Article of the Year 2020
A Fortran-Keras Deep Learning Bridge for Scientific ComputingRead the full article
Scientific Programming provides a forum for research results in, and practical experience with, software engineering environments, tools, languages, and models of computation aimed specifically at supporting scientific and engineering computing.
Chief Editor Professor Tramontana is based at the University of Catania and his research primarily concerns the areas of software engineering and distributed systems.
Latest ArticlesMore articles
Analysis and Research on the Characteristics of Modern English Classroom Learners’ Concentration Based on Deep Learning
There are some problems in modern English education, such as difficulties in classroom teaching quality evaluation, lack of objective evaluation basis in teaching process management, and quality monitoring. The development of artificial intelligence technology provides a new idea for classroom teaching evaluation, but the existing classroom evaluation scheme based on artificial intelligence technology has a series of problems such as high system cost, low evaluation accuracy, and incomplete evaluation. In view of the above problems, this paper proposes a solution of English classroom concentration evaluation system based on deep learning. The program studies the evaluation methods of students’ class concentration, class activity, and enrichment degree of teaching links, and constructs an information evaluation system of students’ learning process and class teaching quality. Based on the edge computing system architecture, a hardware platform with cloud platform AI+ embedded visual edge computing devices managed by an FPGA deep learning accelerated server was built. The design, debugging, and testing of classroom evaluation and student behavior statistics-related functions were completed. This scheme uses edge computing hardware architecture to solve the problem of high system cost. Deep learning technology is used to solve the problem of low accuracy of classroom evaluation. It mainly evaluates the classroom objectively by extracting indicators such as the students' attention in the classroom, and solves the problems of the students’ inattentiveness in the classroom. After the test, the classroom evaluation system designed by the paper runs stably and all functions run normally. The test results show that the system can basically meet the requirements of classroom teaching evaluation application.
A Short Text Similarity Calculation Method Combining Semantic and Headword Attention Mechanism
Short text similarity computation plays an important role in various natural language processing tasks. Siamese neural networks are widely used in short text similarity calculation. However, due to the complexity of syntax and the correlation between words, siamese networks alone cannot achieve satisfactory results. Many studies show that the use of an attention mechanism will improve the impact of key features that can be utilized to measure sentence similarity. In this paper, a similarity calculation method is proposed which combines semantics and a headword attention mechanism. First, a BiGRU model is utilized to extract contextual information. After obtaining the headword set, the semantically enhanced representations of the two sentences are obtained through an attention mechanism and character splicing. Finally, we use a one-dimensional convolutional neural network to fuse the word embedding information with the contextual information. The experimental results on the ATEC and MSRP datasets show that the recall and F1 values of the proposed model are significantly improved through the introduction of the headword attention mechanism.
Detecting Anomaly Data for IoT Sensor Networks
The Internet of Things, or IoT, has been widely recognized as a new perception paradigm for interacting between the digital world and the physical one. Acting as the interface and integral part of the Internet of Things, sensors embedded within the network are the principal components that collect the unprocessed data, and these sensors are usually deployed in unattended, hostile, or harsh areas, which inevitably makes the sensor readings prone to faults and even anomalies. Therefore, the quality of sensor readings will ultimately affect the quality of various data-oriented IoT services, and the sensor data are of vital importance affecting the performance of the system. However, the data anomaly detection is a nontrivial task for IoT because sensors are usually resource-constrained devices with limited computing, communication, and capacity. Therefore, an efficient and lightweight detecting method is needed to meet the requirements. In this study, we deal with the anomaly data by detecting the source sensor nodes through combination methods of the local outlier factor and time series. Simulations show that the proposed method can effectively detect the anomaly data and presents a better normal data rate.
Research on Dynamic Programming Strategy of Bayesian Network Structure Learning
Bayesian network structure learning based on dynamic programming strategy can be used to find the optimal graph structure compared with approximate search methods. The traditional dynamic programming method for Bayesian network structure learning is a depth-first-based strategy, which is inefficient. We proposed two methods to solve this problem. First, the dependency constraints were used to prune the process of calculating redundancy scores. The constraints were obtained by the conditional independence test from the observed data sets. However, it was difficult to guarantee the accuracy of the constraints, which may have led to a decrease in the accuracy of the method. Second, we proposed a breadth-first-based strategy, which enhanced efficiency greatly while also ensuring global optimality. Experimental results showed that on the standard network data sets, compared with the dynamic programming based on depth-first search (DFSDP) algorithm, dynamic programming based on constraints (CBDP) could reduce the average running time by 57.10% and that dynamic programming based on breadth-first search (BFSDP) could reduce the average running time by 50.02%. On the UCI data sets, compared with DFSDP, CBDP reduced the average running time by 40.71%, and BFSDP reduced the average running time by 81.78%.
Simulation Analysis of Combined Mechanics of the Thrust Rotary Guide Drill
The force and deformation analysis of the bottom hole assembly (BHA) is the foundation and essential component of good trajectory control technology. Still, the BHA structure has complexity (multiangle), making it challenging to analyze the force and deformation of the BHA accurately. In this paper, we consider the factors such as well drilling parameters (drill pressure), drill combination structure parameters (multivariable cross section, stabilizer, and flexible short section diameter and position, etc.), a mechanical analysis model of the push-the-bit rotary steering tool is established based on the infinitesimal method and continuous beam-column theory. Taking the push-the-bit rotary steerable drilling assembly as an example, the influence of force and deformation of the rotary guided drilling assembly, such as drill pressure, stabilizer, and flexible short section parameters, was analyzed. The research results of this paper can solve the mechanical modeling problem of the variable cross section in the bottom drill combination, can deal with the combination of infinite multiple stiffnesses according to the actual situation, bring the analysis results closer to the real problem, and provide theoretical support for the design optimization of the bottom drill combination. In addition, edge computing can provide sufficient computing power for the calculation of this paper to ensure operational efficiency. It can realize the online real-time transmission of mechanical analysis results through powerful calculation examples and effectively guide the field operation.
Preparation of Nitrogen-Doped Carbon-Based Bimetallic Copper-Cobalt Catalysis Based on Deep Learning and Its Monitoring Application in Furfural Hydrogenation
In the field of catalysis, the support of the catalyst is often composed of hollow carbon materials. In order to monitor the preparation of nitrogen-doped carbon-based bimetallic copper-cobalt catalysis and its hydrogenation reaction in furfural, using m-aminophenol as the nitrogen source, formaldehyde as a carbon source, and P123 as a template agent, a nitrogen-doped bimetallic copper-cobalt mesoporous carbon catalyst Cu-Co@N-MPC-500 was synthesized by the hydrothermal method. The morphology, structure, and chemical composition of the catalyst were analyzed by means of TEM, XRD, BET, and XPS, respectively. The results show that the nitrogen-doped mesoporous carbon has a stable structure, uniform pore size distribution, and the nano-copper-cobalt particles are uniformly dispersed in the mesoporous carbon surface. Through furfural hydrogenation, the catalyst selectivity and cycle stability were discussed. Under the furfural conversion rate of 96.1%, the yield of cyclopentanone could reach 76.2%. After 5 cycles, the catalytic efficiency of the catalyst did not decrease significantly. It shows that Cu-Co@N-MPC-500 has excellent application prospects in the field of industrial production.