Computational Ability Research Articles

Neuro-inspired hardware solutions for high-performance computing: A TiO2-based nano-synaptic device approach with backpropagation

Computer-based machine learning algorithms that produce impressive performance results are computationally demanding and thus subject to high energy consumption during training and testing. Therefore, compact neuro-inspired devices are required to achieve efficiency in hardware resource consumption for the smooth implementation of neural network applications that require low energy and area. In this paper, learning characteristics and performances of the nanoscale titanium dioxide (TiO2) based synaptic device have been analyzed by implementing it in the hardware-based neural network for digit classification. Our model is experimentally validated by using 32-nm CMOS technology and the results demonstrate that the model provides high computational ability with better accuracy and efficiency in resource consumption with low energy and less area. The proposed model exhibits 20% energy gain and 16.82% accuracy improvement and 18% less total latency compared to the state-of-the-art Ag:Si synaptic device-based neural network. Furthermore, when compared to the software-based (i.e., computer-based) implementation of neural networks, our TiO2-based model not only achieved an impressive accuracy rate of 90.01% on the MNIST dataset but also did so with reduced energy consumption. Consequently, our model, characterized by a low hardware implementation cost, emerges as a promising neuro-inspired hardware solution for various neural network applications. The proposed model has further demonstrated outstanding performance in experiments involving both the MNIST and Fisher’s Iris datasets. On the latter dataset, the model exhibited notable precision (94.5%), recall (91.5%), and an impressive F1-score (92.9%), accompanied by a commendable accuracy rate of 93.04%.

Convolutional Neural Networks to Facilitate the Continuous Recognition of Arabic Speech with Independent Speakers

Automatic speech recognition (ASR) is a field of research that focuses on the ability of computers to process and interpret speech feedback from humans and to provide the highest degree of accuracy in recognition. Speech is one of the simplest ways to convey a message in a basic context, and ASR refers to the ability of machines to process and accept speech data from humans with the greatest degree of accuracy. As the human-to-machine interface continues to evolve, speech recognition is expected to become increasingly important. However, the Arabic language has distinct features that set it apart from other languages, such as the dialect and the pronunciation of words. Until now, insufficient attention has been devoted to continuous Arabic speech recognition research for independent speakers with a limited database. This research proposed two techniques for the recognition of Arabic speech. The first uses a combination of convolutional neural network (CNN) and long short-term memory (LSTM) encoders, and an attention-based decoder, and the second is based on the Sphinx-4 recognizer, which includes pocket sphinx, base sphinx, and sphinx train, with various types and number of features to be extracted (filter bank and mel frequency cepstral coefficients (MFCC)) based on the CMU Sphinx tool, which generates a language model for different sentences spoken by different speakers. These approaches were tested on a dataset containing 7 hours of spoken Arabic from 11 Arab countries, covering the Levant, Gulf, and African regions, which make up the Arab world, and achieved promising results. CNN-LSTM achieved a word error rate (WER) of 3.63% using 120 features for filter bank and 4.04% WER using 39 features for MFCC, respectively, while the Sphinx-4 recognizer technique achieved 8.17% WER and an accuracy of 91.83% using 25 features for MFCC and 8 Gaussian mixtures, respectively, when tested on the same benchmark dataset.

STEM: Its Potential in Developing Students' Computational Thinking

Computational thinking skills are a person’s ability to solve problems with several passed processes. The process is recognizing patterns, decomposing patterns, compiling algorithms, making models, and doing abstractions. This ability is essential for students to master understanding aspects of computational problems. It also helps students evaluate the suitability of computational techniques to solve the problem. In addition, this computational thinking ability can also help students understand the limitations and strengths of computing techniques, apply or adapt computing tools/techniques to new uses, recognize opportunities to use computing in new ways, and apply computational strategies. The STEM approach, is learning that integrates science, technology, engineering, and mathematics to develop creativity in solving problems. The transformation of the educational process in the current era that requires a high level of technology makes STEM one of the keys to progress and innovation in developing human competence. Based on a study of more than 30 articles and research results related to the effect of STEM on student learning outcomes, we can conclude that learning using STEM has the potential to develop students’ computational thinking skills. This article discusses how, theoretically, we can acquire computational thinking skills through education with a STEM approach. Keywords: STEM, developing students’, computational thingking

