Machine Learning and Mechanics Based Soft Computing Applications

Text classification based on thresholds belongs to the supervised learning method which assigns text material to predefined classes or categories based on different thresholds with divergence approach. These categories are identified by a set of documents trained by an automated algorithm. This work presents an approach of text classification using an automatic keyword extraction algorithm based on the Kullback–Leibler divergence approach. The proposed method is evaluated on 2000 documents in Vietnamese, covering ten topics, collected from various e-journals and news portal Web sites including vietnamnet.vn, vnexpress.net, and so on to generate a completely new set of keywords. Such keywords, then, are leveraged to categorize the topic of new text documents. The obtained results verifying the practicality of our approach are feasible as well as outperform the state-of-the-art method.

Water quality is essential in aquaculture, particularly in intensive aquaculture (ponds, lakes). Aquatic animals can grow and thrive when water quality meets the standards and stability, or they will be mentally disabled or die when water is contaminated. Therefore, water’s chemical, biological, and physical indexes, such as dissolved oxygen, temperature, pH, must be monitored to help fish culturists control fish farming. This article proposes the simulation model of water quality monitoring in a pangasius pond based on cellular automata theory. The experiment results are verified using water quality data collected at Phu Thuan fish farm, Hau Giang province, Vietnam.

In this paper, we introduce two heuristic methods for solving Markowitz mean-variance portfolio optimization problem with cardinality constraints and bounding on variables: genetic algorithm (GA) and heuristic branching (HB) with some proposed improvements. There are exact methods for solving the problem: outer approximation, branch-and-bound, etc. They are efficient for small-size problems, which is under five hundred stocks. However, they are not applicable for larger size problems. We implement the algorithms on Vietnam and the United States stock market data. Numerical experiments show that GA and HB give good results and have some advantages, especially in computation times.

E-commerce systems have developed remarkably and provided a considerable profit for commercial companies and groups. Customers also benefit from such systems. However, the rapidly increasing volume and complexity of data lead customers to find that it is a challenge to find suitable products for their interests. Numerous product recommended methods have been researched and developed to support users when they visit E-commerce websites. This study proposes a recommendation system for a Clothes Online Sale system based on analyzing context-based and collaboration-based methods. Each type was divided into memory-based and model-based approaches. The results give the same product, but the cosine distance of the Word2vec + IDF algorithm is the lowest. We have also deployed algorithms including the K-nearest neighbor’s algorithm (KNN), singular value decomposition (SVD), non-negative matrix factorization (NMF), and matrix factorization (MF) for the comparison. The method is evaluated on Amazon women’s clothing, including 50,046 samples and six features. We proposed a content-based memory-based method using Word2vec + IDF and a collaboration-based model-based method using the SVD algorithm with the result of RSME as 1.268 to deploy on the sales system.

Increment of energy consumption in a fast-developing country like Vietnam has put high pressure on conventional power generation such as thermal power and hydroelectricity. In order to supplement energy demand, renewable energy sources are playing crucial roles. Among those sources, biogas energy has great potential, especial in livestock farm, for its ability to convert waste into heat and electricity. In this work, a control system has been developed to facilitate the operation of a biogas-based electrical generator. Self-customized generator of 25–100 kW is used to supply power to a hog farm. As the biogas production which is impacted by the livestock and the electrical load may vary over the time, it is required to manage the system in such a way that it can stabilizes the generator speed and thus its output power frequency. In this paper, we present a control system equipped with an industrial programmable logic controllers (PLC) and adopting PID algorithm which has been integrated to the biogas-based generators. The self-developed biogas base generation systems have been deployed in a few hog farms in the rural area of northern Vietnam. The experimental results have shown that the control system is able to supply power to the farm with stable frequency within 49.5 and 50.5 Hz while the load varies.

Business Process Execution Language (BPEL) specification is transformed into communicating finite state machines (CFSMs) specification for interacting, synchronizable web services. Computational distributed tree logic (CDTL) logic over a set of Kripke structures, called communicating minimal-prefix machines (CMPMs) which together constitute sum machine was proposed in previous work. Sum machine has a much better scalability of variety of peers as opposed to the more traditional product machine. This paper discusses about using sum machine and CDTL logic to model-check a web-service specification by integrating both choreography and orchestration in a modular fashion with bottom-up approach. We do a couple of case studies of Virtual Travel Agency (VTA) protocol and Fresh Market Update (FMU) service and illustrate the model-checking procedure on these protocols to verify the synchronizability and reachability properties of the protocol in an efficient manner, with parallel, distributed algorithm incurring polynomial time complexity.

