Predicting the Occurrence of a Data Breach

Mehdi Barati, Department of Information Sciences, University at Albany, Albany, New York, USA

ABSTRACT

The problem of data breach incidents in the United States has been attracting the attention of researchers and practitioners recently. Commercial entities and government agencies who obtain and process personal information face the risks of the breach of personal information and consequently legal, reputational, and financial damages. The risk management approach to information security is common to manage and mitigate those risks. This paper aims to estimate the probability of the occurrence and size of data breach incidents for government and commercial entities by devising a predictive model with historical data as its input. We used a dataset of all reported data breaches in the US collected and published by Privacy Rights Clearinghouse (PRC). The results show that the trend of a data breach has not been increasing despite all the attention and warnings. The distribution of data breach occurrence follows the Poisson and Negative Binomial distributions closely. Both models proved promising and can predict data breach incidents with low deviance from actual numbers.

KEYWORDS

Data breach, Information security, Information privacy, Predictive modelling, Risk assessment.

LaBelle: A Deep Learning APP that Helps You Learn Ballet

Sarah Fan¹, Kevin Guo², Yu Sun³,¹Sage Hill School, 20402 Newport Coast Dr, Newport Beach, CA 92657, ²University of Southern California, Los Angeles, CA 90007, Irvine, CA 92606,³California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

Human Pose Estimation has proven versatility in improving real-world applications in healthcare, sports, etc. Proper stance, form and movement is instrumental to succeeding in these activities. This paper will explain the research process behind the deep learning mobile ballet app, LaBelle. LaBelle takes in two short videos: one of a teacher, and one of a student. Utilizing Mediapipe Pose to identify, analyze, and store data about the poses and movements of both dancers, the app calculates the angles created between different joints and major body parts. The app’s AI Model uses a K-means clustering algorithm to create a group of clusters for both the student dataset and the teacher dataset. Using the two sets of clusters, LaBelle identifies the key frames in the student-video and searches the teacher cluster set for a matching set of properties and frames. It evaluates the differences between the paired frames and produces a final score as well as feedback on the poses that need improving. We propose an unsupervised guided-learning approach with improved efficiency in video comparison, which is usually both time and resource consuming. This efficient model can be used not just in dance, but athletics and medicine (physical therapy like activities) as well, where stance, form, and movements are often hard to track with the naked eye.

KEYWORDS

Artificial Intelligence, Machine Learning, Deep Learning, K-means Clustering, Computer Vision, Ballet

Using singular value decomposition in a convolutional neural network to improve brain tumor segmentation accuracy

Pegah Ahadian, Maryam Babaei and Kourosh Parand, Department of Data and Computer Sciences, Shahid Beheshti University, Tehran, Iran

ABSTRACT

A brain tumor consists of cells showing abnormal brain growth. The area of the brain tumor significantly affects choosing the type of treatment and following the course of the disease during the treatment. At the same time, pictures of Brain MRI are accompanied by noise. Eliminating existing noises can significantly impact the better segmentation and diagnosis of brain tumors. In this work, we have tried using the analysis of eigenvalues . We have used the MSVD algorithm, reducing the image noise and then using the deep neural network to segment the tumor in the images. The proposed methods accuracy was increased by 2.4% compared to not using it. From the decomposition of singular values, convergence speed has also increased, showing the proposed methods effectiveness.

KEYWORDS

Artificial Neural Network, SV, Brain tumor, Brain MR, Segmentation

Reconfigurable Gamification Platform for the Autonomous Learning of Low Value Medical Practices

César Fernández¹, María Asunción Vicente¹, Susana Lorenzo², Irene Carrillo¹ and Mercedes Guilabert¹, ¹Department of Health Psychology, Miguel Hernández University, Elche, Spain, ²Hospital Universitario Fundación Alcorcón, Madrid, Spain