Exploring machine learning applications in chemical production through valorization of biomass, plastics, and petroleum resources: A comprehensive review

Machine learning (ML) is a subtype of artificial intelligence that uses a computer's ability to learn from a given set of accessible data. ML is becoming prominent in almost every business, including the domain of chemical engineering, where there have been numerous researches and investigations. This article provides a detailed overview of the use of ML in the production and characterization study of biomass, polymers, and petroleum products. Categories of ML, including classification, regression, and clustering, are also investigated to get a deeper understanding of ML. From this review, it can be concluded that ML has aided in numerous domains, such as the prediction of biomass energy, the stability of crude oil based on NMR spectroscopy, the calculation of gasoline's octane number, the estimation of fuel oil's kinematic viscosity, the classification of waste plastics, and the estimation of drilling efficiency in petroleum reservoirs, among others. Apart from this, ML has also been playing a significant role in the microwave-assisted pyrolysis of biomass, polymers, and petroleum resources. ML substantially influences chemical engineering and is especially useful for enhancing system efficiency and monitoring processes that are difficult to understand manually. Although several obstacles are associated with ML, such as black box behavior, the need for a large amount of data, and the difficulty of understanding the predictions, deploying the model in the future is uncomplicated once the learning program has been trained.

Development of Real-Time Unmanned Aerial Vehicle Urban Object Detection System with Federated Learning

In this paper, an urban object detection system via unmanned aerial vehicles (UAVs) is developed to collect real-time traffic information, which can be further utilized in many applications such as traffic monitoring and urban traffic management. The system includes an object detection algorithm, deep learning model training, and deployment on a real UAV. For the object detection algorithm, the Mobilenet-SSD model is applied owing to its lightweight and efficiency, which make it suitable for real-time applications on an onboard microprocessor. For model training, federated learning (FL) is used to protect privacy and increase efficiency with parallel computing. Last, the FL-trained object detection model is deployed on a real UAV for real-time performance testing. The experimental results show that the object detection algorithm can reach a speed of 18 frames per second with good detection performance, which shows the real-time computation ability of a resource-limited edge device and also validates the effectiveness of the developed system.

A novel mode of INS-aided BDS real-time high-rate and precise kinematic relative positioning between two moving platforms

ABSTRACT For kinematic relative positioning users between two moving platforms, limited communication bandwidth and computation ability usually cannot support real-time transmission of high-rate (≥10 Hz) BeiDou Navigation Satellite System (BDS) data. The performance of Ambiguity Resolution (AR) is also a major challenge in signal blocked and loss of lock environments. Based on BDS and Inertial Navigation System (INS) data, we develop a novel mode of INS-aided BDS kinematic relative positioning between two moving platforms, aiming to realize real-time, high-rate and precise positioning with low-cost communication modules in challenge environments. To achieve this goal, the INS-aided high-rate relative kinematic positioning and INS-aided BDS AR re-initialization methods are proposed. In this mode, the baseline bias caused by the different position datum is first defined and resolved. Then, high-rate INS data are assisted to obtain kinematic relative position when real-time kinematic positioning results are unavailable within 1 second. Once the BDS data are available, the predicted relative position is used as additional constraint to facilitate AR re-initialization and improve the kinematic positioning performance. The performance of these methods is discussed in a set of experiments with 1 Hz BDS data and 100 Hz INS data, and compared with the conventional method by sending raw BDS/INS measurements. The results show that the proposed methods can achieve an accuracy of about 5 cm for the INS-aided 100 Hz relative positioning in baseline components and lengths, which is equal to the conventional method, but the transmitted data have been sharply reduced by an average of nearly 80%. With the assistance of INS-predicted baseline constraint, the relative positioning performance has been further improved. The accuracy of baseline length errors is less than 4 cm, and the AR fixing rates keep larger than 95% during the experiment, while the wrongly fixed rates are reduced to less than 0.5%.

Improving mental arithmetic ability of primary school students with schema teaching method: An experimental study.

