A Survey of Networking Cipher Algorithms and how Natural Language Can be Used to Enhance Them

John E. Ortega, Courant Institute of Mathematical Sciences, New York University, New York, New York, USA

ABSTRACT

This work provides a survey of several networking cipher algorithms and proposes a method for integrating natural language processing (NLP) as a protective agent for them. Two main proposals are covered for the use of NLP in networking. First, NLP is considered as the weakest link in a networking encryption model; and, second, as a hefty deterrent when combined as an extra layer over what could be considered a strong type of encryption -- the stream cipher. This paper summarizes how languages can be integrated into symmetric encryption as a way to assist in the encryption of vulnerable streams that may be found under attack due to the natural frequency distribution of letters or words in a local language stream.

KEYWORDS

Networking, Natural Language Processing, Security, Stream Ciphers.

Media Legitimacy Detection: A Data Science Approach To Locate Falsehoods And Bias Using Supervised Machine Learning And Natural-Language Processing

Nathan Ji¹ and Yu Sun², ¹Portola High School, Irvine, CA, 92618, ²California State Polytechnic University, Pomona, CA, 91768

ABSTRACT

Media sources, primarily of the political variation, have a hastening grip on narratives that can easily be constructed using biased views and false information. Unfortunately, many people in modern society are unable to differentiate these false narratives from real events. Utilizing natural language processing, sentiment analysis, and various other computer science techniques, models can be generated to help users immediately detect bias and falsehoods in political media. The models created in this experiment were able to detect up to 70% accuracy on political bias and 73% accuracy on falsehoods by utilizing datasets from a variety of collections of both political media and other mediums of information. Overall, the models were successful as the standard for most natural language processing models achieved only about 75% accuracy.

KEYWORDS

Data Science, Political Bias, Fake News, Supervised Machine Learning and natural-language processing.

Exploring Transformers Models for Emotion Recognition: A Comparision of Bert, Distilbert, Roberta, Xlnet and Electra

Diogo Cortiz, Brazilian Network Information Center (NIC.br), São Paulo, Brasil & Pontifical Catholic University of São Paulo (PUC-SP), São Paulo, Brasil

ABSTRACT

This paper investigates the performance of different Transformers Models for Emotion Recognition, a specific NLU (Natural Language Understanding) task in Affective Computing. We implemented and finetuned different transformers language models (BERT, DistilBERT, RoBERTa, XLNet, and ELECTRA) using a fine-grained emotion dataset with 28 emotions classes Our initial hypothesis was that the model size would not significantly impact this specific task. We evaluated and compared all those models in terms of performance (f1-score) and time to complete. Except for the ELECTRA model, which had the worst F1-score (.33), the other models had more similar results. RoBERTa achieved the best F1-score (.49), followed by DistillBERT (.48), XLNet (.48), and then BERT (.46). However, when we look at the metric of computational cost and time to complete, we argue that the DistillBERT, which is a smaller model, is the most efficient for this type of task.

KEYWORDS

Artificial Intelligence, Natural Language Processing, Natural Language Understanding, Emotion Recognition, Affective Computing.

Deep Multiple Instance Learning for Forecasting Stock Trends using Financial News

Yiqi Deng and Siu-Ming Yiu, Department of Computer Science, The University of Hong Kong, Hong Kong SAR, China

ABSTRACT

A major source of information can be taken from financial news articles, which have some correlations about the fluctuation of stock trends. In this paper, we investigate the influences of financial news on the stock trends, from a multi-instance view. The intuition behind this is based on the news uncertainty of varying intervals of news occurrences and the lack of annotation in every single financial news. Under the scenario of Multiple Instance Learning (MIL) where training instances are arranged in bags, and a label is assigned for the entire bag instead of instances, we develop a flexible and adaptive multi-instance learning model and evaluate its ability in directional movement forecast of Standard & Poor’s 500 index on financial news dataset. Specifically, we treat each trading day as one bag, with certain amounts of news happening on each trading day as instances in each bag. Experiment results demonstrated that our proposed multi-instance-based framework gains outstanding results in terms of the accuracy of trend prediction, compared with other state-of-art approaches and baselines.

