Accepted Papers

Richer Information Transformer for Object Detection

Shunyu Yao1, Ke Qi1, Wenbin Chen1 and Yicong Zhou2, 1School of Computer Science and Cyber Engineering, Guangzhou University, Guangzhou, 2Department of Computer and Information Science, University of Macau, Taipa, Macau

ABSTRACT

Though CNNs can efficiently decrease local redundancy by convolution within a small neighborhood, the limited receptive field makes it hard to capture global dependency. Alternatively, ViTs can effectively capture long-range dependency via self-attention mechanism which also produce quadratic computation complexity to the image size input. In this paper, we propose a module composed of several groups of convs and activation functions to make up for the lack of comprehensive information in ViTs for extracting features, so that conv and transformer can achieve a complementary advantage. We also introduce channel attention module to capture the channel information, which arising from the frequent manipulations to channels during the calculation process of self-attention. In the absence of pretrained data, our model achieves 40.3 box AP and 37.1 mask AP on COCO object detection task, surpassing state-of-art Swin Transformers backbone respectively by +8.8, +6.7 respectively under the similar FLOPs settings.

KEYWORDS

Object Detection, Vision Transformer, Locally Enhanced, Channel Attention.

PDR-CapsNet: A novel capsule network based on parallel dynamic routing

Samaneh Javadinia and Amirali Baniasadi, ECE Department University of Victoria Victoria, BC, Canada

ABSTRACT

Convolutional Neural Networks (CNNs) have produced state-of-the-art results for solving image classification tasks. However, they are not able to identify rotational and viewpoint invariance. This is because of information loss in max-pooling layers. Capsule Network (CapsNet) uses the dynamic routing algorithm to overcome this problem. Despite the advantages, CapsNet does not reproduce competitive results on several benchmark datasets with complex data. Inspired by the InceptionNet, we propose the parallel dynamic routing network, which is part of a deeper CapsNet and results in higher accuracy while avoiding overfitting and achieving faster training and inference. In this work, we obtain 83.55% accuracy on the CIFAR-10 dataset, while requiring 87.26% fewer parameters and achieving 3x faster inference compared to CapsNet.

KEYWORDS

Capsule Network, CapsNet, Deep Learning, Image Classification.

LEON: Lightweight Edge Detection Network

Nasrin Akbari and Amirali Baniasadi, Department of Computer Engineering, University of Victoria, Victoria, Canada

ABSTRACT

Deep Convolutional Neural Networks (CNNs) have achieved human-level performance in edge detection. However, there have not been enough studies on how to efficiently utilize the parameters of the neural network in edge detection applications. Therefore, the associated memory and energy costs remain high. In this paper, inspired by Depthwise Separable Convolutions and deformable convolutional networks (Deformable-ConvNet), we aim to address current inefficiencies in edge detection applications. To this end, we propose a new architecture, which we refer to as Lightweight Edge Detection Network (LEON ). The proposed approach is designed to integrate the advantages of the deformable unit and DepthWise Separable convolutions architecture to create a lightweight backbone employed for efficient feature extraction. As we show, we achieve state-of-the-art accuracy while significantly reducing the complexity by carefully choosing proper components for edge detection purposes. Our experiments on BSDS500 and NYUDv2 show that LEON, while requiring only 500k parameters, outperforms the current lightweight edge detectors. Note that our results are achieved while training the network from scratch and without using pre- trained weights.

KEYWORDS

Edge detection, lightweight neural network, Receptive field, network pruning.

Genetic Algorithm Based C- Means Fuzzy Clustering for Community Detection

Harsavardini, SRM Institute of Science and Technology, India

ABSTRACT

The networks which exist in the real world have different representations and these networks are exchanging the information between them. The real world structure in the networks can be represented as community structure for the better understanding of the complex structure exhibited by the network. In order to put the information together from the different network, integrating approaches have been developed. For community detection, in complex networks it is important to explore the links. In this paper a novel algorithm using the Genetic algorithm based c-means fuzzy clustering for community detection. It is a scalable algorithm for higher proximity preserving along with community structure. The correctness and convergence are measure as performance parameters. The objective function is defined in terms of c-means fuzzy clustering algorithm. From the experimental study it can be seen that the proposed algorithm is more efficient compared to the existing for community detection.