In this paper, we use logistic regression model to study the factors affecting students’ results in the subject named ‘Analysis III (MI1130)’. The data is collected from the results of the questionnaires and the actual students’ academic performance given by the School of Applied Mathematics and Informatics (SAMI) of HUST. Data analysis provides a precise estimate of 5 factors that make an impact on students’ learning results in the subject, including the living environment, the CPA result, self-study time, difficulty in solving some Analysis III problems, and awareness of the application of the subject in reality. Then, some suggestions can be made to improve the quality of training and learning outcomes of students.

This study presents a prototype of the soft-fingered hand based on the operating principle of an asymmetric soft actuator, which is designed as a tube with the different thickness and stiffness of two sides and is activated by the pneumatic actuation. Based on such design, four soft fingers were fabricated from the silicone rubber material by using the molding method and then were assembled to the connectors and base to complete the soft-fingered hand. The ability of the proposed soft-fingered hand is validated by conducting the simulations and gripping experiments. The simulation and experimental results show that under the pneumatic activation, the soft fingers have a good bending deformation, and the proposed hand can grip several objects with different shapes, sizes, and weights.

In this paper, a design of Sinusoidal Pulse Width Modulation (SPWM) 400 Hz inverter is proposed for Ground Power Unit (GPU) in airlines industry. In which, the inductor filter is integrated into the isolated transformer, for reducing the production volume, also enhancing the system flexibility and reliability. In the control scheme, resonant controller with parallel structure is performed to control fundamental harmonic voltages. The computation delay always existed in digital systems is considered to handle through the phase compensator technique. Simulation studies in Matlab software and Hardware in the Loop (HIL) show that the controller can regulate the output voltage, with the Total Harmonic Distortion (THD) is just 1.21% and 2.31%, respectively, even under the nonlinear load condition. The 20 kVA GPU experimental prototype using digital signal processing (DSP) has been implemented verify the validity and feasibility of the proposed controller. The control strategy steady state and the dynamic performance are tested in carefully, which show that the steady state error is relatively small, the transient response is fast and THD is 2.24% for the rectifier load.

Wireless power transfer (WPT) is an appropriate method of delivering power without connecting wires to multiple devices. To further increase WPT performance, metamaterials’ extraordinary properties, such as electromagnetic field focusing, have been used successfully. Normally, metamaterial properties depend on multiple parameters. Several metamaterial designs require a significant amount of time to complete numerical simulation. In this work, we propose a rapid design square-spiral metamaterial method using an artificial neural network (ANN). When ANN is used, the results show an accuracy of 97.4% and a collective mean square error (MSE) less than 0.7 × 10^–3. For synthesizing the design parameters, the results show an accuracy of 95.6% and the MSE less than 7 × 10^–3. Besides, the computation time of 1000 samples can be reduced 93 × 10³ times compared to the HFSS simulation.

Effective land utilization and management will be an important foundation for economic and social development. In recent decades, information technology has developed rapidly and has been applied in all areas of life. However, the application of informatic technology, especially digital map system for professional purposes in land management at the commune level in Vietnam, is still limited. Thu Dau Mot city has a total natural land area of 11,890.58 ha, accounting for 4.41% of the total area of Binh Duong province. In recent years, the urbanization rate in Thu Dau Mot city is very high, every year, the fact that the area of agricultural land converted to non-agricultural is very large causes rapid changes in the reality of land use and pressures for management faced by the local government. With the aim of establishing land use status quo map in 2020 quickly, saving time and money for monitoring and evaluating land use changes over the past 5 years, this study is conducted. On the basis of inheriting the maps and data of the past years, processing remote sensing images and applying GIS software in combination with collecting field data, the map of current land use in 2020 is established, and changes of current land use in comparison with that of the year 2015 are assessed. Research results show that the area and percentage of main land use groups of Thu Dau Mot City by the end of 2020 are as follows: (1) Non-agricultural land group is 8929.53 ha (accounting for 75.10% of natural area of Binh Duong province), that area in 2015 is 7530.03 ha (63.33%), increasing an area of 1399.50 ha in 5 years (11.77%); (2) the group of agricultural land in 2020 is 2961.05 ha (24.90%) while that in 2015 is 3133.07 ha (26.35%), that means in 5 years agricultural land decrease an area of 172.02 ha (1.45%); and (3) the group of unused land is 0.00 ha (0.00%), in compared to 2015, that area is 1227.48 ha (10.32%), decreasing 1227.48 ha in 5 years (10.32%). The combination of GIS software with remote sensing images helps to create a land use status quo map effectively. The results of the research are the necessary information for the local government to ensure the sustainable use of land resources.

