Self-learning System Research Articles

Design of an AI-based Personalized English Self-learning System

Design of an AI-based Personalized English Self-learning System

Theoretical basics of the self-learning system of intelligent locomotive decision support systems

Analysis of works in the field of artificial intelligence allows to make an assumption that today there is a sufficiently developed theoretical basis for the development of intelligent control systems for locomotive control. This will minimize the risks associated with the human factor on the railways. The paper presents the theoretical rationale for the development of a knowledge base for intelligent locomotive control systems. The approach and structure of the self-learning system of intelligent DSS is proposed, the advantage of which is the presence of a fuzzy classifier that works according to the set criteria and determines a fuzzy image of the current train situation. Learning a fuzzy classifier consists in finding a vector K that minimizes the distance between the results of logical inference and experimental data from the sample. The knowledge base is implemented using linguistic variables formalized by methods of fuzzy logic. The use of linguistic values makes it possible to design the base using the usual language of communication, which greatly simplifies both the design process itself and the analysis of the system's performance. Also, the knowledge base has the possibility of constant self-improvement. This happens in two ways. The first is by adding new rules to the knowledge base in case the current situation does not match the existing ones in the base, in which case an additional rule is created and checked for adequacy. The second way is a mechanism for ranking rules in the knowledge base. If the control action of the locomotive driver coincided with the recommendation of DSS in the current situation, then the rating of this recommendation (rule) increases, and in the future the rule selection algorithm will choose one or another control action for the current situation that has the highest rating (that is, it has already been verified several times person). The experiment has shown that the use of intelligent DSS has positive results. On average, the DSS made the correct train control decisions faster than the locomotive driver.

Study on Self-Learning System for Milling Chatter Feature Based on Hybrid Preprocessing Model and Improved Convolutional Clustering

Milling chatter is a self-excited vibration phenomenon during the cutting process, typically occurring in machining operations with lower stiffness. It significantly impacts the machining precision and surface morphology of the workpiece. To summarize the vibration signal features under operating conditions and enhance the precision of the intelligent recognition model for milling chatter features, this paper proposes an adaptive learning model for chatter feature recognition based on the improved convolutional clustering algorithm. This study uses wavelet analysis to extract chatter features and improved convolutional clustering to identify chatter and stable cutting signals. This method selected appropriate mapping techniques based on differences in chatter signals under various conditions and uses convolution for feature extraction. Improve convolutional clustering evaluates sample similarity through energy transfer, less influenced by signal space structure compared to Euclidean or Mahalanobis distances. This method provides a unified criterion for evaluating different signal feature representation spaces, thereby improving the accuracy and efficiency of chatter sample recognition. The recognition accuracy of chatter phenomena at different spindle speed ranges reaches $95$ % and $96.3$ % respectively. The results show that this method can effectively distinguish between chatter signals and stable cutting signals.

Improving students’ problem-solving abilities through answering questions in metaverse games: taking GOXR software as an example

In computer-based dialogue learning, students’ willingness to communicate and ability to interact is one of the important factors. Few studies have examined the testing of mobile metaverse teaching material systems built for interactive communication and learning courses. This study developed and designed course content for a virtual classroom, which is suitable for mobile learning on Android or iOS systems and can provide metaverse space for students to discuss homework using their mobile phones, as well as a system for online self-learning and testing. Students can participate in classes without time or space restrictions. In addition, questions related to life math knowledge and leveling up are included, such as the basic reproduction number of COVID-19, influenza, and enterovirus. There were 100 valid responses from two schools, which recognized the convenience and benefit of the virtual digital technology of the mobile self-learning system in the metaverse space as an auxiliary tool for learning, as it provides a more flexible and diversified educational learning mode. This system can be used for course teaching, after-class review, and testing to facilitate self-testing of learning results.