KEYWORDS

Multiple Instance Learning, Natural language Processing, Stock Trend Forecasting, Financial News, Text Classification.

CalixBoost: A Stock Market Index Predictor using Gradient Boosting Machines Ensemble

Jarrett Shan Wei Yeo and Chai Kiat Yeo, School of Computer Science and Engineering, Nanyang Technological University, Singapore

ABSTRACT

The potential of machine learning has sustained the interest of both academia and industry in stock market prediction for over the past decade. This paper proposes a stock market index predictor using an ensemble of Gradient Boosting Machines (GBMs) called CalixBoost. The predictor is trained on data comprising macro-economic metrics, technical financial indicators and sentiment analysis of social media using a simple and fast but effective rule-based model. The proposed model is evaluated using a holdout method, viz. on two separate test datasets whose datapoints are collected under (i) normal economic activity and (ii) during a black swan (financial downturn). Experimental results show that our model outperforms existing methods and can achieve a high prediction accuracy and the model can generalize well under different economic situations.

KEYWORDS

Gradient Boosting Machines, Time Series Prediction, Game Theory, Ensemble, Bayesian Optimization.

An Introductory Review of Spiking Neural Network and Artificial Neural Network: From Biological Intelligence to Artificial Intelligence

Shengjie Zheng^1,2, Lang Qian³, Pingsheng Li⁴, Chenggang He², Xiaoqi Qin⁵ and Xiaojian Li², ¹University of Chinese Academy of Sciences, Beijing, China, ²Brain Cognition and Brain Disease Institute (BCBDI), Shenzhen-Hong Kong Institute of Brain Science, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ³Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, China, ⁴McGill University, Montreal, Canada, ⁵Beijing University of Posts and Telecommunications

ABSTRACT

Stemming from the rapid development of artificial intelligence, which has gained expansive success in pattern recognition, robotics, and bioinformatics, neuroscience is also gaining tremendous progress. A kind of spiking neural network with biological interpretability is gradually receiving wide attention, and this kind of neural network is also regarded as one of the directions toward general artificial intelligence. This review summarizes the basic properties of artificial neural networks as well as spiking neural networks. Our focus is on the biological background and theoretical basis of spiking neurons, different neuronal models, and the connectivity of neural circuits. We also review the mainstream neural network learning mechanisms and network architectures. This review hopes to attract different researchers and advance the development of brain intelligence and artificial intelligence.

KEYWORDS

Spiking Neural Networks, Brain-Inspired Intelligence, Deep Neural Networks, Artificial Intelligence and Biological Intelligence.

Automid: A Novel Framework for Automated Computer Aided Diagnosis of Medical Images

Ayeshmantha Wijegunathileke and Achala Aponso, Department of Computing, Informatics Institute of Technology, Sri Lanka

ABSTRACT

Machine Learning, a subtype of AI, enables computers to mimic human behaviour without explicit programming. Machine learning models arent used very often in diagnostic imaging because there isnt enough knowledge and resources to do so. Hence, this study aims to apply automated machine learning to the diagnosis of medical images to make machine learning more accessible to non-experts. In this study, a dataset containing 2313 images each of covid-19, pneumonia and normal chest x-rays were selected and divided into testing, training, and validation datasets. The AutoGluon library was used to train and produce a model that would classify an input image and infer the probable diagnosis from the diseases it was trained upon. This study can prove that applying hyperparameter optimization and neuralarchitecture search is able to produce high accuracy models for medical image diagnosis.

KEYWORDS

Automated Machine Learning, Hyperparameter Tuning, Neural Architecture Search, Medical Imaging.

