keyboard_arrow_up
Accepted Papers
An Explainable Template-based Question Answering Approach Over Knowledge Graphs

Enayat Rajabi1, Geethu Susan Ebby2 and Somayeh Kafaie3, 1, 2Cape Breton University, Sydney, NS, Canada, 3Saint Mary’s University, Halifax, NS, Canada

ABSTRACT

With the emergence of knowledge graph technologies, the Semantic Web community encourages users to leverage the knowledge graphs and their underlying reasoning power. Question-answering systems can be built over the knowledge graphs and provide end-users an environment to ask questions in natural language instead of SPARQL or Cypher queries. This work adopts a template-based strategy to match the natural language to Neo4j Cypher queries by setting it apart from conventional automated translation practices. In this approach, we use a language model (BERT) to identify the context of questions and then match it with a set of templates created by domain experts. It also uses a knowledge graph API (Neo4j) to pose the query over a disease knowledge graph and visualize the results to users with an explanation.

KEYWORDS

Knowledge graph, Question answering, Explainability, Neo4j, Disease database.


A Smart Color Resolution Assist System With Augmented Reality Glasses for Colorblind Patients Based on Artificial Intelligence and Machine Learning

Anzhuo Zheng1, Andrew Park2, 1Ruben S. Ayala High School, 14255 Peyton Dr, Chino Hills, CA 91709, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

Color blindness impacts 1 in 12 men and 1 in 200 women, with an estimated 300 million people having the deficiency[1]. There is a growing need for better and more accessible technologies that can help people with color blindness go about their day without being hindered by their condition. We propose creating colorblind glasses assisted by AI technology created from easily accessible parts that can help them identify important objects as well as labeling what color they are. Our prototype consists of a Raspberry Pi Zero 2W along with an I2C display and camera to power a wearable that feeds the camera feed to our custom AI model which then labels and annotates the frames before showing the result to the user who is looking at the display [2]. Achieving this required us to resolve a variety of challenges such as properly setting up and connecting the display while also being fast and efficient for the sake of power consumption and performance as well as creating custom cron jobs to automate as much of the experience as possible to make the experience more seamless. Experimental testing of the AI in a variety of settings show promising results, however there is still an issue when it comes to certain objects as well as a general tendency to have a notable degree of false negatives [3]. We believe that assistive technologies in a format like ours is important for people with colorblindness as it will help them avoid unnecessary confusion and help them in their daily lives.

KEYWORDS

Color Blindness, Artificial Intelligence, Internet of Things, Augmented Reality.


An Intelligent Mobile Program to Generate Scripts and Images Using Artificial Intelligence

Jessica J Lin1, Tongchen He2, 1Bellevue High School, Wolverine Way, Bellevue, WA 98004, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

This paper explores the problem of creative block in writing and art and aims to improve the efficiency and inspiration of brainstorming through AI-powered script and image generation [7]. We propose a solution that integrates artificial intelligence into a user-friendly app, capable of generating diverse scripts and images to inspire creativity. Our program leverages natural language processing and image synthesis algorithms to create original content from user prompts [8]. Challenges encountered included ensuring the coherence and originality of AI-generated scripts and optimizing the speed of content generation. By refining the algorithm and streamlining backend processing, we resolved these issues. Experimentation across different user groups highlighted the app’s versatility in supporting creativity for writers, artists, and casual users alike. Key results showed significant improvements in creative productivity and user satisfaction. Ultimately, our solution stands out for its ability to efficiently provide inspiration and simplify the creative process, making it a valuable tool for professionals and hobbyists alike.

KEYWORDS

Script Generator, Image Generator, FlutterFlow, AI.


Classification Models for Sentiment Analysis

John Tsiligaridis,Department of Math and Computer Science, Heritage University, Toppenish, WA,USA

ABSTRACT

Sentiment analysis (SA) is a Natural Language Processing (NLP) method that helps identify the emotions in text. It is the automated process of identifying and classifying emotions in a text as positive, negative, or neutral sentiment. In this way companies can understand the customers’ sentiments, improve their products and services accordingly, and determine their strategies that would work. The need to discover the algorithm with the best classification performance is obvious. To this end, two different approaches for Sentiment Analysis problems are presented. The first one is based on Machine Learning (ML) models and the second one on Deep Learning (DP) models. Most ML models are flexible depending on their classifier hyperparameters and provide competitive accuracy levels but not all of them. Logistic Regression (LR), Random Forests (RF) of ML and the various models based on Neural Networks (NNs) of DL are applied. Useful results are obtained. Measures for classifiers’ effectiveness are also provided.