ABSTRACT

Failure to follow do-not-do recommendations (also known as low-value practices) is one of the causes of the lack of quality care in all health systems in all countries. Healthcare professionals must be provided with information about these low-value practices that are still frequently performed and their implications for patients and the healthcare Continuous education is a key factor in this scenario, so that health students, health professionals, and even patients are kept updated with the main do-not-do recommendations. Gamified platforms are one of the most valuable options for continuous education, as they combine learning efficiency with a high level of engagement for the students. Besides, the effectiveness of gamification platforms can be improved by adding artificial intelligence techniques.In this paper, a novel gamified platform focused on improving knowledge about low-value practices is proposed. AI techniques, as well as NLP tools are used to optimize the effectiveness of learning by adapting the platform to each user, at an individual level. Besides, the engagement of students is encouraged by their participation in a common project, namely the creation of a specialized dictionary for do-not-do terms. Hardware development is currently in progress. A basic gamification platform has already been developed for the two main mobile operating systems, and the development of IA and NLP techniques to analyse the training outputs and make the platform adaptable to each student is progressing.The proposed learning tool can have an important impact on the improvement of healthcare quality, and can be applied to many other learning fields, particularly when continuous training is required.

KEYWORDS

Low-value practices, Do-not-do, Gamification, Artificial Intelligence, Natural Language Processing

On the Energy Efficiency of Ieee 802.11ax WLANs

Zineb Machrouh and Abdellah Najid, Department of Communication Systems, INPT, Rabat, Morocco

ABSTRACT

Wireless communications evolved in a remarkable way during the last decade and is well on its way to surpass wired internet. The demand shifted towards higher transmission speed for more users and heavier traffics. In this paper, we present an IEEE 802.11ax scenario, in which we study the energy efficiency for the key metrics of the MAC layer. In this latest edition, called High Efficiency WLAN (HEW) energy is a main concern in order to satisfy scenarios of internet of things and wireless sensor networks. we prove that some of the new features such as the higher order modulation and coding schemes enhance remarkably the energy efficiency. We also show the impact of an increase in the number of users on the system and prove the payoff of using IEEE 802.11ax. We evaluate the contention window size performance as one of the most important metrics, on which throughput highly depends.

KEYWORDS

Contention Window, Energy Efficiency, IEEE 802.11ax, Throughput, Wireless Networks

Ensembles for Class Imbalance Problems in Various Domains

Deepakindresh N, Gauthum J, Jeffrin Harris, Harshavardhan J, Shivaditya Shivganesh, School of Computer Science Engineering, Vellore Institute of Technology, Chennai, India

ABSTRACT

The paper is an analysis of class imbalance problems from various domains such as the medical field,sentiment analysis, software de-fects, water portability, and relationship status of students and summarizes the performance of data resampling techniques such as random undersampling and oversampling. Synthetic minority oversampling techniques combined with the power of ensemble methods such as bagging, boosting, and hybrid techniques are generally used to solve the class imbalance problem.

KEYWORDS

Machine Learning, Class Imbalance, Multi Domain Analysis

Blockchain in insurance industry: turning threat into innovative opportunities

Wadnerson Boileau, United States of America

ABSTRACT

Insurance has been around for more than centuries. This risk mitigation strategy has been utilized in maritime commerce as early thousand years ago, where Asian merchant seafarers were pooling together their wares in collective funds to pay for damages of individual’s capsized ship. In 2018, insurance industry made up 6% of global GDP while financial industry amounted to about 7-9% of the US GDP. In 2020, the industry net premiums written totaled $1.28 trillion, created 2.9 million jobs, and recorded $2.0 trillion investments [1]. Despite of growing reform, the insurance market is dominated by intermediaries assisting people to match their insurance needs. While many predictions focused on artificial intelligence, cloud computing, blockchain stands out as the most disruptive technology that can change the driving forces underlying the global economy. We will focus on presenting blockchain use cases in insurance, demonstrating how the sector can turn blockchain threat into innovative opportunities.

KEYWORDS

Blockchain technology, insurance, bitcoin, process.

Understanding People S Awareness Towards Social Engineering with Survey

Aashka Raval, Shreya Chakrabarty, Harshita Jasoliya and Nishant Doshi, Department of Computer Engineering, Pandit Deendayal Petroleum University, Gandhinagar, India

ABSTRACT

The world is recovering from corona, and along with that, it has brought the zeal to use digital media, concepts like work from home, connecting the whole world using applications and social media with good things follow bad, we observe a lot of people affected by social engineering attack via multiple means in some cases they are known with the process but are unaware of the names they are victimized with while different types do not know many of social engineering attack. In this paper, we represent the survey filled by 100+ people from diverse age groups and work profiles seeking their views on the attack and knowledge about social engineering.