In this study, a platform of a fluid-power actuator system with a combination of electro-hydraulic and pneumatic for acceleration tracking control is proposed. Furthermore, a control strategy is provided to obtain high-performance results in controlling the piston's motion. Here, the particle swarm optimization (PSO), a computational method, is appropriately utilized for selecting the parameters of the classical proportional integral derivative (PID) control. The tracking errors are eliminated without the challenge of the tuning process, and the control performance is further enhanced. In order to validate the effectiveness of the control strategy, the numerical simulation results are eventually given. The remarkable result of the paper is that the position tracking control is precisely guaranteed when applying only a traditional PID controller with optimized parameters by using the PSO algorithm.

Object detection is one of the most fundamental tasks in digital image processing and has been widely applied in industries. Advances in deep learning are accelerating object detection methods. This work presents the end-to-end two-stage object detection method and the steel bars counting application. In the first step, we collect data and labeling. Second, data is trained and fine-tuned by the Faster-RCNN FPN model. Finally, predict the test data from the trained model. Based on the mean Average Precision metric, the steel bars detection result is 67%. The experience shows that this approach is feasible for counting the steel bars.

The Red River Delta is formed by the Red River System, which is the greatest river system in Vietnam’s northern region. The primary river, the Red River, has neither an water storage dam nor a hydroelectric dam. Because of this disadvantage, water level forecasting is critical for regulating agricultural water in Vietnam’s second-largest rice-producing region. We present a model to anticipate the water level of Red River water level in Viet Tri, which is near Ha Noi, in this study. The new model is known as the SARIMA-GRU hybrid model, which can fully exploit seasonal patterns in the data. In comparison to the single models SARIMA and GRU, as well as the model ARIMA-RNN, published by Xu et al. in 2019, the new model has produced better results.

Activity-based learning is one of the most trending methodology of learning. It is a teaching methodology which enables a learner to learn as per his or her natural pace using a series of activities which is more interactive, engaging and beneficial for young learners. It also has the facility of monitoring and evaluating the activities. However, the online activity-based learning became a new mode of teaching during the COVID-19 when the entire world went under lockdown. The situation was indeed very difficult for everyone to make their ends meet during that time. People were not allowed to come out of their homes except to purchase the daily needs. All the educational institutions including schools, colleges and universities were shut down for an indefinite period. The faculties were left with no other alternative but to take classes through online mode. This was a challenging task not only for students but also for the teachers. Therefore, the basic objective of the paper would be to discuss about the effectiveness of the implementation of activity-based learning through online mode and how far it has succeeded in creating an impact among the trainers as well as the learners.

Current techniques for electroencephalograph (EEG) emotion recognition still train prototypes equidistant using all EEG measurements. Moreover, since a few of the source (training) samples are significantly different from the target (test) samples, they can negatively impact. As a result, rather than forcing a classification model to be trained using all of the samples, it is crucial to listen carefully to EEG samples with a high transferability. Furthermore, according to neuroscience, not all of the signalling pathways in an EEG study contain emotional information effectively conveyed to the test results. Even some data from specific brain regions would significantly negatively impact learning the emotional classification model. In the light of certain two issues, in this article, we propose a TANN for EEG speech signals that develops emotional discriminant features by emphasizing traceable EEG neural domains data and samples adaptively through locally and globally attention mechanisms. To do so, measure the outputs of different brain discriminators as well as a specific test discriminator. TANN outperforms existing state-of-the-art approaches in comprehensive EEG emotion recognition studies.