An interactive self-learning system using smartphone app and cards enabling braille touch experience for blindness

This research is conducted to address the challenges faced by individuals with visual impairment (VI) in learning braille through the development and evaluation of an innovative self-learning system. Traditional methods of braille education often rely on resource-intensive one-on-one guidance, which may not be feasible for all due to resource constraints. This study introduces a novel approach that leverages smartphone technology and tangible braille cards to create an engaging and effective self-learning experience. The study’s findings highlight the self-learning system’s effectiveness, evidenced by high ratings in technology acceptance model (TAM) evaluations and interviews with students and teachers, conducted collaboratively at the VI Rehabilitation Center in Malang, Indonesia. As an inclusive educational tool, the system bridges the gap between traditional braille learning methods and modern technological advancements, promoting independence and enhanced literacy for the community with VI.

Artificial Intelligence (AI) and Knowledge-Based Engineering (KBE) in Ship Design: Bridging Tradition and Technology Through ACQUAINT

_ Despite the limited use of artificial intelligence/knowledge-based engineering (AI/KBE) in industries with small series or one-off designs, our study demonstrates the technical feasibility and potential benefits of implementing AI/KBE in ship design processes. This research presents the development of “ACQUAINT,” uniquely designed to address the complexities inherent in bespoke shipbuilding. Central to this module is a robust AI-driven inference engine, integrated seamlessly with AutoCAD through a Python-based interface, facilitating a novel approach in shipbuilding’s detail and production design phases. The module’s capability to generate optimal designs autonomously—grounded in a deep understanding of design rules, constraints, and requirements—substantially reduces the reliance on human interaction. Our initial proof of concept with “ACQUAINT” showcases measurable advancements in ship design accuracy and efficiency, highlighting AI KBE’s transformative impact in shipbuilding and setting a foundation for future research and practical applications. Keywords ship design; artificial intelligence; knowledge-based engineering; self-learning system; software development; ACQUAINT; CAD/CAM software; computers in construction; computers in design; modernization; ship structure

Collective Decision of One-vs-Rest Networks for Open-Set Recognition.

Unknown examples that are unseen during training often appear in real-world pattern recognition tasks, and an intelligent self-learning system should be able to distinguish between known examples and unknown examples. Accordingly, open-set recognition (OSR), which addresses the problem of classifying knowns and identifying unknowns, has recently been highlighted. However, conventional deep neural networks (DNNs) using a softmax layer are vulnerable to overgeneralization, producing high confidence scores for unknowns. In this article, we propose a simple OSR method that is based on the intuition that the OSR performance can be maximized by setting strict and sophisticated decision boundaries that reject unknowns while maintaining satisfactory classification performance for knowns. For this purpose, a novel network structure, in which multiple one-vs-rest networks (OVRNs) follow a convolutional neural network (CNN) feature extractor, is proposed. Here, an OVRN is a simple feedforward neural network that is designed to assign confidence scores that are lower than those in the softmax layer to unknown samples so that unknown samples can be more effectively separated from known classes. Furthermore, the collective decision score is modeled by combining the multiple decisions reached by the OVRNs to alleviate overgeneralization. Extensive experiments were conducted on various datasets, and the experimental results show that the proposed method performs significantly better than the state-of-the-art methods by effectively reducing overgeneralization. The code is available at https://github.com/JaeyeonJang/Openset-collective-decision.

Evaluation Method of English-Speaking Self-Learning System Based on Natural Language Processing Technology

In the education, the use of advanced natural language processing techniques has gained prominence for their potential to revolutionize the teaching and learning experience. Self-learning in English teaching has become increasingly popular due to its flexibility and accessibility. With the advent of digital resources and language learning apps, students can now engage in language acquisition on their own terms. This paper presented a novel framework that combines the power of the Bidirectional Encoder Representations from the Transformers (BERT) model with the Hidden Condition Random Model (HCRM) for the enhancement of English teaching. The primary goal is to provide educators and institutions with a robust tool for evaluating the relevance and quality of teaching materials. The HCRM architecture incorporates sentiment analysis, feature extraction, and classification, making it a comprehensive solution for assessing the suitability of documents in the context of English teaching. The model takes into account the opinions of both students and teachers, ensuring a holistic perspective on the teaching materials' effectiveness. By effectively analyzing sentiments and extracting pertinent features, the HCRM facilitates a nuanced understanding of the potential impact of educational content. This paper's findings suggest that the integration of BERT with HCRM has the potential to greatly enhance English teaching by providing a more accurate, holistic, and data-driven approach to material assessment. The innovative framework presented in this research holds promise for improving the quality and relevance of teaching materials in the field of English instruction.