KEYWORDS

Social Engineering, Security, Phishing, Cyber-attack, Scam, Cyber Crime.

A Blockchain based Security Model for IoT Ecosystem

Sarthak Agrawal, Saksham Sharma and Surjeet Balhara, Department of Electronics and Communication Engineering, Bharati Vidyapeeth College of Engineering, New Delhi, India

ABSTRACT

The use of IoT devices has been increasing exponentially with time and this raises serious concerns for the overall protection of the huge IoT Network’s and the database’s associated with it. While there are many proposed approaches to deal with different security related aspects in IoT, one of the potential solutions to such issues is Blockchain. Blockchain is a rapidly emerging technology and is used in various fields. Blockchain technology has features like decentralisation and immutability which guarantees security. A blockchain based security model has been proposed in this paper for securing IoT devices from various security threats. Finally, proposed approach and its implementation using blockchain to secure IoT Ecosystem is discussed to make IoT ecosystem more secure.

KEYWORDS

Authentication, Blockchain, Data Protection, IoT, Security.

Blockchain, Smart Contract, And Global Digital Transformation

Zdenek Chalus, Independent Researcher

ABSTRACT

Global digital transformation (GDT) is a challenging goal with two paths. The first is about future profit and benefits, and the second is about the know-how gained via working on a task with vast integrity. The second direction creates the content of understanding of present digitalization in organizations and projects. Opens space for initiatives, e.g., for a new project paradigm that explains the definition of the United Economy for the GDT development operations. The author uses his work on the SPC Concept development, its operational units (SPC Utility), and rules of limits of the scope of the projects (SPC Drivers) to test the ideas of the target group and final beneficiaries. The author of this article mentions new initiatives and is interested in the experts formation (a new guild) for Blockchain and Smart contact technologies, which is meeting at its second international conference.

KEYWORDS

Philosophy, Business, Scaling, Blockchain and Smart Contract, and New Project Paradigm.

Redactable Blockchains: Current State,Challenges and Applications

Shaleeza Sohail, University of Newcastle, Central Queensland University, Australia

ABSTRACT

Recently the rewritabilty of blockchain data has become increasingly important dueto legal and ethical requirements. A redactable blockchain was recently proposed in 2017, thatallows one to modify the data in the blockchain without causing a hard fork. There are a numberof techniques proposed to construct redactable blockchains in permissioned and permissionlesssettings. In this paper, we provide a systematic literature review of the current state in thisdomain. We briefly discuss the techniques and approaches proposed for providing redactability inblockchain while highlighting their strengths and limitations. We analyse application requirementsrelated to redactability and then discuss how those requirements can be fulfilled by the existingapproaches. As for any new field of research, there are some open challenges in this area that needsolutions. In this paper, we present our effort to state some of these challenges.

The Prospect of Cryptocurrencies in Qatar: Projected Behavioral Response

Abdalftah Hamed Ali, Jad Tayah and Ahlam Alnaqeb, Faculty of Economics, Administration and Public Policy, Doha Institute for Graduate Studies, Doha, Qatar

ABSTRACT

Cryptocurrencies have taken the world by storm and generated massive riches almost overnight. Thus, it is important to develop basic literacy in this topic, as the literature concerning cryptocurrencies in Qatar and Arab world is lacking. Our study aims to assess the perception of the Qatari public towards this novel technology to evaluate whether it can be incorporated into Qatari status quo. As the technology is currently banned in Qatar, we have adopted Berry & Berry’s (2007) policy innovation and diffusion framework to back our assumption that the ban will be revoked. As to quantifying the perception of the local population, the technology acceptance model was used through Likert scale surveys. The analysis resided on linear and singular regression model to understand the impact perception has on Qatar’s population adoption of cryptocurrency. The results showed a positive intention but for a reliable verdict, the sample and analysis must be expanded.