Smart Waste Management System

Rexsan.S, Tharakan.T, Thivyapranath.B and R.K.Sheron, Department of Information Technology, Sri Lanka Institute of Information Technology, New Kandy Road, Malabe, Colombo, Sri Lanka

ABSTRACT

Waste management is a problem that we are thriving worldwide. Still, a majority of the world is trying to figure out a way to administrate and monitor waste in an efficient way to minimize the outburst of environmental perils. Even though various methodologies and technologies have been used to find a solution still none of them has reached their full potential level. This paper proposes an approach to provide a solution for waste management at its starting point with cost-effective benefits and a higher rate of practicality. With the help of the Internet of things (IoT) and Machine Learning (ML) based Image processing we have developed a handheld device with an integrated app to find a solution for this problem. The device will have the application to monitor the bins environment sending data to the admin to make further actions if necessary, regarding the maintenance of the bin. Furthermore, the bin will be able to communicate with the user when requested. To make the user more involved in using the bin properly the app will provide a rewarding system to encourage the user to keep using the bin properly. With the joined eco-system of all these technologies, waste management and segregation can be done more effectively with less manpower and more awareness and encouragement among the users creating a new era of a clean eco-system.

KEYWORDS

Image processing, Machine Learning, Internet of things.

A Machine Learning/deep Learning Hybrid for Augmenting Teacher-LED Online Dance Education

Catherine Hung, Palo Alto Senior High School, Palo Alto, USA

ABSTRACT

For online dancers, learning a dance move properly without the feedback of a live instructor can be challenging because it is difficult to determine whether a move is done correctly. The lack of proper guidance can result in doing a move incorrectly, causing injury. In this work – we explore the use of a hybrid Deep Learning/Machine Learning approach to classify dance moves as structurally correct or incorrect. Given a video clip of the dancer doing a move, such as the grand plie, the algorithm should detect the correctness of the movement. To capture the overall movement, we proposed various methods to process data, starting with deep learning techniques to convert video frames into landmarks. Next, we investigate several approaches to combining landmarks from multiple frames and training machine learning algorithms on the dataset. The distinction between correct and incorrect grand plies achieved accuracies of over 98%.

KEYWORDS

Deep Learning, Machine Learning, Classification, Online Dance Education.

Machine Learning Chatbot for Sentiment Analysis of Covid 19 Tweets

Suha K. Assayed, Safwan Maghaydah, Khaled Shaalan , Manar Alkhatib, Faculty of Engineering and IT, The British University in Dubai, UAE

ABSTRACT

The various types of social media were increased rapidly as people’s need to share knowledge between others. In fact, there are a various types of social media apps and platforms such as Facebook, Twitter, Reddit, Instagram and others. Twitter remains one of the most popular social application that people used for sharing their emotional states. However, it’s increased particularly during the COVID-19 pandemic. In this paper, we proposed a chatbot for evaluating the sentiment analysis by using machine learning algorithms. The authors used a dataset of tweets from Kaggle’s website, and it includes 41157 tweets that are related to the COVID-19. These tweets were classified and labelled to four categories: Extremely positive, positive, neutral, negative and extremely negative, in this study we applied Machine Learning algorithms, Support Vector Machines (SVM) and the Naïve Bayes (NB) algorithms and accordingly we compared the accuracy between them. On the other hand, the classifiers were evaluated and compared after changing the test split ratio. The result shows that the accuracy performance of SVM algorithm is better than Naïve Bayes algorithm, even though Naïve Bayes perform poorly with low accuracy, but it trained the data very fast comparing to SVM.

KEYWORDS

NLP, Twitter, Chatbot, Machine Learning, Sentiment Analysis, SVM, Naïve Bayes.

The Formation of Verbs of Colour in Algerian Arabic

Amel Beldjenna, Department of English Language, University of Oran 2, Oran, Algeria

ABSTRACT

In this paper, the formation of verbs of colour in Algerian Arabic was analysed morphologically, phonologically, syntactically, and semantically. Seven Algerian Arabic verbs of colour were analysed in their verbal forms. The analysis shows that only three forms out of the nine verb forms in Arabic can be derived from this verb class; form II, IV, and IX. Each of these three forms has two patterns; one for the perfect tense and one for the imperfect. The study also shows that the root based approach is most adequate in analysing such verb classification because Form I does not semantically exist in the language and so cannot be used as an input for the derivation of the other forms as Alshdaifat [4] claims.