Transformer based Ensemble Learning to Hate Speech Detection Leveraging Sentiment and Emotion Knowledge Sharing

Prashant Kapil¹ and Asif Ekbal², ¹Department of Computer Science and Engineering, IIT Patna, India, ²Department of Computer Science and Engineering, IIT Patna, India

ABSTRACT

In recent years, the increasing propagation of hate speech on social media has encouraged researchers to address the problem of hateful content identification. To build an efficient hate speech detection model, a large number of annotated data is needed to train the model. To solve this approach we utilized eleven datasets from the hate speech domain and compared different transformer encoder-based approaches such as BERT,and ALBERT in single-task learning and multi-task learning(MTL) framework. We also leveraged the eight sentiment and emotion analysis datasets in the training to enrich the features in the MTL setting. The stacking based ensemble of BERT-MTL and ALBERT-MTL is utilized to combine the features from best two models. The experiments demonstrate the efficacy of the approach by attaining state-of-the-art results in all the datasets. The qualitative and quantitative error analysis was done to figure out the misclassified tweets and the effect of models on the different data sets.

KEYWORDS

BERT, Multi-task learning, Hate speech, Transformer, Ensemble.

Identifying a Default of Credit Card Clients by Using a LSTM Method: A Case Study

Jui-Yu Wu¹ and Pei-Ci Liu², ¹Department of Business Administration, Lunghwa University of Science and Technology, Taiwan, ²Department of Business Administration, Lunghwa University of Science and Technology, Taiwan

ABSTRACT

Detecting fraudulent transactions are critical tasks and a challenge for financial banks and institutes. This study used a deep learning technique, which is a long short-term memory (LSTM) method, for identifying a default of credit card clients (an imbalanced dataset). For the LSTM approach, this study employed three training algorithms based on a gradient method, such as adaptive moment estimation (Adam), stochastic gradient descent with momentum (Sgam) and root mean square propagation (Rmsprop). This study used 10-fold cross-validation. Also, this study compared the numerical results of the LSTM method with those of supervised machine learning methods, which are back-propagation neural network (BPNN) with a gradient descent algorithm (GDA) and a scaled conjugate gradient algorithm (SCGA). This study compared the performance of LSTM- Adam, LSTM-Sgam, LSTM Rmsprop, BPNN-GDA and BPNN-SCGA. Numerical results indicate that the LSTM-Adam method and the BPNN-SCGA have identical performance, and that selecting an appropriate classification threshold value is important for an imbalanced dataset. Based on the numerical results, the LSTM-Adam method can be considered as an effective classifier for dealing with credit scoring problems, which are binary classification problems.

KEYWORDS

Deep Learning, Machine Learning, Long Short-Term Memory, Back-Propagation Neural Network, Credit Scoring.

Comparative Education in the contribution to the construction of the Teaching and Learning Process

Lúcio Rodolfo Rosa¹ and e Wesley Gomes Feitosa², ¹University of Vale do Paraiba (UNIVAP), São Paulo, Brazil, ²Lutheran University of Brazil (ULBRA), Rio Grande do Sul, Brazil

ABSTRACT

The article brings the conception of the current situation of comparative education as a branch of studies and the presence of comparative thinking in the talk about education in the community and society in general. It starts with some examples of where comparative thinking is present today, and then examines the course of comparative education studies and its role in international organizations and in the formulation of public education policies around the world. Since its inception, comparative education has been concerned with observing and describing the educational systems of countries seen as more advanced, with the purpose of offering subsidies for the reform of national school systems and, at the same time, to constitute a science of education. Today, Comparative Education can follow new paths in the face of studies and comparisons, drawing objectives of social equity and promoting competitiveness, and can bring a consensus of regional or social characterization, specific to each civilization.

KEYWORDS

Learning; Comparative education; Educational system.