KEYWORDS

Cryptocurrency, Qatar, Perceived usefulness, Perceived ease of use, Behavioral intention.

Named Entity Recognition Model of Power Equipment based on Multi-Feature Fusion

Yun Wu, Xiangwen Ma, JiemingYang, Anping Wang, Northeast Electric Power University, Jilin, China

ABSTRACT

Aiming at the problems of complex entities and difficult identification of rare entities in Chinese named entity recognition in the field of power equipment, this paper proposes a Chinese named entity recognition model based on multi-feature fusion. First of all, under the guidance of the electric field dictionary, text segmentation and part-of-speech tagging are carried out. Then, use Word2Vec to obtain the corresponding dictionary of character-character vector and word-word vector, and the word category vector is randomly initialized. Finally, the three feature vectors are integrated in series as input vectors, which are input into the BiLSTM-CRF model for sequence labeling. By using the proposed model, the semantic features of words as well as related domain knowledge are effectively applied. The experimental results show that the entity recognition model proposed in this paper improves the recognition effect of Chinese named entities in the field of electrical equipment.

KEYWORDS

Electrical equipment, Chinese named entity recognition, Domain dictionary, Deep learning.

Column Type Detection based on Pretrained Language Models with Crosscolumn Table Encodings

Peining Li and Mizuho Iwaihara, Graduate School of Information, Production and Systems, Waseda University, Japan

ABSTRACT

Real-world tables provide valuable long-tailed facts, and detecting semantic types of table columns is important for table understanding and associated tasks. However, existing methods are often built on heavily-engineered features, such as statistical features and straightforward string matching, which are not robust to dirty data and lack of extensibility. Deep learning models in the field of natural language processing have been successfully adopted to various sequence prediction tasks. In this paper, we discuss utilizing pretrained language model BERT and its variants for table column typing. We present a number of table encoding methods that serialize the target table into a token sequence, which is used for finetuning BERT toward the column typing task. Various orderings of table cells are discussed, where cross-column encodings are introduced to learn relationships between columns and rows simultaneously. We also discuss aggregating prediction results of subtables to deal with tables of many columns and/or rows. Experimental results show that our model outperforms the state-of-the-art method on both weighted- and macro-averaged F1 scores.

KEYWORDS

Column Type Detection, Deep learning，Natural Language Processing，BERT，Pretrained Language Models.

Global Trends and Gaps in Research Related to Latent Semantic Analysis Research Productivity

Ibrahim Hussein Musa^1,2, Ibrahim Zamit³ and Guilin Qi^1,2, ¹Department of Software Engineering, School of Computer Science and Engineering, Southeast University, Nanjing, China, ²Key Laboratory of Computer Network and Information Integration, Ministry of Education, Southeast University, Nanjing 211189, China, ³Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong 518055, China

ABSTRACT

The growth of big data has attracted research on Latent Semantic Analysis (LSA) in recent years. The study aims to provide a comprehensive bibliometric analysis to the LSA scientific landscape researchusing Scopus database. The results were analyzed using Origin Pro 9, VOSviewer Var1.6.6 and Bibliometrix R-package software programs. In total, the 874 retrieved articles were published in 1988 to 2020 and cited 31,464 times and published across 595 journals. Zhang Yin was identified as the most productive author. The USA and China shared the largest amount of research output. The most preferred publication channels in the LSA field were conference proceedings, especially Lecture Notes in Artificial Intelligence and Bioinformatics published as series Lecture Notes in Computer Sciences by Springer. The topic trends of emerging hot topics saw a study increase after 2000. Conclusion: The findings of this study provide concrete evidence of LSA landscape research evolution, and trends.

KEYWORDS

Bibliometric study, Latent Semantic Analysis (LSA), Latent Semantic Index, VOSviewer, visualization.