KEYWORDS

Verbs of Colour, Algerian Arabic, Verbal Derivation, Morphology.

A Preliminary Study of Model-Generated Speech

Man-ni Chu and Yu-Chun Wang

ABSTRACT

This research makes use of two dictionaries of the ChaoShan dialect spanning about 100 years. The same Chinese characters were selected from each of the dictionaries and their contemporary pronunciations were reported. After manually deleting inconsistencies in pronunciation, three machine learning methods were adopted to train the pronunciation of the words from one dictionary to another. These methods were: the attention-based sequence-to-sequence method; DirecTL+; and Sequitur. 5-fold cross-validation was used to evaluate the results and our model was found to have up to 68% accuracy. We probed how the probability of a sound’s following unit determinates the accuracy of the machine learning methods. Thus, machine learning models are not just related to the frequency of input, but also to the probability of the following unit.

KEYWORDS

attention-based seq2seq, Chaoshan dictionary, DirecTL+, Sequitur.

Lexical Features of Medicine Product Warnings in the Philippines

Shielanie Soriano-Dacumos University of Rizal System, Binangonan, Rizal Philippines

ABSTRACT

Over the stretch of years, the Philippines has been facing numerous medical problems since the public outcry against a „dengue? vaccine. As a result, parents refused their children from having an anti-measles vaccine which created a medical outbreak in the country. Product warnings are found to be in their optimal position in safeguarding the life of consumer-patients. This paper anatomizes the lexical features of medicine product warnings in the Philippines which are crucial in the response discourses. A range of linguistic frameworks were applied and significant findings were drawn. Lapses on the use of noun abstractness, synthetic personalization, field continuum, adjectives, and adverbs were identified. Such an investigation brought up the transparency of communicative features of medicine safety texts. In the end, linguistic components create a vital impact on the legal content adequacy of medicine product warnings, unfolding the vitalities of these messages in facilitating informed decision-making among consumer-patients.

KEYWORDS

medicines, consumer-patients, linguistic components, product warnings.

Knowledge-Enriched Moral Understanding Upon Continual Pre-Training

Jing Qian1, Yong Yue1, Katie Atkinson2 and Gangmin Li3, 1School of Advanced Technology, Xi’an Jiaotong Liverpool University, China, 2Department of Computer Science, University of Liverpool, Liverpool, UK, 3School of Computer Science Technology, University of Bedfordshire, Luton, UK.

ABSTRACT

The aim of moral understanding is to comprehend the abstract concepts that hide in a story by seeing through concrete events and vivid characters. To be specific, the story is highly summarized in one sentence without covering any characters in the original story, which requires the machine to behave more intelligently with the abilities of moral perception and commonsense reasoning. The paradigm of “pre-training + fine-tuning” is generally accepted for applying neural language models. In this paper, we suggest adding an intermediate stage to build the flow of “pre-training + continual pre-training + fine-tuning”. Continual pre-training refers to further training on task-relevant or domain-specific corpora with the aim of bridging the data distribution gap between pre-training and fine-tuning. Experiments are basing on a new moral story dataset, STORAL-ZH, that composes of 4,209 Chinese story-moral pairs. We collect a moral corpus about Confucius theory to enrich the T5 model with moral knowledge. Furthermor, we leverage a Chinese commonsense knowledge graph to enhance the model with commonsense knowledge. Experimental results demonstrate the effectiveness of our method, compared with several state-of-the-art models including BERT-base, RoBERTa-base and T5-base.

KEYWORDS

Moral Understanding, Continual Pre-training, Knowledge Graph, Commonsense

Smart Energy Trading for the Blockchain Miners in the Interconnected Multi-Microgrids

Saeed Alishahi1 and Mohammad Hossein Yaghmaee2, 1Mashhad Electrical Energy Distribution Company, Mashhad, Iran 2Ferdowsi University of Mashhad (FUM), Mashhad, Iran