Self-learning system and methods of selection of weight coefficients of neural network

The present work is devoted to the problems of determining the parameters of neural networks. The methodology, algorithm and solution paths of finding the gradient descent, one of the main parameters in a neural network during machine learning, are presented. It is very important to minimize the error of finding the weight parameters of the neuron.

Role of cognitive computing in enhancing innovative healthcare solutions

Cognitive computing represents a groundbreaking development in healthcare, encompassing technological platforms that emulate the human brain's functionality. While cloud computing offers on-demand internet access to computing resources and services, cognitive computing focuses on modelling human mental processes to tackle complicated issues. Cognitive computing enhances human decision-making by integrating reasoning, machine learning, speech, natural language processing (NLP), and human-computer interaction. In the healthcare sector, it facilitates the analysis of clinical and genetic data to forecast diseases, tailor therapies, and elevate drug development. Additionally, it combines data analysis with adaptive page displays to tailor content based on the audience. Relevant papers in cognitive computing for healthcare were identified and studied. This paper aims to undertake an extensive scopic review of the pertinent literature from various sources, including articles and documents from numerous journals and conference proceedings. It delves into the need for cognitive computing in healthcare, elucidates supportive technologies, and expounds on its features within the healthcare domain. Furthermore, it identifies and discusses the substantial applications of cognitive computing in healthcare. These systems utilise computer models to replicate human cognitive processes, streamlining administrative tasks through artificial intelligence and cognitive computing. As a result, healthcare administrators can allocate more of their valuable time to patient care. Cognitive computing enhances outcomes and practitioner productivity and improves treatment decisions. The self-learning system of cognitive computing relies on real-time patient data, medical transcripts, and other pertinent information. These technologies can reduce the administrative burden on healthcare workers by automating tasks such as invoicing, claims processing, and appointment scheduling. This technology is poised to become increasingly indispensable in precision medicine.

Sequence-Type Classification of Brain MRI for Acute Stroke Using a Self-Supervised Machine Learning Algorithm.

We propose a self-supervised machine learning (ML) algorithm for sequence-type classification of brain MRI using a supervisory signal from DICOM metadata (i.e., a rule-based virtual label). A total of 1787 brain MRI datasets were constructed, including 1531 from hospitals and 256 from multi-center trial datasets. The ground truth (GT) was generated by two experienced image analysts and checked by a radiologist. An ML framework called ImageSort-net was developed using various features related to MRI acquisition parameters and used for training virtual labels and ML algorithms derived from rule-based labeling systems that act as labels for supervised learning. For the performance evaluation of ImageSort-net (MLvirtual), we compare and analyze the performances of models trained with human expert labels (MLhumans), using as a test set blank data that the rule-based labeling system failed to infer from each dataset. The performance of ImageSort-net (MLvirtual) was comparable to that of MLhuman (98.5% and 99%, respectively) in terms of overall accuracy when trained with hospital datasets. When trained with a relatively small multi-center trial dataset, the overall accuracy was relatively lower than that of MLhuman (95.6% and 99.4%, respectively). After integrating the two datasets and re-training them, MLvirtual showed higher accuracy than MLvirtual trained only on multi-center datasets (95.6% and 99.7%, respectively). Additionally, the multi-center dataset inference performances after the re-training of MLvirtual and MLhumans were identical (99.7%). Training of ML algorithms based on rule-based virtual labels achieved high accuracy for sequence-type classification of brain MRI and enabled us to build a sustainable self-learning system.