A diachronic shift in Japanese word length distribution

Wenchao Li, Department of Japanese Studies, Zhejiang University, Hangzhou, China

ABSTRACT

Given the typological differences between the Indo-European languages, which are fusional, and Japanese, which is agglutinative, the debate around the measuring unit of Japanese word length is unsurprising. This study delved into diachronic issues and calculated word length in Old, Early Middle, Middle, Early Modern, and Modern Japanese using data from eight writing systems, including 21 genres. This study aimed to clarify how word length distribution has shifted throughout history. The findings revealed that word length is associated with the writing system. Old Japanese bore the longest length, as it was utterly logographic. Since Early Middle Japanese, Japanese text has been written using a phonographic and logographic mix, and word length appears shorter. Furthermore, word length is associated with the diversity of genres. Moreover, an investigation of word length and frequency indicated that textbooks, sharehon, ninjoohon, and tales, which appeared after the Nara Period and used mixed Chinese character and kana writing, fit into the power law function.

KEYWORDS

Japanese, word length, genre, measuring unit, mora, syllable, writing system.

MusicApp, A Music App that can Record and Transcribe Music into Sheet Paper

Daniel Wang¹ and Yu Sun², ¹Troy High School, 2200 Dorothy Ln, Fullerton, CA 92831, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA92620

ABSTRACT

As technology advances, we have found more practical uses for it. This ranges from such things as cleaningthehouse using machines to serving restaurants with robots. Using technology, what if we can use machines toautomatically write sheet music for us, transcribing it from audio [1]. This paper designs an application todoexactly that. We used Java to write a program and app that would be able to transcribe audio into sheet music andstore it on an app. We applied our application to multiple cases and conducted a qualitative evaluation of theapproach. The results show that it is possible with some fine tuning and may be usable in the near future.

KEYWORDS

Music, Sheet Music, Audio to Sheet Music, Flutter.

Prediction and Key Characteristics of All-Cause Mortality in Maintenance Hemodialysis Patients

Mu Xiangwei¹, Zhu Mingjie², Liu Shuxin², Li Kequan¹ and You Lianlian², ¹School of Maritime Economics and Management, Dalian Maritime University, Dalian 116026,Liaoning, China, ²Dalian Key Laboratory of Intelligent Blood Purification, Dalian Municipal ,Central Hospital affiliated with Dalian Medical University, Dalian116033,Liaoning, China

ABSTRACT

Predict and analyze key features of all-cause death in maintenance hemodialysis patients to provide guidance for later diagnosis and treatment. Four machine learning methods were used to establish anall-cause death prediction model for maintenance hemodialysis patients and compare their performance. Analyze the key characteristics that have an important impact on all-cause death, andconduct user portraits for patients of dif erent ages and genders. After comparison, the randomforest algorithm works best, and an important factor af ecting the all-cause death of patients is obtained. Among them, the all-cause death of all patients is related to factors such as albumin, blood potassium, blood magnesium, and urea; With age, the importance of factors such as blood sodium and phosphorus increases, and the importance of factors such as cardiac ultrasound ejection fraction decreases. Finally, there were also dif erences in the importance of analyzing patients of dif erent ages anddif erent sexes af ecting their all-cause death.

KEYWORDS

Maintenance Hemodialysis, All-cause Mortality, Randomized Forest, Feature Importance.

Design and Analysis of All-Optical Logic Gates based on a Germanium Dielectric Material Micro Ring Resonator using Photonics Crystal Technology

Mayur Kumar Chhipa^1*, Vishwa Nath Maurya², B.T.P. Madhav³ and B. Suthar⁴, ^1,3Department of Electronics and Communication Engineering, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, Andhra Pradesh, 522502 India, ²Pro-Vice Chancellor and Distinguished Professor, Faculty of Science and Technology, West Coast International University of Sciences, Technology, Management and Arts (WCIU), Delaware/Washington, D.C., USA, ⁴Department of Physics, MLB Govt. College, Nokha, Bikaner, Rajasthan, India

ABSTRACT

In this paper, plan and examination of optical NOT and OR logic gates based on 2-Dimensional (2D) Photonic Crystals (PhC) design is proposed. The proposed structure is shaped with the combination of line abandons and rectangular ring resonator structure. The execution of the structure is analyzed utilizing 2D Finite Difference Time Domain (FDTD) strategy. The band gap examination is gotten utilizing Plane Wave Expansion (PWE) strategy. The structure has the lattice constant and refractive index as 580 nm and 4.0, respectively. The response time is about 0.306 ps and the dimension of the proposed structure is about 10.6 × 11.6μm2 which is highly compact and integrable. The wide band range is achieved between 1272.0 nm and 1987.4 nm with center wavelength at 1550 nm. All-optical logic gates satisfy their truth tables with reasonable power contrast ratio between logic ‘1’ and logic ‘0’.