ABSTRACT

In blockchain networks, each new block should be validated by some especial computers called miners. Knowing that the mining process is a complex and time-consuming operation that consumes too much electrical energy, it will be more beneficial for miner owners to buy their necessary electrical energy from microgrids instead of buying energy from the more expensive power grid. In this paper, we present a peer to peer energy trading algorithm for energy trading in an interconnected multi-microgrid (IMM) system consists of several microgrids. We first evaluate the profitability of Bitcoin miners, as a currency powered by blockchain technology, in terms of electrical energy price, blockchain network parameters and communication network characteristics. Then, we introduce an integer linear program (ILP) optimization problem which is run by the miner controller to optimally decide to switch off nonefficient miners. We present a smart energy trading algorithm for trading energy inside each microgrid and between microgrids in the IMM system. A convex optimization problem to find the optimal values of traded energy between each pair of MGs in the IMM system is presented. The simulation results confirm the superiority of the proposed smart energy trading algorithm for both energy buyers (miner farms) and energy sellers (microgrids).

KEYWORDS

Energy Trading, interconnected multi-microgrid, smart grid, Blockchain, Bitcoin miners.

Link Quality Prediction for Wireless Networks: Current Status and Future Directions

NiuMingxiao1 LiuLinlan2, and ShuJian3, 1,2School of Information Engineering, Nanchang Hangkong University, Nanchang, China 3School of Software, Nanchang Hangkong University, Nanchang, China

ABSTRACT

In wireless sensor networks (WSNs) and smart gird, link quality prediction (LQP) is an important part of routing design. Effective LQP can select high-quality links for communication and improve the reliability ofdata transmission. Therefore, it hasbeen attracting a vast array ofresearch works. Reported works on LQP are typically based on different link quality parameters and consider different models. This article provides a comprehensive survey on related literature, covering the fundamental concepts of LQP, the description of common link quality parameters and a summary of existing LQP models. Based on these investigations,we canprovidereferenceforLQPresearchinthefuture.

KEYWORDS

wireless sensor networks, smart gird, link quality prediction,machine learning.

Network Models of Studies on Parkinson Disease

Fuad Aleskerov1, Olga Khutorskaya2, Vuacheslav Yakuba2, Anna Stepochkina3 and Ksenia Zinovyeva*3, 1National Research University Higher School of Economics, Institute of Control Sciences of Russian Academy of Sciences, 20 Myasnitskaya Str., 101000 Moscow, Russia, 2Institute of Control Sciences of Russian Academy of Sciences, 65 Profsoyuznaya Str., 117997 Moscow, Russia, 3National Research University Higher School of Economics, 20 Myasnitskaya Str., 101000 Moscow, Russia

ABSTRACT

More than 10 million people worldwide are living with Parkinson Disease (PD). We analyze publications on studies of PD over the period from 2015 to 2021. We have collected about 70 thousand papers from 4164 different journals. After the data preprocessing 39811 publications and 3292 journals are left.Methods of centrality analysis have been applied to identify the most significant publications and journals in the field. A citation network for papers has been built. This network is constructed as a directed graph in which the papers are vertices and edges represent citation between them. Similar network has been constructed for journals. The new centrality indices have been evaluated for the whole period and for each year with different vertices’ parameters. It can be noticed that changing the parameters makes it possible to extract groups of specialized journals and fields of study, which intensively cite each other.

KEYWORDS

Citation network, network analysis, centrality indices, Parkinson Disease.

The Importance of MANETs: A Survey of Applications and Challenges

Fatemeh Safari, I. Savi´c, H. Kunze, J. Ernst, and D. Gillis, School of Computer Science, University of Guelph, Guelph, Canada

ABSTRACT

Recent research in mobile ad hoc networks (MANETs) has shown much promise in expanding the uses of wireless networks. MANETs are a collection of mobile and information of things (IoT) devices (nodes) that are used to form a wireless network that can provide network functionality and cloud services in areas where fixed information and communications technology (ICT) infrastructure might not already exist or be nonoperational, damaged, or insufficient to support effective and sustained communication. The network does not require a node to be connected to wireless or cellular infrastructure, but the connection of even just one node to infrastructure opens many doors for additional research topics and applications. Due to the ubiquity of devices that can support such networks, MANETs have been used to improve communication in a variety of applications, however, a full survey of their use has not (to our knowledge) been documented. As such, in this paper, we present a summary of existing applications that have used MANETs in a variety of sectors, with special attention paid to the challenges and opportunities this technology provides.

KEYWORDS

Mobile ad hoc network, MANET, Applications, Challenges, Military, Natural Disaster, Smart City, Emergency Services.