Vietnam is the world’s second biggest producer of coffee and has been an exporter for several decades. However, Vietnam has been facing many well-documented issues and challenges in the whole coffee supply chain such as climate change; low productivity, poor quality and high cost; excessive use of fertilizing products and irrigations; poor collection, processing and storage solution; low sustainability and limited value-added coffee products; and low applications of smart and sustainable agriculture solutions. This paper introduces an innovative and smart agriculture (INNSA) platform for the creation and operation of a sustainable coffee value chain, with the focus on enhanced quality and added values for key elements of the coffee supply chain in Vietnam. The platform is designed based on the foundations of the key enabling digital transformation and smart agriculture technologies: Smart devices and Internet of Things, big data, artificial intelligence, blockchain and source traceability technologies, and sustainable design and manufacturing. The INNSA platform has the following key features: a smart database with real-time data inputs and updates, cost-effectiveness, and open-architectures for effective integrations of the enabling digital transformation and smart agriculture technologies. The smart platform and ecosystem of INNSA provides the information portal and open-access database for all key factors of coffee supply chain, enabling them to join and interact with each other to bring the coffee industry of Vietnam to a higher level, well-recognized in terms of values, branding and sustainability.

The mechanical property of the tissue can be used for medical diagnosis. In a study about shear wave elastography.

Breast cancer is one of the worst diseases in the world and the most common cancer affected by women. Early detection of cancers allows for faster treatments. Women rarely visit a clinic or hospital for routine tests unless they are ill because of long lines, expensive tests, and life difficulties. Recent studies have focused on early breast cancer diagnosis utilizing non-invasive UWB technologies. This article presents several appropriate supervised machine learning algorithms to detect breast cancer, worked with a user-friendly microwave ultra-wideband (UWB) device within the breast tissue. Two models for compressed breast tissue were created using the CST Microwave Studio simulator. These models generated two patient datasets with differing dielectric properties similar to human tissue. These two datasets are used to train the decision tree (DT), support vector machine (SVM), and nearest neighbor (NN) in order to develop an intelligent classification model that can assist doctors in identifying malignant breast cells. KNN can classify the breast data for the first group with 78% accuracy while the SVM 93% accuracy for the second group.

In recent decades, lots of robots are designed and produced all over the world because of their important applications. Nowadays, using the robot is more and more popular in many different fields. In practice, the modeling and control of most of the robots are performed with an important assumption that all links of a robot are rigid bodies. This is to simplify the modeling, analysis, and control for a robot. The elastic deformation of a link always exists during a robot’s operation. This elastic deformation of a flexible robot has significant effects on several characterizations and specifications of the robot such as the robot strength, the accuracy of the robot motion, the robot control, etc. In the literature, there have been many studies addressing the dynamics modeling and control of flexible robots. This paper presents an overview of the mathematical methods which have been used for the kinematic and dynamic modeling of the flexible manipulators.

Digital twin is one of the emerging areas of research and technology development and the enabling technologies of Smart Manufacturing and Industry 4.0. This study aims to develop and demonstrate a proof-of-concept prototype with a case study of the digital twin of a robotic system. The system has two main elements: the virtual element and the physical or the real element. The virtual element of system has been built based on the Unity platform, which is a cross-platform game engine developed by Unity Software Inc., and the physical element was built with the use of two servomotors and the NVIDIA^® Jetson Nano™ Developer Kit. The virtual and the physical elements are connected and communicated via using the TCP socket protocol suite. A digital twin model of the ABB IRB 120 robot was successfully developed and demonstrated. The collected data include the joint angle position values of the physical and virtual models, and they are stored both locally and in the cloud for the future system development, which can be used as for minimizing the errors between the physical and virtual models of digital twins of robotic systems. The successfully developed digital twin model can be considered as the cost-effective solutions for demonstrating and evaluating potential applications of digital twins in industrial practices as well as in higher educations and research.