A Study on the Advantages and Disadvantages of Online Learning in Thailand

Miss Pitchsinee Oimpitiwong Triam Udom Suksa School, Bangkok, Thailand

ABSTRACT

This paper investigates students online learning experience during COVID-19, specifically aiming to identify points of improvement within the current distance-learning infrastructure in Thailand. The research consolidates students’ opinions toward online learning; namely their ease in adapting to the new learning environment, which depends not only on each students learning style but also on their teachers as well as social and economic factors. Identifying the advantages and disadvantages of learning from home, the research presents students needs and suggestions for improvement. As such, this work may guide future adjustments to online learning.

KEYWORDS

Learning Online, Students, Factors, Advantages, Disadvantages.

Flipped Classroom Model for Assisting Unprivileged students During COVID-19

Suroyo¹ and Bima Maulana Putra², ¹Faculty of Teaching and Education, Universitas Riau - Indonesia, ²Faculty of Social Sciences and Humanities, Universiti Kebangsaan Malaysia – Malaysia

ABSTRACT

Avoiding COVID-19 causes turmoil, resulting in the collapse of key government offices. Individuals tried to regard instruction as one of the most important things to consider, and many schools and instruction offices have closed as a result of widespread. Regardless, online courses are widely regarded as a means for individuals to learn at their own pace. Online learning that employs technology as a platform for lecturing is known as flipped classroom. The video or podcast that lecturers send out a day before class is said to help underprivileged students better understand the material, allowing students who are unable to attend class to receive a lesson. The goal of this study is to show how a flipped classroom can benefit lecturers who are teaching underserved students by providing them with additional resources. The method of this research is descriptive qualitative, and the subject is 185 (participants who completed the survey). The outcome demonstrates that flipped classrooms use technology like video platforms and podcasts to provide many learning opportunities for underprivileged children. However, this is thought to provide an opportunity for all underprivileged students by evenly distributing subjects so that underprivileged kids can learn whenever they want.

KEYWORDS

Education, Flipped Classroom, Lecturers, Unprivileged Students.

Video Content Development Guides based on Teaching Experiences

Zolzaya Badamjav¹ and Uranchimeg Tudevdagva², ¹Department of Didactics, Mongolian National University of Education, Ulaanbaatar, Mongolia, ²Faculty of Computer Science, Chemnitz University of Technology, Chemnitz, Germany

ABSTRACT

This paper describes research study on video content development. Due to the COVID-19 pandemic all kind of education and training switched from traditional classroom teaching to online and distance teaching. High impact of online teaching will stay as one of main teaching and learning mode for higher education system. The challenge of online teaching in higher education is to prepare learning materials for students with corresponding quality in various types. The video contents are one of important type of teaching and learning materials. This is one of most welcomed learning materials by students during online and distance teaching. Advantages of video contents are easy to follow focus of lesson, can hear and watch simultaneously, or just can hear if want, or just can watch if not possible to hear, more realistic, gives feeling like takes lesson in classroom. But, to prepare video contents requests a lot of time and preparation. It needs corresponding skills from teacher and it is costly. To support video content with high quality can be offer well defined guidance which helps to prepare good video contents. In this study authors are explained experience-oriented guidance for video content development.

KEYWORDS

Online teaching, distance teaching, higher education, quality of video lessons, SURE model, evaluation.

Tribes and Education in India

Vrishali, School of Development Studies, Tata Institute of Social Sciences, Mumbai, India

ABSTRACT

Of the several issues faced by the tribals despite development-oriented policies and programmes introduced and implemented by the government in India, the right to education still evades the tribals in the country, placing them in a precarious position, lower than the mainstream caste or non-Hindu population. The formality of the modern set up, which was mostly aimed at producing bilingual bureaucrats and directed towards administrative conveniences in difficult terrains of remote and isolated tribal areas, sadly and quite obviously hardly took into account the needs or considered the conventionalities of the already existing knowledge systems in India. This paper attempts to study the effects of colonial education and the Hindu caste system on tribes in India. Traversing through the realms of internal and external colonisation, the study, referring to the gap between policy and practice, argues that it is essential to recognise this as a conflict between individualism and tribalism.

KEYWORDS

Tribal education, inclusion, social justice, internal and external colonisation, gap between policy and practice.