Measuring Performance of HTTP With Updated TCP for Faster Web Browsing

Ziaul Hossain and Department of Computing, University of Fraser Valley, Abbotsford, BC, Canada

ABSTRACT

There are existing solutions which focuses on performance improvement for bulk and interactive applications. But popular Internet applications such as web browsing, web video download or variablerate voice still suffer from those modifications as their transmission rate and pattern are different from the previous ones. These variable-rate traffic poses challenges to the widely deployed Transport Control Protocol (TCP). Previous works have analysed the interaction of these applications with the congestion control algorithms and proposed Congestion Window Validation (CWV) as a solution. But this method was incomplete and have shown drawbacks. Finally, a newer version NewCWV have been proposed with a practical mechanism to estimate of the available path capacity and corresponding congestion control behaviour. This benefits variable-rate applications with shorter transfer durations. This technique was implemented and tested by experiments in the Linux TCP/IP stack, where results indicate that NewCWV improves the performance of Web applications significantly.

KEYWORDS

Network Protocols, HTTP, TCP, Congestion Control, NewCWV, Bursty TCP traffic.

A Traffic Flow Prediction Method based on Spatio-temporal Graph Attention Network

Bailin Li and Mi Wen,School of Computer Science and Technology, Shanghai University of Electric Power, Shanghai, China

ABSTRACT

In the field of traffic forecasting, Graph Neural Networks (GNNs) have achieved remarkable results by modeling traffic flow. However, existing GNNs-based methods rely on the physical topology of road networks in predefined graphs. To more accurately represent the hidden features of traffic flow, a Spatiotemporal Graph Attention Network (STGAT) is proposed for traffic prediction. The graph structure in the network is learned end-to-end from the physical topology of the road network and traffic data, and a more accurate description of the relationships between traffic flows is found. Then, a spatial attention mechanism based on the traffic graph and a temporal attention mechanism based on multiple time periods are proposed to capture the spatio-temporal dependence of the changing traffic data. Finally, predictions are made by merging the hidden state and the original time series. Experimental results on two real datasets demonstrate that STGAT achieves better performance than state-of-the-art methods.

KEYWORDS

Traffic flow prediction, graph attention network, spatio-temporal characteristics

Efficient Data Attack Detection Algorithm in Hadoop

Jing Wen1,2,3, Fengjia Chang2, Ningwei Li1, 1College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics, Jiangsu Nanjing 211106, 2College of Computer Science and Technology Huaiyin Normal University, Jiangsu Huaian 223000, 3Huaian Key Laboratory of Big Data Intelligent Computing and Analysis, Jiangsu Huaian 223000

ABSTRACT

Due to the large-scale and distributed nature of the hadoop cluster, it is difficult to detect internal data attacks in the cluster. A malicious user only needs to attack a node in the cluster to prolong the execution time of the task or destroy the correctness of the task execution result. Although there are currently some studies on detecting data attacks, these studies are only applicable to a specific scenario or hosting data on a third-party trusted platform, but users cannot rely on the reputation of the third party to ensure data security. Ensuring data security through a certain technical mean embedded in the hadoop framework can users be more assured. However, there is currently no general solution for detecting hadoop internal data attacks. Aiming at the shortcomings of the existing internal attack detection schemes such as poor versatility, low efficiency, and low detection success rate, this paper proposes an efficient hadoop data attack detection algorithm. The algorithm reduces the performance overhead caused by attack detection by adding data sampling and result comparison modules in hadoop, and through the analysis of experimental results, the algorithm has a higher detection success rate for data attacks.

KEYWORDS

Big data, data attack, attack detection, hadoop.

Paddy Rice Smart Farming

Hakkin Chethiya Kaushila De Silva, Srisinthujan Shanmuganathan, Mathusan Anantharajah, Sivanujan Sivanathan, Thamali Dassanayake and Sanath Jayawardena, Department of Information Technology, Sri Lanka Institute of Information Technology, Malabe, Sri Lanka

