z020-1CLG484C FPGA. Reducing latency in the range of 2.6% to 16% is observed from the Vitis AI platform. With white-box adversarial attack applications on SNNs, the accuracy of the proposal is approximately 70% higher robustness than the previous works.

In this day and age, supplying electricity for an entire nation during the Industry 4.0 era is not a simple mission. Therefore, the power system reliability must be one step ahead. Troubles related to the Distribution network reliability are considered the most important for they directly affect the customers. Many latest technologies are often applied to effectively operate this sort of network. The most difficult Distribution network reliability problem is its vastness and complexity that makes manual calculation seem impossible. However, thanks to the modernization of Electrical engineering software, these issues are now easily solved. This paper briefly introduces the Distribution network reliability and proposes a method to establish an optimal plan to improve the reliability by using the ETAP's Reliability function. The research applies ETAP in evaluating the reliability of an actual network to find out its deficiencies. Based on this, we can propose manners to improve this network reliability. Finally, we verify the effectiveness of these manners through economic indicators.

The recent introduction of pseudo-public space to Vietnam has significantly contributed to increasing the area of public spaces in urban area. Not only is coffee-shop space where people drink coffee and gossip as before, but now it is also a place for entertainment, relaxation and social interactions. Similarly to studies on such privately owned public spaces in foreign countries as pocket park, this research aims at understanding the usage nature of coffee-shop space in Vietnam’s cities, particularly Vinh Long city. The research also discovers the distribution of coffeeshop space network in urban configuration, types of coffee-shop space and their spatial properties on basis of data collected by observation, investigation, survey and Open Street Map. Furthermore, the research uses analytical tools deriving from space syntax theory. Findings will be detailed in this paper.

We present an approach to detecting small cell populations in flow cytometry (FCM) samples based on the combination of unsupervised manifold embedding and supervised random forest classification. Each sample consists of hundred thousands to a million cells where each cell typically corresponds to a measurement vector with 10 to 50 dimensions. The difficulty of the task is that clusters of measurement vectors formed in the data space according to standard clustering criteria often do not correspond to biologically meaningful sub-populations of cells, due to strong variations in shape and size of their distributions. In many cases the relevant population consists of less than 100 scattered events out of millions of events, where supervised approaches perform better than unsupervised clustering. The aim of this paper is to demonstrate that the performance of the standard supervised classifier can be improved significantly by combining it with a preceding unsupervised learning step involving the Uniform Manifold Approximation and Projection (UMAP). We present an experimental evaluation on FCM data from children suffering from Acute Lymphoblastic Leukemia (ALL) showing that the improvement particularly occurs in difficult samples where the size of the relevant population of leukemic cells is low in relation to other sub-populations. We show that the positive effect of the UMAP becomes more noticeable for smaller training sets. Further, the experiments indicate that in this situation the algorithm also outperforms other baseline methods based on Gaussian Mixture Models.

This study aims to prove the influence of wetness on the mechanical characteristic of carbon fiber–reinforced plastic (CFRP). There is no suitable method for using lubrication fluids in the composite industry. There is a significant impact on CFRP machining capabilities by the geometry of the cutting tool and cryogenic liquid cooling. This paper selects the optimal parametric combinations of the carbon fiber reinforced plastic drilling process by using Taguchi optimization technique base on grey relation analysis (GRA) technique (TGRA). The parameters of the drilling process such as the helical angle of the blade, the coolant gas, and the feed rate were found by observing multi-responses output such as surface roughness and diameter of the drill hole. The following parameters resulted in a larger value of grey relational grade (GRG) to reduce the surface roughness and the error deviation of the diameter dimension: the speed of the spindle motor was 6100 rpm, the feed rate was 1006 mm/min, the helical angle of the cutting blade was triple, the coolant gas was CO 2 -N 2 mixture. Finally, an analysis of variance (ANOVA) is used to analyze the effect of each parameter process in the estimation of GRG. The results showed that the coolant gas had an outstanding effect on the surface roughness and error deviation of the diameter dimension.

Reflecting various physical phenomena of the primary visual cortex (V1) in hierarchical approaches has become prevalent over the years when the semiconductor miniaturization reaches its limit. In this paper, we propose a V1 hardware model to extract features. Our architecture with the edge, slit, left-right parallax, XY movement direction, and approach detection functions are a partial mimic from V1. The fundamental of the model is the sum of absolute differences and the AND function of the slit angles, which are increased gradually every 22.5° in the range of 0° to 157.5°. The handwritten letter recognition application is the first application that is assessed by replacing the first convolutional layer with the slit detection function. Our model is evaluated on Xilinx Zynq-7020 FPGA by the Vivado HLS tool. We reduce 46% in latency and 30% hardware resources at a 97.34% accuracy with other works on the MNIST database. It is therefore suggested that the proposal demonstrates high-efficiency energy in the hardware architecture.

