Fast Processing Time Research Articles

Classification methods performance on human activity recognition

There are around 650 million people from all over the world who lived with disabilities. One of the fundamental rights of people with disabilities is the existence of a companion to supervise his activity. Meanwhile, the use of mobile phones for monitoring the activities of people with disabilities has been widely carried out. The human activity monitoring mobile application requires human activity recognition methods that provide high accuracy, precision, and recall to reduce the error rate of the activity estimation. Some researches use machine learning algorithms like K-Nearest Neighbour (k-NN) algorithm, Artificial Neural Networks (ANN), Support Vector Machine (SVM), and Random Forest for human activity recognition methods. However, the results of these studies have not been compared apples to apples. Therefore, this study presents a performance comparison of SVM, KNN, and Random Forest machine learning methods. Based on our findings, the SVM method with Support Vector Classifier (SVC) and Radial Basis Function (RBF) kernels can achieve the highest precision and recall, 87% and 85% respectively. The fastest processing time is obtained using the SVM method with the Stochastic Gradient Descent. However, in general, the best performance is shown by Random Forest. The Random Forest method with a depth of 100 and 300 trees can reach an accuracy of 96% within 0.45 minutes.

Epigenetic Algorithm-Based Detection Technique for Network Attacks

Nowadays, the cybersecurity issue involves new strategies to protect against advanced threats and unknown attacks. Intrusion detection system (IDS) is considered a robust system dealing with attacks detection, particularly unknown attacks and anomalies. Several IDS-based algorithms have been recently inspected in the literature, among them the well-known strengthen algorithms, i.e. Genetic algorithm (GA). Moreover, Epigenetic-based algorithm (EGA) is known as an improved version of GA ensuring high performance with reduced computational complexity. Its main goal is to converge within a short time towards an optimal solution by acting on genetic operators, namely mutation and crossover. In this article, we propose a new classifier based on EGA for IDS. Especially, based on a database of network traffics, EGA is applied to classify attacks. The results, performed through EGA simulation, show that the performance of the proposed technique outperforms the ones of GA classifier by obtaining a high detection rate up to 98% and a faster processing time than that of GA and other algorithms that we have compared in this article.

Lightweight Multilayer Random Forests for Monitoring Driver Emotional Status

This study proposes a lightweight multilayer random forest (LMRF) model, which is a non-neural network style deep model consisting of layer-by-layer random forests. Although a deep neural network (DNN) is a powerful algorithm for facial expression recognition (FER), the requirement of too many parameters, careful parameter tuning, a huge amount of training data, black-box models, and a pre-trained architecture remain significant burdens for a current DNN, particularly for real-time processing. To overcome the limitations of a DNN, our FER system uses LMRF consisting of a two-layer structure and a small number of trees per layer for fast FER. The proposed LMRF achieves a similar performance as a DNN even with a small number of hyper-parameters, and a faster processing time using a CPU. We conducted experiments using a benchmark database captured indoors and a real driving database captured using a near-infrared (NIR) camera. Based on a performance evaluation against a few other state-of-the-art FER methods, the proposed method showed a more uniform performance than DNN-based methods, and required a reduced number of parameters and operations without a loss of accuracy when compared to DNN model compression algorithms. As a replacement for deeper and wider networks, the proposed model can be embedded in low-power and low-memory in vehicle systems for the monitoring of a driver's emotion.

Mining Integrated Sequential Patterns From Multiple Databases

Existing work on multiple databases (MDBs) sequential pattern mining cannot mine frequent sequences to answer exact and historical queries from MDBs having different table structures. This article proposes the transaction id frequent sequence pattern (TidFSeq) algorithm to handle the difficult problem of mining frequent sequences from diverse MDBs. The TidFSeq algorithm transforms candidate 1-sequences to get transaction subsequences where candidate 1-sequences occurred as (1-sequence, itssubsequenceidlist) tuple or (1-sequence, position id list). Subsequent frequent i-sequences are computed using the counts of the sequence ids in each candidate i-sequence position id list tuples. An extended version of the general sequential pattern (GSP)-like candidate generates and a frequency count approach is used for computing supports of itemset (I-step) and separate (S-step) sequences without repeated database scans but with transaction ids. Generated patterns answer complex queries from MDBs. The TidFSeq algorithm has a faster processing time than existing algorithms.