KEYWORDS

Photonic Crystals, Band Gap Structures, Optical Logic Gates, FDTD Method, PWE Method, Optical Ring Resonator.

Handwritten Digit Recognition System based On CNN and SVM

Yousra Berrich and Zouhair Guennoun, Smart Communications Research Team - ERSC, Mohammadia School of Engineering, Rabat, Morocco

ABSTRACT

The recognition of handwritten digits has aroused the interest of the scientific community and is the subject of a large number of research works thanks to its various applications. The objective of this paper is to develop a system capable of recognizing handwritten digits using a convolutional neural network (CNN) combined with machine learning approaches to ensure diversity in automatic classification tools. In this work, we propose a classification method based on deep learning, in particular the convolutional neural network for feature extraction, it is a powerful tool that has had great success in image classification, followed by the support vector machine (SVM)for higher performance. We used the dataset (MNIST), and the results obtained showed that the combination of CNN with SVM improves the performance of the model as well as the classification accuracy with a rate of 99.12%.

KEYWORDS

Classification, feature extraction, convolutional neural network, support vector machine, MNIST.

Measurement Study on 5G NSA Architecture over Fading Channel

Bruno S. da Silva and Iury da S. Batalha, Sidia Institute of Science and Technology, Manaus, Brazil

ABSTRACT

The 5G NR network with the Non-Standalone (NSA) architecture aims to advance with regard to throughput. When compared to fourth-generation mobile communication (4G LTE), the 5G has a higher data exchange capability through the gNB and the UE (User Equipment). For evaluation and optimization, it is necessary to carry out practical studies on the behaviour of the system in different environmental conditions, subject to attenuation processes, such as large-scale fading (Shading) and small-scale fading (Multipath propagation). This work has analysed the effect of the MCS (Modulation and Coding Scheme) variation on Throughput/BLER for, initially, a channel degraded by default AWGN, then the analysis extends to the multipath fading effect, which emulates more realistically a mobile communication network. The analysis confirmed the need for robust decision process algorithms in terms of MCS switching to maintain adequate data rates according to the requirement of each scenario with specific QoS (Quality of service), considering both 64 QAM and 256 QAM. The throughput degradation effect was more evident in higher-order modulations due to the higher probability of error inherent in the symbol arrangement. This study can be a key for understanding and developing huge modulation and coding schemes for fifth generation communications.

KEYWORDS

5G, NSA, Fading, MCS, Throughput, Modulation, Coding scheme, BLER, Signal-Noise Ratio.

Clustering method in hierarchical wireless sensor networks

Rahul Das¹, Mona Dwivedi², ¹Research Scholar. Department of Computer Science, Mansarovar Global University, ²Professor, Department of Computer Science. Mansarovar Global University

ABSTRACT

Thousands of resource-constrained sensors are often used in Wireless Sensor Networks (WSNs) to monitor their surroundings, gather data, and send it to distant servers for additional processing. Despite the fact that WSNs are thought of as highly flexible ad-hoc networks, network management in these kinds of networks has been a fundamental challenge due to the magnitude of the deployment and the quality issues like resource management, scalability, and dependability connected with them. Topology management is seen as an effective method to address these issues. The most popular topology management technique in WSNs is clustering, which groups nodes for administration and/or distributes various duties including resource management. Although energy consumption reduction is the primary goal of clustering approaches, other quality-driven goals can also be achieved.

KEYWORDS

wsn, clustering, CH, intra-cluster.

ComputerBank: A Community-based Computer Donation Platform using Machine Learning and NFT

Eric Gong¹ and Yu Sun², ¹University High School, 4771 Campus Drive, Pomona, CA 91768, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA92620

ABSTRACT

Many people donate money to fund organizations, but very rarely do those donors have information about wherethose donations go. Donation platforms are both non-transparent and also leave a large portion of potential donorsunnoticed: gamers [1]. This paper explores the concept of utilizing blockchain technology and its existence as aweb3 token-based platform in order to provide transparency for donation routes, showing donors and othercompanies exactly where donations are coming from and where that money is going. Our application utilizes HTTPrequests in order to greatly increase compatibility, and also uses multiple private key encryptions in order to ensurethat any user data or information and monetary transactions are kept secure and private [2].