IMPROVING MATHEMATICAL PROBLEM-SOLVING SKILLS WITH AN INTELLIGENT SELF-LEARNING SYSTEM: A RANDOMIZED CONTROLLED TRIAL STUDY

The research paper discusses an intelligent self-learning system for analyzing the learning process of mathematical problem solving, which uses advanced machine learning algorithms to identify learners' strengths and weaknesses, and provides personalized feedback and recommendations to help learners overcome difficulties and develop their strengths. The system is based on analyzing data on test scores and assignments completed by learners. Machine learning models are used to determine the knowledge and competency level of each learner in different math topics. Based on this analysis, the system provides personalized feedback, pointing out strengths and weaknesses in the approach to problem solving and offering individual recommendations for independent skill development. In addition, the system uses natural language processing techniques to analyze students' written responses to mathematical problems. This makes it possible to identify students' errors and difficulties and provide real-time feedback. The proposed intelligent system has the potential to significantly improve the educational process by providing students with personalized support based on their individual needs and abilities in mathematics.

Evaluation of spoken English self-study system considering speech knowledge recognition algorithm

Abstract This paper builds and evaluates a self-learning system for spoken English pronunciation suitable for PC mobile terminal or mobile terminal users, and the core technology of speech recognition is built on Hidden Markov Model (HMM), which is used to decode the speech signal in spoken English learning. This paper studies the related speech recognition theory and signal processing technology, builds a comprehensive English self-learning system in more complex situations and more user types and conducts a comprehensive evaluation of the system. The results show that the overall accuracy of the HMM model in the spoken English recognition and evaluation system built in this paper is good, and the accuracy of the input audio for people of all ages is greater than 90%. In the younger population, the accuracy of male speech signals was the highest in both closed space and open space, reaching 98.12% and 96.53%, respectively. In addition, it is observed that the accuracy gradually decreases as more wrong judgements are made on the speech input signal. When the evaluation is poor, the accuracy of the scoring results decreases to 55%, and when there are fewer voice judgement errors, it is in the excellent range, where the accuracy reached 88%.

A Study on the Construction of the Evaluation System of the Teaching Ability of Students using Pattern Recognition for Studying Majoring in Badminton in the Mixed Learning Model of Physical Education Majors and Self-Learning System

With my country's ongoing education reform and the continuous development of information technology-enabled education methods, the teaching environment and conditions of various universities have greatly improved, and information technology is changing the learning methods of college students at an alarming speed. This is the advantage of the blended learning model. In badminton as a specialization of physical education, strengthening the professional ability training of college badminton students is an important reform in the direction of achieving for the current talent training goal of our country. In the present teaching environment, the traditional classroom teaching mode and online learning mode coexist and each has their own advantages. Based on the investigation and analysis of the current situation of badminton students' teaching ability, this article proposes methods such as the literature method, expert interview method, questionnaire survey method, derivative Delphi method, and so on, and designs and constructs a future-oriented badminton student teaching ability evaluation system. An in-depth study of the blended learning model has provided a certain theoretical and practical basis. The experimental results of this article show that by selecting two periods of indicators and calculating the weights, it can be seen that in the large-scale badminton teaching ability evaluation system, the badminton teaching organization and management ability (26%), and badminton skills and tactics are satisfactory, and the execution ability is also satisfactory (35%), which is relatively large, and the sum of the two exceeds 60%. The blended learning model exploits the advantages of traditional teaching, combined with online learning to promote the development of students' professional badminton practice ability.