ABSTRACT

It is anticipated that machine learning (ML) and the internet of things (IoT) would significantly impact smart farming and engage the entire supply chain, in particular for the production of rice. Rice smart farming offers new capabilities to foresee changes and find possibilities thanks to the growing amount and variety of data gathered and obtained by emerging technologies in the Internet of Things (IoT). The accuracy of the models created through the use of ML algorithms is significantly impacted by the quality of the data obtained from sensor readings. These three components, machine learning (ML), the internet of things (IoT), and agriculture have been used extensively to improve all aspects of rice production processes in agriculture. As a result, traditional rice farming practices have been transformed into a new era known as rice smart farming or rice precision agriculture. We do a study of the most recent research that has been done on the application of intelligent data processing technology in agriculture, namely in the production of rice, in this paper. We analyze the applications of machine learning in a variety of scenarios, including smart irrigation for paddy rice, predicting paddy rice yield estimation, predicting droughts and floods, monitoring paddy rice disease, and paddy rice sample classification. In each of these scenarios, we describe the data that was captured and elaborate on the role that machine learning algorithms play in paddy rice smart agriculture. This paper also presents a framework that maps the activities defined in rice smart farming, data used in data modeling, and machine learning algorithms used for each activity defined in the production and post-production phases of paddy rice.

KEYWORDS

AI, Deep Learning, IoT, Machine Learning .

Development and Demonstration of Monitoring System for Position and Operation Management of Medical Devices

Kazuto Kakutani1, Nobuhiro Ito1, Kosuke Shima1, Shintaro Oyama2 and Takanobu Otsuka1, 1Nagoya Institute of Technology, Showa, Nagoya, Aichi, 466-8555, Japan, 2Nagoya University, Chikusa, Nagoya, Aichi, 464-8601, Japan

ABSTRACT

In recently years, according to sophistication of Medical Devices (MD), many portable MDs have been used and maintained with central management. However, the central management lends hospital staff the MDs only with managing by a ledger, therefore, missing or subletting may be caused. Furthermore, while the demand for the MDs is increasing due to the COVID-19, there is an issue that it is difficult to operate due to the shortage of clinical engineers against management duties of the MDs. In this study, we develop a power strip device which can measure electricity usage of plugged MD and its position and propose a visualization system for position and operation ratio of the MDs. We implemented 75 developed devices in three hospitals and confirmed that the system was effective to evaluate whether the number of the MDs owned by the hospital is appropriate.

KEYWORDS

Internet of Things (IoT), Wireless Network, Indoor Positioning, Medical Device, Management System.

Sales Forecasting of Perishable Products: A Case Study of a Perishable Orange Drink

T. Musora, Z. Chazuka, A. Jaison, J. Mapurisa and J. Kamusha, School of Natural Sciences and Mathematics, Department of Mathematics, Chinhoyi University of Technology , Pivate Bag-7724, Chinhoyi, Zimbabwe

ABSTRACT

Companies are in a continuous attempt to increase their profits and reducing their costs. Sales forecasting is an inexpensive way to meet the aforementioned goal, since this leads to improved customer service, reduced lost sales, reduced product returns and more efficient production planning. For food industries, successful sales forecasting systems is much importance, due to the reduced shelf-life of food products and the importance of the product quality which is linked to human health. This paper investigates the application of the ARIMA model in forecasting sales of a perishable orange drink. The methodology is applied successfully. (0, 1, 1)(0, 1, 1)12 was concluded as the most appropriate model. Model diagnostics were done and it was realised that no model assumption was violated. Fitted values were regressed against observed values. A very strong linear relationship was evident with an R2 value of over 90% which is very plausible.

KEYWORDS

Sales forecasting, Orange Drink, ARIMA, Model Diagnostics, R2- value.

Keep an Eye Out! A Literature Review on Phishing Susceptibility Assessment Through Eye-Tracking

Noon Hussein, Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, Ontario, Canada

ABSTRACT

As of 2021, it has been reported that around 90% of data breaches occur on account of phishing, while about 83% of organizations experienced phishing attacks [1]. Phishing can be defined as the cybercrime in which a target is contacted through e-mail, telephone or text message by someone impersonating a legitimate institution [2]. Through psychological manipulation, the threat actor attempts to deceive users into providing sensitive information, thereby causing financial and intellectual property losses, reputational damages, and operational activity disruption. In this light, this paper presents a comprehensive review of eye-tracking in association with phishing cyberattacks. To determine their impact on phishing detection accuracy, this work reviews 20 empirical studies which measure eye-tracking metrics with respect to different Areas of Interest (AOIs). The described experiments aim to produce simple cognitive user reactions, examine concentration, perception and trust in the system; all in which determine the level of susceptibility to deception and manipulation. Results suggest that longer gaze durations on AOIs, characterized by higher attention control, are strongly correlated with detection accuracy. Eye-tracking behavior also shows that technical background, domain knowledge, experience, training, and risk perception contribute to mitigating these attacks. Meanwhile, Time to First Fixation (TTFF), entry time and entry sequence data yielded inconclusive results regarding the impact on susceptibility to phishing attacks. The results aid in designing user-friendly URLs, visual browsing aids, and embedded and automated authentication systems. Most importantly, these findings can be used to establish user awareness through the development of training programs.