Skillful utilization of mental arithmetic can significantly improve students' mathematical computation ability. However, it was observed that primary school students often resort to reiterating the process of written arithmetic in their minds during mental arithmetic, which is not conducive to their numerical ability improvement. This paper devises a set of graphic teaching aids for primary school students' mental arithmetic improvement based on mental arithmetic strategies, schema theory, and working memory. To validate the effectiveness of schema teaching in enhancing mental arithmetic ability among primary school students, a controlled experiment was conducted with two groups of third-grade students randomly selected from a primary school in Jingshan City. The results, obtained through descriptive statistical analysis and the multitrait-multimethod approach (MTMM), indicated that the experimental group (n = 52) demonstrated significant improvements in speed, accuracy, and stability in mental addition and subtraction after a 14-day instruction period in schema teaching. This study offers a potent mental arithmetic teaching strategy for elementary mathematics education, which can lead to a comprehensive enhancement of students' mental calculation abilities. It also holds promise for inspiring innovative teaching methodologies in primary and secondary mathematics education in the future.

The Impact of Artificial Intelligence on Human Resource Practices

This research investigates the effects of Artificial Intelligence on practices within Human Resources Management. It examines various important results, including precision, automation, computational ability and capacity, real-time interaction, customization, and time and cost savings. The aim is to identify the potential advantages of incorporating AI. Data was gathered from 274 IT workers in Chennai City through a well-designed online survey. The study introduces a new research framework using IBM SPSS version 21 software for analysis as well as AMOS version 21.The results show that factors such as Precision, Computational Ability & Capacity and Customization significantly impact Time-Saving & Cost Reduction while Automation and Real-Time Interaction do not have significant influence. This study's unique contribution lies in investigating specific outcomes when utilizing AI technologies in Human Resources Management Practices by focusing on key variables like Precision, Automation, Computing Power & Capacity Real-time experience, Customization, and Time-Saving Cost Saving capturing an extensive understanding of anticipated outcomes during AI implementation within Human resources management along with their interrelationships.

106 Characteristics of Cognitive Impairment in Moyamoya Disease

INTRODUCTION: Cognitive impairment in adult moyamoya disease (MMD) is thought to be the result of ischemic stroke; however, the presence and extent of cognitive decline in asymptomatic patients remain largely unknown. METHODS: After classification using T2-weighted fluid attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI), a total of 19 MMD patients with a history of cerebral infarction, 21 asymptomatic MMD patients, and 20 healthy controls matched for age, sex, and years of education were prospectively included in this study. A detailed neuropsychological evaluation of two moyamoya subgroups and normal controls was conducted. In addition, participants underwent arterial spin labeling and resting state fMRI sequences. RESULTS: Asymptomatic patients showed varying degrees of decline in intelligence (Raven’s Standard Progressive Matrices, P = 0.001), spatial imagination (mental rotation, P = 0.014), working memory (verbal working memory-backward digit span, P = 0.011), and computational ability (simple subtraction, P = 0.014; complex subtraction, P < 0.001) compared with normal controls. MMD patients with cerebral infarction had more severe impairment in complex arithmetic (P = 0.027) and word short-term memory (P = 0.01) than those without symptoms. The asymptomatic MMD group also showed decreased CBF in the left anterior central and left inferior frontal gyri of the island fap with multiple node abnormalities in the brain network and reduced network connectivity. There was a significant association of these changes with cognitive decline in the MMD group. CONCLUSIONS: In asymptomatic MMD patients, a variety of cognitive impairments precedes the onset of clinical symptoms such as cerebral infarction, which may be a long-term complication of conservative treatment. Besides, disturbance of CBF and impaired brain network connections may be important causes of cognitive decline.

Discussion on the Reform of the Cultivation of Computational Thinking Ability for Information Engineering Major under the Background of Engineering Education Accreditation

Abstract Computational thinking is the basic ability that contemporary college students should have, and it is more demanding for engineering students. Combining with the orientation and characteristics of information engineering specialty in our university, and comparing with the graduation requirement index system in engineering education certification, this paper puts forward the idea of strengthening the cultivation of computational thinking ability, reconstructs the specialty curriculum system, strengthens the teaching of computational thinking series courses to optimize the teaching content, and implements the reform measures of “problem solving” series courses with “domain” as the teaching unit to innovate the practical teaching method. Years of teaching reform practice shows that the reform of computational thinking series courses has achieved good results and met the requirements of talent training objectives.