KEYWORDS

Random Forest, Machine Learning, Deep Learning.


Multi-cloud 6g Networks Orchestration-based Throughput Maximization on Uav-assisted Wireless Sensor Networks

Betalo Mesfin Leranso, Supeng Leng, and Wondemu Woyo Ganebo, University of Electronic Science and Technology of China (UESTC), China

ABSTRACT

The management and coordination of unmanned aerial vehicles (UAVs) in wireless sensor networks (WSNs) is achieved through the use of multiple cloud computing platforms. To achieve redundancy, scalability, and efficient resource allocation, multiple cloud providers are used and the potential of 6G networks is exploited to enable high-speed, low-latency communication between UAVs and WSN nodes. However, UAVs have limited energy and poor quality of service (QoS), consume a large amount of energy, and cause network latency problems when collecting data from multiple WSNs. This paper considers a multi-agent distributed deep deterministic policy gradient (MAD3PG)-enabled UAV-supported WSN for agricultural monitoring in rural areas. The UAVs are equipped with energy harvesting (EH) techniques to provide sustainable power to the UAVs and SNs. In particular, we formulate the data collection from the SNs and the EH from the environment as joint optimization problems to maximize the energy-efficient data throughput (EEDT) network performance and quality of service (QoS) in a large-scale WSN. To solve the dimensional constraints and complex optimization problems, we propose a MAD3PG algorithm to find the optimal data collection and EH decisions using multiple UAVs’ sensor scheduling techniques. The simulation results show that our proposed scheme outperforms the benchmark schemes in terms of the average number of UAVs, offloading capacity, and improving system reliability.

KEYWORDS

Unmanned aerial vehicles, multi-cloud 6G network orchestration, deep reinforcement learning, and data throughput.


A New Classification of Clustering-based for Different Problems in Different Wireless Ad-hoc Networks

Adda Boualem1, Marwane Ayaida2, Hichem Sedjelmaci3, Chaima Khalfi1, Kamilia Brahimi1, Bochra Khelil1, Sanaa Bouchama1, 1Departement Informatique Universite Ibn Khaldoun Tiaret, Algerie, 2IEMN UMR CNRS 8520 Universite Polytechnique Hauts-de-France 59300 Valenciennes, France, 3Ericsson R&D Security Massy Palaiseau, France

ABSTRACT

The ad-hoc networks have been particularly designed for establishing communications in environments where they are extremely complicated or infeasible for creating the specific network infrastructure. The introduction of clustering concept in various ad-hoc networks problems such as (Wireless Sensor Networks (WSN), Mobile Ad-hoc Networks (MANET) as Vehicular Ad-hoc Networks, and Delay-Tolerant Networks (DTN), Wireless ad-hoc Network (WANET), Underwater Wire- less Sensor Network (UWSN), Unmanned Aerial Vehicle Net- work (UAV), often called “drones”, Space Network (SN), and Satellites Networks (TN) . . . ) offers more opportunities to propose ameliorated strategies for tracking events for WSN, SN,. . . , monitoring areas in MANET, VANET,. . . ,in the de- terministic and uncertain environments. This paper presents a comparative study of different proposed strategies based on the clustering concept to address coverage in deterministic and uncertain environments. Consequently, it addresses the current and future challenges of clustering-based WANET and shows the shortcomings, strengths, and weaknesses of clustering models. Finally, some avenues for exploring coverage problems are cited in present and forthcoming new technologies.

KEYWORDS

Clustering-based, WSN, MANET, DTN, UWSN, WANET, UWSN, UAV, SN, TN, Coverage, Connectivity, Deterministic-based models, Uncertainty-based models, Coverage Current Challenges, Coverage Future Challenges.