The Effectiveness of English for Spesific Purposes in a Biology Class

Tatu ZakiyatunNufus, English Department of University Muhammadiyah Jakarta

ABSTRACT

English is important to learn because English is the official language most commonly used in the international world. At least one in five people can speak at least English. English is the official language of 53 countries in the world. English is the language of science, aviation, computers, diplomacy, and tourists. Obviously there is a connection with biology, because biology is a science. Many scientists who come from abroad who use English in explaining their findings, but in biology use more Latin than English. This indicates that English courses are really needed by biology students to fight in competition in the era of globalization. Competition in looking for work, knowing and understanding English will increase opportunities to get good jobs in multinational companies in the country and abroad. In addition, English is very important to learn for science students because in science lessons and materials there are many theories written in English.

KEYWORDS

Teaching English, Science Students, Biology Subject, English for Specific Purposes.

Development and Validation of Simulation-based Instructional Materials on Central Dogma of Molecular Biology for Senior High School

Junar S. Cano, Integrated Basic Education Department-Senior High School, Notre Dame of Marbel University, Koronadal City, Philippines

ABSTRACT

Despite the advantages proffered by technology in science education, little has been done to develop and validate innovative technology-based instructional materials on Central Dogma of Molecular Biology. It is imperative to test any newly generated instructional materials to validate their quality before being widely used. Hence, this study aimed to validate the developed simulation-based instructional materials on Central Dogma of Molecular Biology for Senior High School. This study utilized Research and Development (R&D) design involving 50 Grade 12 STEM learners and 15 Biology education experts chosen through purposive sampling. Results revealed that experts strongly agreed that the developed simulation-based instructional materials on Central Dogma of Molecular Biology have content (M = 4.78 ± 0.36), technical (M = 4.88 ± 0.18), and instructional (M= 4.84 ± 0.16) qualities. Meanwhile, the pretest and posttest results revealed that the learners demonstrated significant conceptual improvement from approaching proficiency to advanced mastery level in Central Dogma of Molecular Biology concepts. Further, the learners pretest and posttest mean scores on the concepts differed significantly (p<0.05). Therefore, it is recommended that the developed simulation-based instructional materials be used to complement in teaching Central Dogma of Molecular Biology.

KEYWORDS

Instructional materials, Computer simulations, Science education, Development, Validation.

Online Reading Strategies: Examining Undergraduate Students Preference

Franklin Jose Almonte Butial and Ericson Olario Alieto, College of Teacher Education, Western Mindanao State University, Zamboanga City, Philippines

ABSTRACT

The study aims to determine the online reading strategies employed by 127 university students. In addition, the study attempts to know the frequency of use of the overall reading strategies and when they are classified into three different categories. It also explored significant differences in respondents’ utilization of the overall reading strategies and for individual strategy use when they are grouped according to gender. Results reveal that the overall reading strategies employed by the respondents when reading academic texts are high in level. As for individual strategy use, support reading strategies ranked first, followed by problem-solving and global strategies. Furthermore, female readers showed a higher frequency of using all the reading categories compared to male readers. Finally, there is a significant difference between male readers and female readers in their use of global reading strategies while no significant difference is found from other strategies.

KEYWORDS

Reading strategies, online reading, global reading strategy, problem-solving strategy.

The Development and Application of the Problem based Learning Model in China’s Education

Maojia Sun¹ and Weijia Sun², ¹Faculty of Educational Studies, Universiti Putra Malaysia, Selangor, Malaysia, ²School of Business and Economics, Universiti Putra Malaysia, Selangor, Malaysia

ABSTRACT

The Creative Teaching Model is a novel model of teaching and learning. In this paper, the development and application of the problem-based learning (PBL) model in China’s education will be chosen as the topic of discussion. The paper will describe the elements of the PBL model and the issues related to the innovative application of the model in the Chinese school curriculum to develop a creative teaching and learning model and make recommendations for this creative teaching and learning model.