Progressive Feature Matching: Incremental Graph Construction and Optimization

We present a novel feature matching algorithm that systematically utilizes the geometric properties of image features such as position, scale, and orientation, in addition to the conventional descriptor vectors. In challenging scenes, in which repetitive structures and large view changes are present, it is difficult to find correct correspondences using conventional approaches that only use descriptors, as the descriptor distances of correct matches may not be the least among the candidates. The feature matching problem is formulated as a Markov random field (MRF) that uses descriptor distances and relative geometric similarities together. Assuming that the layout of the nearby features does not considerably change, we propose the bidirectional transfer measure to gauge the geometric consistency between the pairs of feature correspondences. The unmatched features are explicitly modeled in the MRF to minimize their negative impact. Instead of solving the MRF on the entire features at once, we start with a small set of confident feature matches, and then progressively expand the MRF with the remaining candidate matches. The proposed progressive approach yields better feature matching performance and faster processing time. Experimental results show that the proposed algorithm provides better feature correspondences in many challenging scenes, i.e., more matches with higher inlier ratio and lower computational cost than those of the state-of-the-art algorithms. The source code of our implementation is open to the public.

Non-Dimensional Feature Engineering and Data-Driven Modeling for Microchannel Reactor Control

Catalytic plate microchannel reactors (CPRs) are a promising means for modular hydrogen/fuels production from distributed natural gas resources. However, the equipment miniaturization presents challenges for process control, including spatially-distributed models, limited availability of measurements, and fast process time constants. In the present paper, we investigate the use of data-driven models—specifically, artificial neural networks (ANNs)—to estimate temperature “hotspots” within CPRs. We prescribe nonlinear transformations of the model inputs in the form of well-known dimensionless quantities (e.g., Reynolds number), and we show that these engineered features can improve the prediction capability of computationally parsimonious ANNs using a first-principles reactor model. Finally, we present a simulation case study that demonstrates the use of a trained ANN for inferential model predictive control.

Fingerprints Classification through Image Analysis and Machine Learning Method

The system that automatically identifies the anthropometric fingerprint is one of the systems that interact directly with the user, which every day will be provided with a diverse database. This requires the system to be optimized to handle the process to meet the needs of users such as fast processing time, almost absolute accuracy, no errors in the real process. Therefore, in this paper, we propose the application of machine learning methods to develop fingerprint classification algorithms based on the singularity feature. The goal of the paper is to reduce the number of comparisons in automatic fingerprint recognition systems with large databases. The combination of using computer vision algorithms in the image pre-processing stage increases the calculation time, improves the quality of the input images, making the process of feature extraction highly effective and the classification process fast and accurate. The classification results on 3 datasets with the criteria for Precision, Recall, Accuracy evaluation and ROC analysis of algorithms show that the Random Forest (RF) algorithm has the best accuracy (≥96.75%) on all 3 databases, Support Vector Machine (SVM) has the best results (≥95.5%) 2 / 3 databases.

Analysis of Frequent Itemsets Mining Algorithm Againts Models of Different Datasets

Data mining is a study that uses statistical knowledge, mathematical calculations, artificial intelligence methods, machine learning by extracting and identifying useful information and related knowledge from various large databases.. One of them is looking for itemsets combination from the data stack, the search process can be done using the Apriori Association Rules algorithm, the FPGrowth Association Rule and Closed Association rule. The three algorithms are some of the implementations of frequent itemsets search methods. Two datasets will be tested, namely retail datasets and accidents datasets derived from fimi datasets, with the aim of knowing the behavior of the three algorithms against the dataset model we tested. Each algorithm will test both datasets, with minimum support values for retail datasets ranging from 0.01-0.05 and min-conf 0.01-0.05. Likewise on the accident min-sup and min-conf 0.6-1 datasets. For retail algorithm data with the fastest processing time, the FP-Growth AR algorithm is more than 85% compared to Apriori-AR, followed by AR-Apriori and Apriori-Close-AR. For detailed memory usage results the Apriori-AR algorithm is lighter and outperforms. Accident data the experimental results show that FP-Growth-AR is a little more memory efficient and the fastest process is Apriori-Close, while the Apriori-AR with the longest processing time.