One-Parameter Statistical Methods to Recognize DDoS Attacks

Within our academic high-speed network infrastructure which is used for connecting all universities and high schools in our country to the Internet, there are thousands of cybersecurity attacks occurring every day. That is why, within our SANET II project, an effort has been made to create a self-learning system without a teacher, which would be able to quickly adapt to arbitrary traffic and recognize DDoS attacks on time, even in high-speed networks, with a potential simple implementation into a hardware probe. In the article, we deal with the Hurst and autoregression coefficients and the coefficient of variation. We test the coefficients on simulated data and on real records of attacks. For early machine recognition of the attack, we propose the so-called predicting σ-tunnel. The obtained results can lead to the investigation of other prediction methods that would improve the early recognition of an attack.

Characteristics of the ideal healthcare services to meet adolescents' mental health needs: A qualitative study of adolescents' perspectives.

Despite increased focus on development of mental health services worldwide, many adolescents still hesitate to reach out to the services when they need them. This might be linked to the lack of adolescent involvement in the development of services. This study aimed to explore adolescents' perspectives on the ideal healthcare services to meet their mental health needs. A two-stage qualitative study was carried out to explore the perspectives of adolescents on the healthcare services to support their mental health. In the first stage, data from 295 adolescents attending a mental health seminar were analysed using qualitative content analysis. In the second stage, in-depth perspectives of 10 adolescent coresearchers were gathered through self-reflection and group discussions to further explore the categories developed in the first stage. Thematic analysis was used to develop the themes answering the research question. Ten adolescent coresearchers planned the study, collected and analysed data and authored the manuscript, with the support of a senior researcher. Five themes describe adolescents' perspectives on the ideal healthcare services to meet their mental health needs: (1) Culturally Sensitive and Responsive; (2) Communication of Information; (3) Easy Access; (4) Variety of Support; and (5) Consistency. Culturally Sensitive and Responsive services influenced all other themes. The themes describe suggestions for mental health service improvement, including how to manage the barriers that adolescents face to receive help from the mental health services. This study highlights the importance of culturally sensitive and responsive services. It raises the need for an expanded definition of culture going beyond nationality and ethnic background. Adolescents need flexible services that meet their individual mental health needs. This has implications for practitioners, educators, system organizers and researchers, who should also involve adolescents in planning, implementing and assessing the services. There is a need for a self-learning system to continuously adapt to user feedback. This study provides insight into adolescents' perspectives on the ideal mental health services. It offers suggestions for ways to improve services to better meet the individual mental health needs of adolescents. Additional research is needed to further develop and implement service changes, as well as to assess their acceptability, effectiveness, cost-effectiveness and safety. This is a study lead by adolescents. Adolescent coresearchers have, with the support of a senior researcher, planned and carried out the study, collected and analysed data and authored the manuscript.

Self-Study System Evaluation of Oral English Based on Limit Theory

With the globalization of the contemporary social economy and its ever-increasing influence, it is particularly important to master a common language in the world at this stage English. For the subjects who have the opportunity to learn English, their listening, reading, writing, and other abilities are relatively good, but the speaking ability is relatively lacking. Therefore, the self-study of spoken English is particularly important for the improvement of the practical application ability of English, and an efficient self-study system will also play a better role in the self-study process. The efficiency of a system depends on the maximum and minimum magnitudes that the system can carry. Therefore, for the self-learning system of spoken English in this paper, it is necessary to calculate the maximum and minimum carrying capacity. Therefore, it is necessary to deal with the limit thinking of the maximum and minimum carrying capacity of the system. The evaluation based on limit theory in this paper is a kind of evaluation of the maximum and minimum carrying capacity of the English oral self-learning system by using the convergence and dispersion characteristics of limit thinking. In order to evaluate the performance of the oral English self-learning system, this paper implements the dynamic evaluator of the oral English self-learning system, and constructs a hardware system for the self-learning oral English. This paper also uses the hidden Markov model in the self-learning system and the application of the central limit theory. The corresponding experimental results show that the mean value of the spoken English test data is 25.1337, and the standard deviation is 2.01385, which is in line with the normal distribution. It shows that the evaluation based on the limit theory can make the operation of the spoken English self-learning system stable.