KEYWORDS

Creative, Creative teaching model, Problem based learning, Education.

using assistive technology to promote adult education and learning among learners’ with disability in an academic setting

Florence Olufunbi, Adeyemo B.Sc(Osun), M.Ed(Ib), Department of Adult and Non-Formal Education, Federal College of Education(Special) Oyo

ABSTRACT

The marginalization and social exclusion of learners’ living with disabilities from meaningful participation undermine their prospects to learn, grow and develop. Leaners disability issues has significantly grown across every spheres of life. Society classified individuals with disability as severely disabled exposing them to discrimination and denying them access to quality education and all. Learners, regardless of their disability deserves nothing less but quality education and training that will provide them with opportunities for life-long learning, the world of work and meaningful participation in society and to become productive citizens. The use of technologies is a key requirement to help adult learners with disability to develop independent thinking skill, self-reliance, increase autonomy, develop problem-solving skills, facilitate a sense of continuity and become more actively involved in their educational activities at home, schools and community. Hence, the paper examine using assistive technology to promote adult education and learning among learners with disability.

KEYWORDS

assistive technology, Adult Education, Learners, and disability.

IoT-based Vital Signs Monitoring System with GPS Tracker and Alert System for the Elderly

Allyssa Kimberly Gabion, Jefferson Pons and Jessie Jaye Balbin, School of Electrical, Electronics, & Computer Engineering, Mapúa University, Manila, Philippines

ABSTRACT

A wide variety of vital signs monitoring systems are already available in the market. However, accuracy is a big issue when it comes to these devices. There have also been few studies about wearable vital signs monitor, which can send out alerts when the device detects an abnormality in the users vital signs. This study mainly focuses on the monitoring of the vital signs, such as temperature, pulse rate, oxygen saturation, and blood pressure, as well as creating a mobile application that may be used by the guardian of the wearer in the monitoring of the vital signs, tracking of the wearers location, and alerting if the vital signs are out of normal range. The researchers have successfully created a device with the features mentioned and based on the results; the device has achieved an accuracy of 95% and above.

KEYWORDS

IoT, vital signs, monitoring system, GPS tracking, alert system.

An IoT-based Vending Machine using Blockchain for Enhanced Security

Sheldon Henriques, Shaun Lewis, Gautam Kotian, St. Francis Institute of Technology (Mumbai University) Mumbai, India

ABSTRACT

Internet of Things (IoT) is a well-established technology, and it is influencing our day-to-day lifestyle, the threat of these IoT devices being misused also increases. To protect IoT devices from being hacked blockchain technology which provides strong security is used, helping solve the drawbacks of an IoT system. The automated vending machine works by distributing a variety of vended goods and helps provide disintermediation. The main aim of the vending machine is to increase the swiftness of customer service and to reduce human contact. In this paper, the development of a system for cashless payment is described. This helps vending companies with problems like cash handling, fake or slug currency, encashment, and change. The infrastructure of IoT vending machines which is based on the blockchain network is discussed in detail and at last, a model has been provided for the security of the internet of things using blockchain.

KEYWORDS

Blockchain, IoT, Vending Machine.

Review on Vulnerabilities, Attacks and Countermeasures for the Internet of Things in WLAN

Tahani M. Alshammari, Department of Computer and information sciences, Aljouf University

ABSTRACT

Internet of things (IoT) and Wireless Sensor Networks (WSNs) are impacting our society in a multitude of ways, and have a wide range of applications in our surrounding environment. Wireless Local Area Networks (WLAN) are considered one of the cheapest ways to enable network connection in the Internet of Things (IoT). The maintenance operations associated with WLAN are easier than other ways to ensure the appropriate network connection. However, an attack on IoT devices can be both easy and have destructive consequences because hackers can control the local wireless network if the appropriate security measures that prevent such intrusions are not taken. Since the security mechanisms used are considered expensive and difficult to provide in full, it is better to study the network in terms of weaknesses and security risks, and then check whether the security mechanisms used are sufficient to repel potential attacks, and in the case of the opposite, these mechanisms must be upgraded. In this work, we will focus on the use of WLANs in the Internet of Things, and accordingly, we will describe various vulnerabilities, attacks, and countermeasures related to wireless LANs.