Characteristic of phenol and 2,4-dichlorophenol synthetic wastewater degradation in a DBD (dielectric barrier discharge) reactor

In this study the content of phenol and 2,4-DCP (2,4-dichlorophenol) in synthetic wastewater was decomposed using the excitation technique of a mixture of waste liquid and air in a cold plasma Dielectric Barrier Discharge (DBD) reactor. The purpose of this study was to study the degradation process of organic compounds of phenol and 2,4-DCP liquid waste into simpler compounds. Plasma technology applied to the liquid-air mixture in the DBD reactor has the ability to oxidize and degrade organic synthetic wastewater into simpler compounds with relatively faster processing times without forming new waste compounds. Plasma air (excited air) will degrade wastewater by breaking the atomic bonds of synthetic wastewater compounds at high voltages between 220 - 330 V. The performance of the degradation process of synthetic waste can be known through analysis of phenolic compounds, hydroxyl and dissolved ozone, COD, and the final product. The parameters studied in this study are air gas flow rate 2 - 2.5 L/min, waste water flow rate 52 - 100 mL/min, plasma voltage between 220 - 330 V, and volume of waste water. This study also compared the degradation process of phenol and 2,4-DCP in cold plasma DBD reactor with the degradation process in multi-injection bubble column reactor. The removal efficiency of the process achieved 57.5% for phenol and 89.55% for 2,4-DCP.

Automated Detection of Intracranial Large Vessel Occlusions on Computed Tomography Angiography: A Single Center Experience.

Background and Purpose- Endovascular thrombectomy is highly effective in acute ischemic stroke patients with an anterior circulation large vessel occlusion (LVO), decreasing morbidity and mortality. Accurate and prompt identification of LVOs is imperative because these patients have large volumes of tissue that are at risk of infarction without timely reperfusion, and the treatment window is limited to 24 hours. We assessed the accuracy and speed of a commercially available fully automated LVO-detection tool in a cohort of patients presenting to a regional hospital with suspected stroke. Methods- Consecutive patients who underwent multimodal computed tomography with thin-slice computed tomography angiography between January 1, 2017 and December 31, 2018 for suspected acute ischemic stroke within 24 hours of onset were retrospectively identified. The multimodal computed tomographies were assessed by 2 neuroradiologists in consensus for the presence of an intracranial anterior circulation LVO or M2-segment middle cerebral artery occlusion (the reference standard). The patients' computed tomography angiographies were then processed using an automated LVO-detection algorithm (RAPID CTA). Receiver-operating characteristic analysis was used to determine sensitivity, specificity, and negative predictive value of the algorithm for detection of (1) an LVO and (2) either an LVO or M2-segment middle cerebral artery occlusion. Results- CTAs from 477 patients were analyzed (271 men and 206 women; median age, 71; IQR, 60-80). Median processing time was 158 seconds (IQR, 150-167 seconds). Seventy-eight patients had an anterior circulation LVO, and 28 had an isolated M2-segment middle cerebral artery occlusion. The sensitivity, negative predictive value, and specificity were 0.94, 0.98, and 0.76, respectively for detection of an intracranial LVO and 0.92, 0.97, and 0.81, respectively for detection of either an intracranial LVO or M2-segment middle cerebral artery occlusion. Conclusions- The fully automated algorithm had very high sensitivity and negative predictive value for LVO detection with fast processing times, suggesting that it can be used in the emergent setting as a screening tool to alert radiologists and expedite formal diagnosis.

Comparison of SMART and SAW Methods in Decision Making

In this study using a comparative analysis of SMART and SAW methods. The results of the SAW method calculation analysis have a faster processing time with a time of 0.40 s. While the SMART method calculation analysis has a longer processing time difference of 0.47 s. In this study the authors suggest using the SMART method. Where the process of calculating the SMART method has more varied results, because the criteria value between one candidate and another candidate influences each other. While the calculation of the SAW method results are less varied, because the criteria value between one candidate and another candidate does not affect each other. Weighting is done without an algorithm. In this study the authors suggest using the SMART method because there is a final value of <60 is not feasible given the provision of venture capital.

Deep learning‐based embedded license plate localisation system

In this study the authors propose novel neural network architecture for license plate localisation (LPL) based on an inverted residual structure where the shortcut connections are between the linear bottleneck layers. This residual structure is used for feature extraction in a modified single shot detector for object detection, where standard convolutions are replaced with depthwise separable convolutions in classification layers. The proposed deep learning (DL) solution was tested against three popular international research databases and achieves state-of-the-art results, proving that the proposed model is accurate and robust. Across those databases, the proposed model surpasses other recent LPL works, including DL-based methods, in terms of accuracy and speed. The authors show the proposed architecture to have significant speedup and computational efficiency gains over other DL models, and to have fast per-image localisation processing times sufficient for applications deployed on expensive and commodity hardware alike. Using a novel multi-threading video capture with motion detection then inference algorithm, the authors increase computational efficiency and drop fewer frames overall, allowing for increased performance. Repeated tests show that the proposed method is well-suited to real-time and highly accurate LPL, regardless of hardware.

Cancer driver genes: a guilty by resemblance doctrine.