DEVELOPMENT FEATURES OF COGNITIVE TECHNOLOGIES IN MASS MEDIA

The article examines the use of cognitive technologies implemented in the media, which affect the entire complex of mass media. The need to predict the impact of their extensive use on society makes this issue relevant. The purpose of the article is to identify the main factors influencing the cognitive technologies implemented in the field of mass media and the effectiveness of their use. The research methodology consists in systematic and logical analysis, as well as methods of comparison, synthesis, deduction, and classification, which made it possible to consider in detail the object of research as a system, to identify the driving forces in the field of mass media and categories of various cognitive technologies, their key features. The scientific novelty is to reveal the implementation of cognitive technologies in the mass media, which have the potential for shaping not only consumer loyalty but also determining the preferences of the audience for a long period of time thanks to the self-learning system. Conclusions. It is determined that the integration of cognitive technologies into the professional spheres of mass media is capturing more and more of its branches, helping to create modern products and processes, taking over computing, recognition, reproduction, and other tasks that were previously done manually with a significant loss of time. It is proved that the use of modern information technologies in the media gives the opportunity to increase the loyalty of the target audience, expand the presence on the market through new placement platforms and social networks, provide customer support and some accompanying and supporting functions for employees, significantly reducing the cost of operating activities of the company, automating it, and perform many other tasks. It is noted that although cognitive technologies are a powerful business tool, the pace of innovation is accompanied by the risks they pose to their understanding, restraining, and controlling.

AI Based Self Learning Intelligent Information Leak Protection System for TI Companies using LSTM

Preventing the leak of sensitive information, also popularly known as data leak or data loss to an unauthorized recipient, is the primary goal of an organization's information security system. A data leak can occur through multiple channels. While it may not always be possible to prevent it entirely, measures can be taken to minimize the possibility of the occurrence. Like all other financial institutions, TI companies collect sensitive personal information of their customers for business purposes. This information is often categorized into three primary types; NPI, PII, and PI are the designated types in the descending order of sensitivity. The detection of sensitive documents and redaction of sensitive information is required if it is needed to be shared. Inspection of such digital documents to find any sensitive information is by far a human-driven process, and thus time-consuming and costly. An and costly. An intelligent and robust system is required where the content is analysed by state-of-the-art data mining, statistical and machine learning techniques from various data dimensions. An AI based self-learning Intelligent Information Leak Protection System using LSTM is proposed in the project that mines and extracts information and categorizes the document images, to SD or NSD, based on the presence of NPI and PII semantic signatures without any explicit rule configuration. The system is designed to be used proactively as an early warning system to tag the SD images while resting in the data store. It can also act as a real-time checkpoint for the information loss by the documents in transit or use. The proposed model prescribes an information loss protection mechanism using a binary classifier based on the state-of-the-art LSTM technique within the paradigm of Artificial Intelligence.

Artificial Intelligence Driven Chatbot

Today, most large-scale conversational AI agents (e.g. Alexa, Siri, or Google Assistant) are built using manually annotated data to train the different components of the system. Typically, the accuracy of the ML models in these components are improved by manually transcribing and annotating data. As the scope of these systems increase to cover more scenarios and domains, manual annotation to improve the accuracy of these components becomes prohibitively costly and time- consuming. In this paper, a group of Amazon researchers propose a system that leverages user-system interaction feedback signals to automate learning without any manual annotation. Users here tend to modify a previous query in hopes of fixing an error in the previous turn to get the right results. These reformulations, which are often preceded by defective experiences caused by errors in ASR, NLU, ER or the application. In some cases, users may not properly formulate their requests (e.g. providing partial title of a song), but gleaning across a wider pool of users and sessions reveals the underlying recurrent patterns. The proposed self-learning system automatically detects the errors, generate reformulations and deploys fixes to the runtime system to correct different types of errors occurring in different components of the system. The results show that the approach is highly scalable, and able to learn reformulations that reduce Alexa-user errors by pooling anonymized data across millions of customers.