Social Sphere: Developing a Gamified Intervention for Enhancing Real-world Communication Skills in Socially Anxious Youth

Kai Zhang1, Moddwyn Andaya2, 1Moreau Catholic High School, 27170 Mission Boulevard, Hayward, 2Computer Science Department, California State Polytechnic University, USA

ABSTRACT

Social Anxiety has been a large problem overall with many kids as well as me. However, we have made a proposed solution to the feeling of not knowing how to express yourself. Tens of millions of children go to school staying quiet, most because they are scared to talk or do not know what to say. With resources arising, this issue should have been fixed a long time ago. However, the truth is, that those who are scared to talk to others, are also scared to receive help. Most would rather sit behind screens and talk on social media, a group of platforms that make socialization completely inconsequential and useless [8]. This does not allow for the people to reach out for help. Standing behind the normalization of the screen, which allows for lying and manipulation. It is a safer world out there in person than on social media, which is why I created Social Sphere. This game will make the experience of talking feel real, without severe consequence, or anything that can scare them back off.

KEYWORDS

Digital Therapy, Communication Skills, Virtual Socialization.


An Intelligent Wearable and Mobile System for the Automated Falling Detection and Protection using Machine Learning and Internet-of-things

Zimi Zeng1, Ziyu Huang1, Jonathan Sahagun2, 1University High School, 2Computer Science Department, USA

ABSTRACT

In todays society, elderly adults living alone face various challenges. From this we can feel the indifference of modern society and the danger of falling for elderly people living alone. By recounting my grandfathers personal experience of a fall, the urgent need for assistance among elderly adults and the potential dangers of isolation are highlighted. The core question is: How can technology be used to improve the safety and overall quality of life of elderly adults living alone [1]? To solve this serious problem, we developed a fall detection device. When elderly adults wear fall detection devices, it can address these challenges by providing a mechanism to send instant alerts to family members through a connected app in the event of a fall or emergency. The device aims to improve the safety and well-being of the elderly, provide timely intervention, reduce social burdens, and solve potential social problems [2]. In addition to direct benefits, fall detectors have significant commercial value and can have a positive impact on the quality of life of older adults. The device enhances physical and mental health, increasing independence and safety in outdoor activities with fall detection and emergency alert features [3]. Fall detectors not only have commercial value, positive impact and wider social contribution, they can also play a role in addressing social neglect and inspiring elderly adults to make positive changes.

KEYWORDS

Internet-Of-Things, Machine Learning, Mobile System.


Real-time Monitoring and Visualisation of the Water Quality of Rivers in Ghana Using Wireless Sensor Networks (Wsns)

Martin Adane1 and Laurene Adjei2, 1Department of Mathematics and ICT Education, University of Cape Coast, Cape Coast, Ghana, 2Ho Technical University, Ho, Ghana

ABSTRACT

There has been a public concern and outcry over the impact of artisanal mining on rivers in Ghana which serve as a constant source of water supply for domestic, industrial, and irrigation purposes. Monitoring the quality of water in the affected river basins is mainly carried out using traditional sampling techniques which do pose a significant challenge to water regulatory bodies. The water quality parameters that need to be monitored are spatiotemporal in nature, hence, the need for the use of a more efficient approach that will perform real-time water quality monitoring is relevant. The focus of this work is to use existing opensource tools, to demonstrate how the visualisation capabilities of commercially available Wireless Sensor Networks (WSNs) may be extended allowing them to be embedded in custom-made applications. The data collected is analysed and presented in real-time thus providing timely information to aid in efficient analysis, decision-making, and planning.

KEYWORDS

Artisanal mining, spatiotemporal, real-time water quality monitoring, visualisation, wireless sensors network.


An Intelligent Website to Provide Easy Access to Coloring Pages using Edge Detection

Grace Zan1, Samin Salman2, 1University High School, 4771 Campus Drive, Irvine, CA 92612, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

Coloring books are used by children and adults to relax and relieve stress. However, there are limited subject matters, and it may be hard to find exactly what you want to color. By creating a program that converts any image into a custom coloring page, you no longer have to spend as much time searching for a suitable coloring book and have more variety in what you want to color. Our program allows the user to import an image and convert it into a coloring page using edge detection. It provides three different types of edge detection that the user can pick from and also allows the user to pick two to blend together to create a more detailed, fuller image. There is also a webcam feature that displays an edge-detected live feed to provide an augmented reality of the user’s environment in an outlined format. By using this program, you can quickly and easily create a coloring page that will suit your tastes.

KEYWORDS

Edge Detection, Coloring Pages, Stress Relief, AI.