A major benefit of expansive cancer genome projects is the discovery of new targets for drug treatment and development. To date, cancer driver genes have been primarily identified by methods based on gene mutation frequency. This approach fails to identify culpable genes that are not mutated, rarely mutated, or contribute to the development of rare forms of cancer. Due to the complexity of the disease and the sheer volume of data, computational methods may encounter a NP-complete problem. We have developed a novel pathway and reach (PAR) method that employs a guilty by resemblance approach to identify cancer driver genes that avoids the above problems. Essentially PAR sifts through a list of genes of biological pathways to find those that are common to the same pathways and possess a similar 2-reach topology metric as a reference set of recognized driver genes. This approach leads to faster processing times and eliminates any dependency on gene mutation frequency. Out of the three pathways, signal transduction, immune system, and gene expression, a set of 50 candidate driver genes were identified, 30 of which were new. The top five were HGF, E2F1, C6, MIF, and CDK2.

Label-free discrimination of single nucleotide changes in DNA by reflectometric interference Fourier transform spectroscopy

Phenotypic variation – such as disease susceptibility and differential drug response – has a strong genetic component. Substantial effort has therefore been made to identify causal genomic variants explaining such variation among humans. Point mutations (PMs), which are single nucleotide changes in the genome, have been identified to be the most abundant form of causal genomic variants, making them useful, reliable diagnostic markers. Methods developed to genotype PMs have moved towards solid-phase assays, which not only show greater sensitivity and specificity, but also enable scalability and faster processing time. Most current assays are, however, based on fluorescent probes, which makes them relatively expensive. To develop a more cost-effective label-free genotyping method, we used a porous silicon (PSi) base as an efficient support for DNA biosensing and coupled it with reflectometric interference Fourier transform spectroscopy (RIFTS). To assess the versatility of this approach, we tested both a single nucleotide substitution in VKORC1 (-1639G > A; rs9923231) and a single nucleotide insertion in BRCA1 (5382insC; rs80357906). We demonstrate that the PSi-RIFTS method can efficiently detect both PM types with high sensitivity where hybridization of complementary DNA can be quantifiably differentiated from mismatch and non-complementary hybridization events. In addition, we show that the PSi base with immobilized DNA not only can be re-used to type further samples, but it also remains stable for 14 days, suggesting its potential for high-throughput applications.

Characteristics study and robustness comparison of various star pattern recognition algorithm for star sensor using digital simulation

The last few years the use of star sensor, which is the most accurate sensor on satellites, began to penetrate into micro satellites and nano satellites, previously used only on high-end satellites. Hence we need to develop star sensor according to the specific needs of microsatellite and nano satellites. One of star sensor’s aspect is star pattern recognition algorithm. A good star pattern recognition algorithm must have a high level of matching success when there are a lot of disturbances (robust) and fast processing time. There are different types of star pattern recognition algorithms, which have their respective advantages and disadvantages, so it is necessary to characterize and compare the various algorithms. In this paper, the characterization and comparison of robustness in different types of star pattern recognition algorithm are done using digital simulation. The simulation results show that the success rate does not always rise with the number of stars used for matching because it depends on how the algorithm works. In addition, the more the number of stars are used, the longer processing time, and vice versa. The best algorithm in terms of success rate is an algorithm that uses a combination pattern of all possible triangles, in other words most robust, but has the longest processing time. This long processing time can be overcome by optimizing GSC (Guide Star Catalogue). From the results the multitriangles 2 algorithm has the best success rate compared to the other algorithms, with success rate for 5, 6, and 7 stars configuration are 97.8%, 98.7%, and 99.8% respectively.

Formula omitted]-SLAM: A fast RGB-D SLAM approach for large indoor environments

We present an RGB-D based SLAM system capable of building consistent 3D maps of large indoor environments. The system performs motion estimation by using keypoint correspondences between frames. A novel data association approach is developed for accurate mapping with robust loop closure detection. To achieve this, the proposed method uses a Keyframe Autocorrelogram Database and an adaptive thresholding technique. Keyframe Autocorrelogram Database indexes and clusters past frames based on their spatial color correlations and returns short and long loop closure candidates as k-sized clusters by using a data structure that can effectively handle high dimensional data. The adaptive thresholding technique, which was introduced in our previous work, filters out more outliers in the clusters and selects better loop closure candidates. The proposed approach incrementally produces environment maps without needing any training step. The system is tested on widely used public datasets with a large group of sequences. The experimental results show the robustness and efficiency of our system in challenging conditions, in comparison to other state-of-the-art systems. The system produces accurate maps for medium and large-scale environments with stable and fast processing times on CPU. Thus, the proposed approach makes possible real-time operation with efficient implementations.