KEYWORDS

Wireless LANs, Internet of Things (IoT), Wireless Sensor Networks (WSNs), CIA (confidentiality, integrity, and availability), Attacks, and Vulnerabilities.

Blockchain-based Price Priority Matching Two-Stage Energy Trading Algorithm

Qingyu Meng and Zhaogong Zhang, Department of Computer Science and Technology, Heilongjiang University, Harbin, China

ABSTRACT

With the rapid development of energy Internet, the implementation of smart meters and distributed energy resources (DER) provide new opportunities for consumers. Distributed energy needs a new settlement mechanism. Blockchain technology can provide a means for the electricity trading market. This paper proposes a decentralized, two-stage double auction energy trading algorithm. The proposed market is a distributed and decentralized application and its rules may be specified through smart contracts. Market participants conduct energy transactions by interacting with smart contracts. The smart contract first collects the quotations and performs the same price priority matching, and then performs a unified price clearing on the remaining quotations to obtain the market clearing price. Compared with the traditional double auction, our method improves the consumption capacity and security, improves the economic income of the overall market, and makes the profits of producers and consumers higher.

KEYWORDS

Blockchain, Smart contract, double Auction, Smart Grids.

An Intelligent sensor mobile phone assisting systemusingAIand Machine Learning

Ruilang Liang¹, Yu Sun², ¹Brea Olinda High School, 789 N Wildcat Way, Brea, CA 92821, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA92620

ABSTRACT

Technology is taking over the world [7]. Thus, how elderly people can request for help when they use a mobiledevice if there is anybody around them? In this paper, we address this issue by providing a systemthat can shareand remote control a mobile device in real time [8]. An Android mobile app has been developed as an assistant tool. Thus, when a user needs help, she/he uses an unique ID, sends a request and shares the mobile screen, so the helpersees the sharing screen in his/her device and assists the person who needs help. We applied our application todataanalysis and accurate measurements. For the accurate measurement, we conducted diverse experiments to observethe stability of use in dif erent devices, and the influence of geographic, environmental, and network factors. Theresult shows there are no interrupts during the 30 experiments, which means that the system is stable for use andthenetwork speed is the main factor which af ects the average connection delay. For the data analysis, we advertisedthe Mobile App in communities and schools and received a total of 20 feedback questionnaires. We observe that users from 66 - 70 yield the highest positive score.

KEYWORDS

Machine Learning, Screen Remote Sharing, Mobile APP.

An Intelligent Mobile Application for Depression Relief using Artificial Intelligence and Natural Language Processing

Zhishuo Zhang¹, Yu Sun², Ryan Yan³, ¹Arnold O. Beckman High School, 3588 Bryan Ave, Irvine, CA 92602, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620, ³California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

“What is an simple yet effective method to improve the mental health of individuals?” is the question that we chose to tackle [7]. The solution that we came up with was having a deep conversation with another person. From personal experiences, having deep conversations with another person seemed to be one of the most effective ways to keep someones mental health issues under control and maintain a more positive outlook on life. Sharing similar experiences with another person can demonstrate to people that they are not alone and there is always someone who can relate to them and lead them down the right path.In order to provide people with an easier method to have deep conversations with one another, we decided to create an application called Affinity, which was developed using Flutter [8]. In this application, users with various mental health issues will be able to talk with other users who have shared similar experiences. Users can connect to each other based on similar mental health issues, and they can engage in deep conversations with one another through a chat messaging system. We tested the results by providing twelve participants with two surveys. One survey measured a self-given score regarding the participant’s levels of stress and anxiety before using Affinity as well as after one week of using Affinity, and the other survey asks participants to tally the number of conversation partners that shared at least one mental health issue or experience with them compared to the total number of conversation partners. The results we have found are that daily usage of this application will generally reduce levels of stress and anxiety, and the majority of the individuals that the application will offer as conversation partners will be able to connect to a user through at least one additional similar shared experience or mental health issue.