KEYWORDS

web3, donation, platform, decentralized.

An Efficient AI Music Generation mobile platform Based on Machine Learning and ANN Network

Jiacheng Dai¹ and Yu Sun², ¹International Department of the Affiliated High School of SCNU, Zhongshan Road West No.1, Guangzhou, China, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA92620

ABSTRACT

The aim of this paper is to provide a solution to the growing need for fresh music to use in media, as adding musiccan greatly enhance the media’s atmosphere and the viewers’ experience [6]. Our solution to this issue was thecreation of a mobile application named MFly that can output music using the sentiment from an inputted message. To test the ef ectiveness of this new music-generating method, an experiment was conducted in which twenty-threeparticipants inputted a message with a positive and negative sentiment each and recorded whether each outputtedmusical piece accurately represented the sentiment from the message [7]. A post-experiment survey was alsoprovided to each of the participants to gauge the convenience and practicality of the application. The resultsindicated that MFly was largely successful at conveying messages into appropriately fitting music. However, thepracticality of the application could use some work, as generating music based on the sentiment does not alwaysseem to match up with the original inputted messages sentiment, especially with messages that have a negativesentiment. Furthermore, feedback from participants indicated that the application could still improve withtheaddition of more features, such as the ability to save the generated music for later use.

KEYWORDS

Machine Learning, AI, Mobile application.

Sentiment Analysis of Different Ecommerce Platforms in Nepal using Machine Learning

Manil Vaidhya, Aayush Bhattarai, Sanjaya Neupane, Department of Mechanical and Aerospace Engineering, Pulchowk, Kathmandu, Nepal

ABSTRACT

There are many e-commerce platforms in Nepal where the users buy product based on the basis of reviews made online. For the analysis of such reviews, sentiment analysis is done. Since the data are in huge numbers, machine learning algorithms are used for the fast and effective calculation and analysis of these product reviews. These reviews can be done quickly using machine learning as the model is created where thousands of reviews are made. For the better accuracy of the model, the datasets are subjected to different pre-processing techniques. Then, machine learning method is used to classify the sentiments of the dataset in positive or negative class. From different analysis of the algorithm, deep learning model exhibited higher accuracy i.e., LSTM model provided 75% accuracy whereas supervised learning model like Naive-Bayes exhibited has 68% accuracy and SVM classifier exhibited 71% accuracy whereas Vader classifier exhibited 68% accuracy.

KEYWORDS

Sentiment analysis, machine learning, deep learning, supervised learning, unsupervised learning, LSTM, VADER, SVM, Naive-Bayes.

An Intelligent System to Automate the Detection of Online Cheating Activities using AI and Context-Aware Techniques

Qinyuhan Zhao¹, Mingze Gao², Yu Sun³, ¹Pacific Academy Irvine, 4947 Alton Pkwy, Irvine, CA 92604, ²University of California, Irvine, 204 Aldrich Hall. Irvine, CA 92697, ³California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

In the environment of online courses and online exams, cheating in online courses is prevalent [1]. To better ensure fairness in exams, schools and educational institutions need to use technology to detect and deter cheating [2]. Starting from practical application, this paper discusses 3 different methods to detect cheating behavior, and proposes a new way for online exam supervision.

KEYWORDS

Machine Learning, Audio Detection, Context-Aware.