Ti3SiC2-Cf composites by spark plasma sintering: Processing, microstructure and thermo-mechanical properties

MAX phases, and particularly Ti3SiC2, are interesting for high temperature applications. The addition of carbon fibers can be used to reduce the density and to modify the properties of the matrix. This work presents the densification and characterization of Ti3SiC2 based composites with short carbon fibers using a fast and simple fabrication approach: dry mixing and densification by Spark Plasma Sintering. Good densification level was obtained below 1400 °C even with a high amount of fibers. The reaction of the fibers with the matrix is limited thanks to the fast processing time and depends on the amount of fibers in the composite. Bending strength at room temperature, between 437 and 120 MPa, is in the range of conventional CMCs with short fibers and according to the resistance of the matrix and the presence of residual porosity. Thermo-mechanical properties of the composites up to 1500 °C are also presented.

Land use change simulation and analysis using a vector cellular automata (CA) model: A case study of Ipswich City, Queensland, Australia

The loss of accuracy in vector-raster conversion has always been an issue for land use change models, particularly for raster based Cellular Automata models. Here we describe a vector-based cellular automata (CA) model that uses land parcels as the basic unit of analysis, and compare its results with a raster CA model. Transition rules are calibrated using an artificial neural network (ANN) and historical land use data. Using Ipswich City in Queensland, Australia as the study area, the simulation results show that the vector and raster CA models achieve 96.64% and 93.88% producer’s spatial accuracy, respectively. In addition, the vector CA model achieves a higher kappa coefficient and more consistent frequency of misclassification, while also having faster processing times. Consequently, the vector-based CA model can be applied to explore regulations of land use transformation in urban growth process, and provide a better understanding of likely urban growth to inform city planners.

A review of the impact of utilising electronic medical records for clinical research recruitment.

Recruitment is an important aspect of clinical research, as poor recruitment could undermine the scientific value of a trial or delay the development process of new treatments. The development of electronic medical records provides a new way to identify potential participants for trials by matching the eligibility criteria with patients' data within electronic medical records. A literature search was performed to examine the effectiveness and efficiency of the electronic medical record recruitment method using MEDLINE, PubMed, PubMed Central, CINAHL Plus with Full Text, ScienceDirect and Cochrane Library databases. These searches generated 11 articles that met the eligibility criteria, and handsearching reference lists generated two additional articles bringing the total number of articles to 13. These articles were subjected to critical appraisal utilising the Effective Public Health Practice Project tool. Out of the 13 included articles, 11 provided quantitative data on recruitment effectiveness while seven articles provided quantitative data on recruitment efficiency. The automation in screening and patient identification by using alerts, a notification system, to notify research staff of a potential participant, was observed to contribute to higher recruitment yield and reduced workload due to its specificity on participant screening. The use of electronic medical record alerts was found to be associated with better recruitment outcomes when they were sent to dedicated research staff rather than physicians. Using electronic medical records for recruitment was found to be effective due to its capability for patient identification outside working hours and fast processing time, which was particularly useful for clinical trials in acute conditions. Several challenges may hinder the impact of the electronic medical record recruitment method, including the lack of conformity of clinical trial eligibility criteria and electronic medical record data structure and missing data. 'Alert fatigue' could also impact on the effectiveness of this method in the long term. The results from this review supports electronic medical record being an effective and efficient method for clinical trial recruitment. Recommendations were made in order to maximise the potential of the electronic medical record recruitment method and also for future research in order to improve the quality of evidence to support this strategy for recruitment.

Automating direct-to-PCR for disaster victim identification

ABSTRACTDirect-to-PCR methodology adds samples directly to PCR tubes offering gains in efficiency and sensitivity. The approach has been applied to a variety of biological sources including blood, saliva, tissue, hair and nail. We added various preservative solutions to a range of biological samples to leech DNA into solution, whilst preserving at room temperature. Tubes containing ‘free DNA’ then followed automated workflows for amplification and capillary electrophoresis. Routine FASS-automated workflows (including DNA extraction and quantification) were compared with published direct-to-PCR methodology and automated amplification of an aliquot of preservative solution. Applying preservative solutions to ~30-year-old blood stains stored at room temperature resulted in recovery of a larger quantity of DNA and more alleles (using PowerPlex 21) when compared with routine automated typing. Trials were extended to blood, saliva, hair and nail, mimicking ante-mortem samples collected in a disaster victim identification effort. Despite slightly lower allelic recovery, the faster processing times, lower costs and storage potential offers advantages for the processing of ante-mortem samples.