An optimal motion tracking for humanoid robotic arms (HRAs) is proposed based on Kinect sensor. HRA models are designed using inverse kinematics (IK) with aid of Denavit–Hartenberg (D-H) method for both arms. Kinect camera is used to transfer the desired position and feed the point to HRAs simulation. Next, IK is implemented to obtain the required angles for HRAs. Then, apply two optimization methods: black hole (BH) and most valuable player algorithm (MVPA) in order to minimize the computation time (CT) and root mean square error (RMSE) between the desired and actual position. After that, forward kinematics (FK) is applied to find the actual position of the obtained angles of HRAs. Finally, the simulation of graphical user interface (GUI) is designed to show the optimal motion characteristic of HRA models. Calculation and simulation results show the efficiency of the optimization methods implementation after they are compared with each other. The simulation results showed that the MVPA method is more accurate than BH method. The RMSE and CT calculated by MVPA are less than those calculated using BH.

In the coming years, additional iterations of the 5th generation (5G) of wireless communication will be introduced. However, due to the inherent constraints of 5G and the development of new applications and services with demanding specifications like latency, energy/bit, traffic capacity, and peak data rate, telecom researchers are now concentrating on conceptualizing the following generation of wireless communications, known as sixth-generation wireless communications (6G). The Internet of Things (IoT) is anticipated to transform consumer applications and services, ushering in a future of fully intelligent and autonomous systems leveraging sixth-generation networks. A collaborative effort between industry and academia has started to conceptualize the sixth generation of wireless communication systems with the aim of laying the groundwork for stratification of communication needs in the 2030s in order to meet these demanding requirements and maintain wireless networks’ competitive edge. This work also goes deeply into the challenges, demands, and trends related to 6G while also providing a future vision for 6G wireless communication and its network architecture. This work includes some fresh, intriguing 6G services and use cases based on the requirements and solutions, which cannot be adequately supported by 5G. Furthermore, this study provides information on key research directions that contribute to successful 6G conceptualization and implementation.

In a radio over fiber (RoF) transmission, the light wave transmitted through an optical fiber is modulated by the radio signal. This technique provides a better path for the transmission of wireless signals over optical media in broadband wireless networks. RoF has developed over the past three decades along with a plethora of studies in the field. However, RoF still encounters several challenges, many of which have already been overcome, yet many still need to be addressed. In newly evolving networks such as 5G and what follows, bandwidth demands, response time, jitter, and fidelity on front-end networks cause significant challenges to RoF systems. Moreover, the movement from the lower microwave scope to the microwave scope was of direct advantage to wireless operations in terms of bandwidth. However, this movement poses more challenges to RoF expansion because combining wired and a wireless (fiber) network into one basic structure is a task of considerable challenge. Therefore, this paper provides an overview of RoF technology with a specific focus on linear and nonlinear effects, mitigation methods, and a discussion of future challenges.

Growth and development of the coffee plant depend on many environmental and management factors that must be continuously supervised and controlled. However, most of the coffee farms in Vietnam do not have any remote automatic control and supervisor systems for them. The care of coffee plants fully depends on weather and farmer experience, which have caused many issues and challenges such as low quantity, poor quality, and common pests. The present study describes the design of local network architecture of data acquisition controllers from the environmental sensors and cameras used for measuring, monitoring, and recording environmental parameters such as temperature and humidity as well as image of coffee leaves damaged due to disease. The slave controllers collect data and transfer them to a master controller through the Lora network, so that the master can send data to server using Internet. The slave controller also controls the actuator such as water pump and electrical fan to keep the environmental parameters within a suitable range. This design is the base to fabricate the real systems and experiments in coffee farms in Western Highlands of Vietnam later.

New age automation technologies that are digital and disruptive in nature are penetrating all spheres and processes of the modern economy and corporate world with a profound impact and at a global level. The function of human resources too has definitely undergone a paradigm shift in tune with the pace of technologies changes, in contrast to what it was several years ago. In this context, a set of challenges and opportunities await the profession. Artificial intelligence stimulated and transforming job roles, in terms of redefined work roles and enacting newer skills and tools and preparing the workforce for newer challenges. Artificial intelligence will make our employees work easier; monotonous task can be further automated, and relevant information from a huge volume of data will be more identifiable. The HR department must take proactive steps to adopt these technologies and update itself in terms of necessary skill. The chapter shows the strategic partner role of human resource management and artificial intelligence. The chapter shows how AI impacts the role of HRM practices in changing economy.