KEYWORDS

Artificial Intelligence, NLP, Mobile Application.A Context-Aware Vocabulary Management and Reading Assistance System using Machine Learning and Natural Language Processing

Zhanhao Cao¹, Yu Sun², ¹Troy High School, 2200 Dorothy Ln, Fullerton, CA 92831, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

Through the increase in the popularity of online reading, many people rely on online dictionaries to further understand the text [1]. However, looking up a word manually is a great inconvenience as well as a form of distraction [2]. This paper develops a chrome extension to automatically detect the difficult words for each user, and provide the words’ associated definition with a mouse hover. The chrome extension can be customized by adding and removing personal difficult words and personal easy words [3]. Also, the chrome extension offers a deeper level of analytic, including the system analyzing part of speech of the world, to further understand the definition of a selected word or sentence. The chrome extension is applied to a school/work setting in order to improve the working efficiency by providing a simple model to analyze the word definition; it is also useful for casual reading, especially to those that aren’t fluent in English. Following the strict SDLC model, the end of the testing stage reflects that most of the users gave positive feedback to the chrome extension with most of the comments centered around convenience and accuracy [4]. Through alpha testing and a small sample of beta testing, the Chrome extension presents productivity improvement on difficult texts.

KEYWORDS

Chrome Extension, NLP, Cloud Computing.

Combining Process Mining and Machine Learning Results in Better Analysis for Predictions of Adult Sepsis Events

Hilda B. Klasky, Lauren Flemmer, Lingtao Chen, Hong-Jun Yoon, 1Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA

ABSTRACT

Sepsis, a bacterial illness in the body, accounts for a sizable amount of hospital spending. According to the Centers for Condition Control, at least 1.7 million people in the United States suffer sepsis each year, and around 270,000 people die as a result of the disease. Analyzing and forecasting sepsis patient discharges will assist the hospital in better managing its teams, resources, schedules, and expenditures. The aims of this study were to (1) better understand adult sepsis incidents as they progressed through the healthcare system and (2) build machine learning models to predict patient discharges. We effectively deduced the sepsis patient clinical care process in a hospital using information technology records, and our best-implemented model attained a classification accuracy of 99%, which exceeded our expectations. Herein, the details of the implementation and outcomes of Logistic Regression models classification of events are presented.

KEYWORDS

Machine Learning, Process Mining, Healthcare, Logistic Regression, Sepsis.

AI_Birder: An Intelligent Mobile Application to Automate Bird Classification using Artificial Intelligence and Deep Learning

Charles Tian¹ and Yu Sun², ¹University High School, 4771 Campus Dr. Irvine, CA 92612, ²California State Polytechnic University, Pomona, CA, 91768, Irvine, CA92620

ABSTRACT

Birds are everywhere around us and are easy to spot. However, for many beginner birders, identifying the birds is ahard task [8]. There are many apps that help the birder to identify the birds, but they are often too complicatedandrequire good internet to give a result. A better app is needed so that birders can identify birds while not dependingon internet connection. My app, AI_Bider, is mainly built in android studio using flutter and firebase, and the AI engine is codedwithTensorFlow and trained with images from the internet [9]. To test my AI engine, I made six dif erent prototypes, each having a dif erent number of times that the code will train from the dataset of pictures. I then selected 5 birdsthat are in my dataset and found 5 pictures on the internet for each of them, which I then uploaded to the app. Myapp will then give me 3 bird species that most closely resemble the image, as well as the app’s confidence initschoices, which are listed as percentages. I recorded down the percentages of accuracy for each picture. After takingthe average percentage of all the models, I selected the most successful model, which had an average percent of accuracy of 79%.

KEYWORDS

Machine Learning, AI platform, Computer vision.