In recent years, the traceability of product origins is strongly concerned, particularly for food products as they directly influence human health. Therefore, there have been some efforts to develop product origins tracking systems. In this paper, we propose an approach to building a supply chain management system based on the blockchain technology for agriculture product origins tracking. The supply chain model is borrowed from Walmart's and it is implemented based on the Ethereum framework using the PoA (Proof of Authority) consensus algorithm. Our experiment shows that the proposed system not only fulfills the requirements of a product origins tracking but also takes the advantages of the blockchain technology such as the immutability and security of data, the low cost in making the transactions, and so on.

The oscillation mechanisms with large amplification rates, the ability of the body to respond to vibration is very important. Achieving a high amplitude of vibration with high frequency (Freq.) is the target to be achieved of amplification mechanisms. If the frequency of excitation is more than the own oscillation frequency of the structure, it will not meet the stable amplitude of the amplification. This study investigates the effect of the geometry parameter of the flexure hinge on the displacement and the frequency of the output of the bridge-type compliant. The modal function is used to analyze via finite element analysis (FEA) in Ansys. The result of the analysis as the frequency is optimized by the Taguchi method in Minitab to meet the best oscillation frequency. The result shows that the simulation model can reach stable at the amplitude 2.07mm with the operating frequency of 207.27 HZ.

this study explores the potential impacts of Automated Mobility on Demand on the long-term mobility choices using SimMobility, a multi-agent-based microsimulation platform. Synthetic population of Singapore in a near future is constructed and used as the input of simulation and different scenarios are simulated in terms of the availability of different transport modes. Results demonstrate how microsimulation could be utilized to evaluate future scenarios in urban transportation and, more specifically, long-term mobility choices such as residential and job locations as well as vehicle ownership.

Baseline detection is a simplified text-line extraction that typically serves as pre-processing for Automated Text Recognition. The cBAD competition benchmarks state-of-the-art baseline detection algorithms. It is the successor of cBAD 2017 with a larger dataset that contains more diverse document pages. The images together with the manually annotated groundtruth are made publicly available which allows other teams to benchmark and compare their methods. We could also evaluate the winning method of cBAD 2017 on the newly introduced dataset which now serves as baseline. This competition shows that the performance of automated baseline detection increased substantially since 2017.

Nowadays, science and technology and the industrial revolution 4.0 are growing rapidly. The field of object recognition has achieved significant result, and applied in many important tasks such as security monitoring, surveillance systems, autonomous systems, human-machine interaction and so on. The intelligent system based on deep learning technique is being used extensively in today’s life. A smart system brings to many benefits in living assistant systems. The contribution presents a solution based on combination of facial recognition and RFID (radio frequency identification) tags for the office checkup task in surveillance monitoring system (SMS). The SMS is constructed based on two main techniques to building intelligent systems which consist of face recognition technology and RFID tag recognition to monitor employee attendance when they are entering or leaving the office. In this system, the deep neural network is studied for face recognition. The system is connected to the SQL Server database to store the connection and ensure to synchronization is superior to the normal monitoring management systems.

In this paper we present a template-based table structure matching using association graphs for handwritten/printed historical documents. The recognition of the table structure consisting of column and header information is the prerequisite for the subsequent row detection and handwritten text recognition used for information extraction. The table matching is done by detecting the maximum clique in an association graph, which represents the matching of the line information of the template and a document of interest. This allows for variations of widths and heights of rows and columns. The presented methodology is evaluated on historical register books (death records) of the Archive of the Diocese of Passau. The method shows a reliable detection of the structure of handwritten/printed tables with a mean cell match of 88.28%.

Acute Myeloid Leukaemia (AML) is a rare type of childhood acute leukaemia. During treatment, the assessment of the number of cancer cells is particularly important to determine treatment response and consequently adapt the treatment scheme if necessary. Minimal Residual Disease (MRD) is a diagnostic measure based on Flow CytoMetry (FCM) data that captures the amount of blasts in a blood sample and is a clinical tool for planning patients' individual therapy, which requires reliable blast identification. In this work we propose a novel semi-supervised learning approach, which is acquired whenever large amounts of unlabeled data and only a small amount of annotated data is available. The proposed semi-supervised learning approach is based on Wasserstein Generative Adversarial Network (WGAN) latent space embeddings learned in an unsupervised fashion and a simple Fully connected Neural Network (FNN) trained on labeled data leveraging the learned embedding. We apply our proposed learning approach for semi-supervised classification of blasts vs. non-blasts. We compare our approach with two baseline approaches, 1) semi-supervised learning based on Principal Component Analysis (PCA) embedding, and 2) a deep FNN that is trained only on the annotated data without leveraging an embedding. Results suggest that our proposed semi-supervised WGAN embedding outperforms semi-supervised learning based on PCA embeddings and if only small amounts of annotated data is available it even outperforms an FNN classifier.