Analysis of the text matching problem

As a matter of fact, with the rapid development of computation ability as well as machine learning scenarios, the natural language model had widely use in society in recent years. To be specific, lots of software and applications have been developed, such as Chat-GPT, machine translation and text generation. Among various applications, text matching constitutes a foundational challenge within the realm of natural language processing (NLP). With this in mind, this study delineates the advancements encapsulated by contemporary models pivotal to text matching methodologies, i.e., LSTM, Transformer as well as BERT. At the same time, this study will make a comprehensive demonstration of them, including the principle and the related works. In the meantime, this study will also provide 10 well-known dataset of Text Matching problem. Finally, this research presents analysis of the limitation for this task as well as gives the future prospect for text matching at the same time.

Medication Recommendation System Based on Natural Language Processing for Patient Emotion Analysis

Natural Language Processing (NLP) is an interdisciplinary field of computer science, artificial intelligence, and linguistics that focuses on the ability of computers to understand, process, generate, and simulate human language in order to achieve the ability to have natural conversations with humans. The underlying principles of natural language processing are at multiple levels, including linguistics, computer science, and statistics. It involves the study of language structure, semantics, grammar and pragmatics, as well as the statistical analysis and modeling of large-scale corpora. In the process of concrete implementation, it is necessary to process natural language at multiple levels. Based on this, this paper combined deep learning and natural language processing technology to conduct sentiment analysis on patients' comments, so as to recommend drugs that are more suitable for patients, thus achieving accurate drug prescribing and personalized recommendation.

Analysis of the methods and performances for data augmentation: Image, text and audio

With the rapid development in computation ability as well as machine learning scenarios, various artificial intelligence applications can be achieved in recent years. With this in mind, this study will explore the application of data augmentation in machine learning and deep learning. To be specific, this paper first introduces the background and research history of data augmentation and then discusses the research progress in recent years. The basic description of this study describes the definition, common methods, and evaluation metrics of data augmentation in detail. At the same time, three data augmentation models, AutoAugment, AugGPT, and SpecAugment++, are introduced respectively, including their principles, experimental results, as well as evaluation. Finally, according to the analysis, the limitations and prospects of the field are discussed and demonstrated, as well as summarize the main findings and research implications of the full paper. Overall, these results shed light on guiding further exploration of data augmentation.

Analysis of the implementation for big data analysis in sales prediction: LSTM, ANN and DNN

As a matter of fact, big data analysis is a choice made by the sales forecasting industry in response to the trend of the times, especially with the rapid development of computation ability and machine learning models. In general, it consists of two parts, i.e., big data and machine learning. To be specific, machine learning has received widespread attention in this field after being supported by big data. On this basis, this study will select the implementation of three machine learning models, LSTM, ANN and DNN. The prediction accuracy of these three models all meets practical requirements, and their performance in complex data is better than traditional models, but their costs are still not affordable for most companies. The purpose of this article is to help readers understand the development of big data analysis in the sales forecast industry, its current advantages and disadvantages, as well as possible future development directions.

A Hybrid Adaptive Simulated Annealing and Tempering Algorithm for Solving the Half-Open Multi-Depot Vehicle Routing Problem

The half-open multi-depot vehicle routing problem (HOMDVRP) is a typical decision optimization problem in the field of collaborative logistics that considers resource sharing. This study aims to develop an effective meta-heuristic algorithm for solving the HOMDVRP. Firstly, a mixed-integer programming model of HOMDVRP is established to minimize the total travel distance of the vehicles. After that, a novel hybrid adaptive simulated annealing and tempering algorithm (HASATA) is proposed based on the features of HOMDVRP. The proposed algorithm combines the strengths of the simulated annealing algorithm and the large-neighborhood search algorithm to balance the algorithm’s searching capabilities in both breadth and depth. Meanwhile, an adaptive Markov chain length mechanism and a tempering mechanism are designed to improve the algorithm’s computational efficiency and convergence ability. Finally, simulation experiments are conducted to verify the effectiveness of the proposed model and the computational performance of the proposed algorithm. Four comparison algorithms are selected and analyzed using 24 groups of problem instances. The comparison results show that the proposed HASATA can solve the HOMDVRP more efficiently and obtain a solution with better optimization performance and satisfactory stability.