Research on Creative Thinking Mode based on Category Theory

Tong Wang, Guangzhou College of Technology and Business, China

ABSTRACT

The research on the brain mechanism of creativity mainly has two aspects, one is the creative thinking process, and the other is the brain structure and functional connection characteristics of highly creative people. The hundreds of millions of nerve cells in the brain connect and interact with each other. The human brain has a high degree of complexity at the biological level, especially the rational thinking ability of the human brain. Starting from the connection of molecules, cells, neural networks and the neural function structure of the brain, it may be fundamentally impossible to study the rational thinking mode of human beings. Humans rational thinking mode has a high degree of freedom and transcendence, and such problems cannot be expected to be studied by elaborating the realization of the nervous system. The rational thinking of the brain is mainly based on the structured thinking mode, and the structured thinking mode shows the great scientific power. This paper studies the theoretical model of innovative thinking based on category theory, and analyzes the creation process of two scientific theories which are landmarks in the history of science, and provides an intuitive, clear interpretation model and rigorous mathematical argument for the creative thinking. The structured thinking way have great revelation and help to create new scientific theories.

KEYWORDS

category theory, creative thinking mode, structured thinking way.

A Gradient Descent Inspired Approach to Optimization of Physics Question

Feihong Liu¹ and Yu Sun², ¹Crean Lutheran High School, 12500 Sand Canyon Ave, Irvine, CA 92618, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

Many people believe that the crouch start was the best way to start a sprint [1]. While it seems intuitive, when the process of running is dissected using specific physical and mathematical mrepresentations, the question of “what is the best starting position” becomes harder to answer [2]. This paper aims to examine this phenomenon through a computer science approach inspired by gradient descent. Specifically, this paper aims to maximize the distance covered by a runner in ten steps. By massuming that every runner has a maximum amount of force that they can exert on each step, that their horizontal velocity is not slowed by friction, among other factors, we will generate a hypothetical environment to study what the best strategy is for reaching the furthest distance within ten steps.

KEYWORDS

Gradient Descent.

An Intelligent Video-based Application to Automate Student Attendance Checking using Computer Vision and Artificial Intelligence

Jasmin Liao¹ and Yu Sun², ¹Trabuco Hills High School, 27501 Mustang Run, Mission Viejo, CA 92691, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

Engaging online students is a challenge for many teachers [1]. While I was a student, I saw teachers struggling to take attendance due to the number of students leaving their classes after attendance. Students would be held responsible for their work using facial recognition technology [2]. To simplify the process of applying absences to students in each class, this paper proposes an application that would allow teachers to stay on top of their work. We applied our software to test “students” in the classroom and used various libraries/csc styles to create a classroom that is easy for both the student and the teacher to read. Our designs are built upon OpenCv and PIL which are used as geometric classifiers to determine if the student is present [3]. We tested several faces to see if the algorithm was suitable for the program. After conducting a qualitative evaluation of the approach, we’ve begun to implement registration, creating new classrooms with different databases, and applying verification. With the addition of HTML code, we were able to create a classroom that is safe, engaging, and easy to use.

KEYWORDS

Facial Recognition, JavaScript, Artificial Intelligence.

Listing2Speak: A Data-Driven Analytical System to Evaluate the E-Commerce Product Listing using Artificial Intelligence and Big Data Analysis

Zihan Xu¹ and Yu Sun², ¹Crean Lutheran High School, 12500 Sand Canyon Ave, Irvine, CA 92618, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

In e-commerce, product pages are important to the success of a website [4]. The ultimate goal of an e-commerce site is sales and it takes a great product page to achieve that. However, today, e-commerce entrepreneurs are confused about how to make their product page more attractive [5]. This paper designs a data-driven analytical system to analyze the relationship between different web page features with sales, in order to give users feedback on how to improve their product web pages [6].

KEYWORDS

Data Analysis, E-commerce, Artificial Intelligence.

A Data-Driven and Collaborative Mobile Application to Assist Sensors using Artificial Intelligence and Machine Learning

Aaron Tse¹ and Yu Sun², ¹Arcadia High School, 180 Campus Dr, Arcadia, CA 91006, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

There are numerous arguments as to what the best GPS software is [1]. However, there are no definitive answers as to which is the best. In this paper, we use multiple GPS applications tracking user locations to determine what GPS is best in terms of tracking users through a mobile app [3]. The app utilizes a GPS as well as a Google Firebase Realtime Database to manage, pinpoint, and track users’ locations [2]. The application is applied specifically to track locations of people that need taking care of, such as the elderly. This will allow concerned caretakers to help keep track and take care of people in need.

KEYWORDS

Android, GPS, Flutter, Firebase.