KEYWORDS

Cybersecurity, eye-tracking, phishing, human factors.

Analyzing and Personalizing the Learning Performance for Special Needs Students Using Machine Learning and Data Analytics

Eric Xiong1 and Yu Sun2, 1Crean Lutheran High school, 12500 Sand Canyon Ave, Irvine, CA 92618 and 2California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620

ABSTRACT

We design a server-client system that collects students’ engagement information and reports it to a centralized server to help teachers assist neurodivergent students in order to provide a visual representation of students’ engagement status aiming to promote an equal learning opportunity for neurodivergent students [6]. In recent years, everyone throughout the globe are all seeking higher education, either for themselves, or for their children. Students are learning an increasing amount in classes and have needed to spend a lot more effort and attention to succeed. In this race for higher education, a specific group of underrepresented minorities has been left behind. This group being the neurodivergent population, specifically high-functioning people with ASD(Autism Spectrum Disorder) [7]. These students often require more attention due to hypersensitivity, and a shorter attention span than the neurotypical populace. These students have all thats necessary to learn and understand the material, although teachers are often stuck to a faster pace curriculum that does not easily allot so much attention to a singular student. Due to this problem many teachers believe that a efficient way to passively gauge these students attentiveness would significantly benefit their education. This paper develops a server-client system that collects students’ engagement information and reports it to a centralized server to help teachers assist neurodivergent students in order to provide a visual representation of students’ engagement status aiming to promote an equal learning opportunity for neurodivergent students. We applied our application to [Class] and conducted an Evaluation of the approach based on the qualitative data collected from the students.

KEYWORDS

Facial features, information collection, Education.

An Integrative App Producing an Optimal Pathforthevessel in Order to Reduce the Impacts of Cargoshipsonthe Environment

Chenyu Zuo1 and Yu Sun2, 1Sage Hill School, 20402 Newport Coast Dr, Newport Beach, CA92657 and 2California State Polytechnic University, Pomona, CA, 91768, Irvine, CA92620

ABSTRACT

Almost every business in the world relies in some way on the shipping industry, whether it is to ship goods ornatural resources, the shipping industry is undeniably the global industry, However, these very ships that drive theeconomy also produce close to 1 billion metric tons of carbon dioxide per year. In this project, we explore the useof machine learning to improve the performance of cargo ships in the ocean by implementing a genetic algorithmAI and a virtual simulation environment. An app was made based on using the training developed by the Ai to be ableto be deployed on cargo ships as part of their navigation system. Once suf icient data regarding a vessel’senvironment was collected, the algorithm could then produce an optimal path for the vessel. Experiments showthat the Ai system could suf iciently adjust to varying conditions and produce optimal paths for vessels.

KEYWORDS

Machine Learning, AI, Mobile APP, environment.

A Belief Revision Mechanism With Trust Reasoning Based on Extended Reciprocal Logic for Multi-agent Systems

Sameera Basit and Yuichi Goto

ABSTRACT

When an agent receives messages from other agents, it does belief revision. A belief revision includes, i) a trust reasoning process, i.e., it obtains new belief related to the messages, and deduces implicitly unknown beliefs from the obtained belief; ii) in the case of contradiction in the belief set, it resolves the contradiction. So, trust reasoning, and belief revision must be included in the decision-making process of an intelligent agent in multi-agent systems. Although a belief revision mechanism with trust reasoning is demanded to construct multi-agent systems, there is no such belief revision mechanism. We, therefore, present a belief revision mechanism with trust reasoning based on extended reciprocal logic for multi-agent systems.

KEYWORDS

Multi-agent systems, trust relationship, trust reasoning, strong relevant logics, belief revision.