Composing music based on machine learning

As a matter of fact, the music is an important tool and pathway for human emotional communication, with great significance, especially in recent years. However, it should be noted that traditional music production methods require a lot of time, manpower, and funds, and have high costs. With the rapid development of machine learning technology as well as computation ability, the computer automatic composition based on machine learning technology is receiving more and more attention. With this in mind, this study introduces and discusses the automatic emotion recognition and emotion based music generation based on this advanced technology, as well as describes the process involved. According to the analysis, the advanced applications and implementations results have been demonstrated. At the same time, the current limitations as well as the future prospects will be clarified and discussed. Overall, these results shed light on guiding further exploration of machine learning implementations of music composition.

Intelligent computation in cancer gene therapy.

In recent years, the use of gene therapy for the treatment of disease has gained substantial interest, both in academic research and in the biomedical industry. Initial experimentation in gene therapy has generated positive results, as well as questions regarding safety. However, lessons have been learned from these first investigations, among them a realization that such treatments require a method to fine-tune the expression of therapeutic genes in real-time. A logical solution to this problem arose through the field of synthetic biology in the form of synthetic gene circuits. Thus, the synthetic biology community today aims to create "smart cells" for a variety of gene therapy applications, in an attempt to precisely target malignant cells while avoiding harming healthy ones. To generate safer and more effective gene therapies, new approaches with emerging computational abilities are necessary. In this review, we present several computational approaches which allow demonstrating artificial intelligence in living cells. Specifically, we will focus on implementing artificial neural networks using synthetic gene regulatory networks for cancer therapy and discuss the state-of-the-art computational developments.

Broadcast Swin Transformer for Music Genre Classification

As a matter of fact, with the rapid development of computation ability as well as machine learning models, music genre classification (MGC) has been widely explored in recent years, which is crucial to the development of modern digital music media platforms. With this in mind, this paper proposes a novel architecture called Broadcast Swin Transformer (BST). To be specific, it adds the Broadcast Mechanism to Swin Transformer, which can effectively convey as well as utilize the low-level information of the spectrogram at multiple scales. According to the analysis, the model has been experimented on Mel-spectrograms extracted from the audio dataset GTZAN with a Top-1 accuracy of 99.0%. At the same time, its excellent performance has also been demonstrated and evaluated in ablation study and comparison with the state-of-the-art methods. In the meantime, the current limitations as well as further prospects are presented as well. Overall, these results shed light on guiding further exploration of music genre classification.

Capitalized Comparison of Three Machine Learning Models: Linear Model, Decision Tree, Neural Network

As a matter of fact, with the rapid development of computation ability, machine learning has boosted rapidly with much faster training speed in recent years. In reality, machine learning is transforming industries with its ability to derive insights and patterns within massive datasets. Among numerous algorithms available, certain foundational models stand out due to their efficiency and capability. With this in mind, this study compares three typical models among various machine learning scenarios, i.e., linear models, decision trees, and neural network models. According to the analysis, the basic principle, concepts as well as parameters will be demonstrated. While all three has its unique pros and cons, this study aims to guide readers in choosing most fitting model with their tasks, by clarify the difference and future outlooks of these models. At the same time, the future development trends for the machine learning models will be proposed based on the analysis. Overall, these results shed light on guiding further exploration of machine learning.

Nuclear matter radii from molecular rotations using ultra-high-resolution spectroscopy

The rotational constant parametrizes the relative spacing between a molecule's rotational energy levels. It depends on the molecule's classical moments of inertia, which, in all studies, are expressed by treating the constituent nuclei as point masses separated by the bond length. We point out that treating the finite nuclear size leads to a correction to the rotational constant at the hertz level, which is resolvable with recently developed ultra-high-resolution molecular spectrometers. Nuclear-model-independent measurements of nuclear matter radii can thus be envisioned in the future using such apparatuses, advancing beyond the existing hadronic scattering experiments and further developing the intersection of nuclear and molecular physics. At the present time, it appears that the computational ability of quantum chemistry might be the limiting factor to the technical readiness of the approach. To test the premises of the proposed method, we call for benchmark experiments using HD+ that are feasible with state-of-the-art experiment and theory. Published by the American Physical Society 2024