Autismsphere: an Immersive Interactive Social Skills Training Platform for Children with Autism using Virtual Reality

Carter Qin1, Francis Rodriguez2, 1Stevenson High School, 3152 Forest Lake Road, Pebble Beach, CA 93953, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

In the summer of 2023, we volunteered at Xingle Family in Shanghai, an educational institution for autistic people. After mentoring and helping them develop their problem-solving and social skills, we were inspired to create a project that allows another avenue for those who could not easily visit Xingle Family. Using VR technology, we created three games that help develop the aforementioned skills without having to leave their homes: a pipe-connecting game, a shape-driven balloon-popping game, and a color sorting game [1]. This project was made possible through Unity3D, a game engine running on C#, and Blender, an open-source 3D modeling tool [2]. To experiment with the games, we performed playtesting with three friends that provided feedback through survey questions. We found that the games were generally easy to pick up but didn’t warrant any replays due to both the lack of difficulty or the lack of appeal; harder levels may keep the player more engaged and thus help develop their motor and problem-solving skills. We want the project to prove that VR experiences can genuinely help development for autistic people with the advantage of it being more convenient for those who cannot easily go outside [3].

KEYWORDS

Autism, Hand-eye coordination, Visual Perceptual Skills, Computer Games.


Modeling, Simulation and Control of Anaerobic Fermentation of Wastewater

Honorato Ccalli Pacco, National University of Moquegua, Peru

ABSTRACT

This research work was carried out for the purpose of modeling and simulation of temperature control in the anaerobic fermentation of waste water in a bioreactor or controlled fermentation reactor. Research carried out at the Professional School of Agroindustrial Engineering of the National University of Moquegua, Peru.It consists of using the LabVIEW software from National Instrument that allows graphic programming without the need for a programming specialist. Temperature control is applied in the on/off mode, which has control of the minimum and maximum temperature based on a Set point, where if the Set point value is modified, the minimum and maximum temperature is automatically modified. . For this, an analog temperature sensor is used, which is Lm35, which in this case is simulated and is an input variable and the output variable is the system cooler. In the block panel, a While Loop cycle is created for the installed indicators, the voltage to temperature converters are created. A SubVI is opened for the Lm35 to temperature converter. Using all this instrumentation, it was possible to carry out the modeling, simulation and temperature control in the bioreactor for the anaerobic fermentation of waste water, obtaining the results expected by this research using the LabVIEW software.

KEYWORDS

Modeling, Simulation, Bioreactor, Fermentation, Programming.


Combining Discrete Wavelet and Cosine Transforms for Efficient Sentence Embedding

Rana Salama1, 3, Abdou Youssef1, and Mona Diab2, 1School of Engineering and Applied Science, George Washington University, 2Language Technologies Institute, Carnegie Mellon University, 3Faculty of Computers and Artificial Intelligence, Cairo University

ABSTRACT

Wavelets have emerged as a cutting edge technology in a number of fields. Concrete results of their application in Image and Signal processing suggest that wavelets can be effectively applied to Natural Language Processing (NLP) tasks that capture a variety of linguistic properties. In this paper, we leverage the power of applying Discrete Wavelet Transforms (DWT) to word and sentence embeddings. We first evaluate, intrinsically and extrinsically, how wavelets can effectively be used to consolidate important information in a word vector while reducing its dimensionality. We further combine DWT with Discrete Cosine Transform (DCT) to propose a non-parameterized model that compresses a sentence with a dense amount of information in a fixed size vector based on locally varying word features. We show the efficacy of the proposed paradigm on downstream applications models yielding comparable and even superior (in some tasks) results to original embeddings.

KEYWORDS

Natural Language Processing, Sentence Embedding, Discrete Wavelets Transform, Discrete Cosine Transform.


Summarizing Arabic Articles Using Large Language Models

Bader Alshemaimri, Ibrahim Alrayes, Turki Alothman, Fahad Almalik, Mohammed Almotlaq, Department of Software Engineering, King Saud University, Saudi Arabia

ABSTRACT

This paper explores abstractive and extractive Arabic text summarization using AI, employing finetuning and unsupervised machine learning techniques. We investigate the adaptation of pre-trained language models such as AraT5 through fine-tuning. Additionally, we explore unsupervised methods leveraging unlabeled Arabic text for generating concise and coherent summaries by utilizing different vectorizers and algorithms. The proposed models are rigorously evaluated using text-centric metrics like ROUGE Lin (2004). The research contributes to the development of robust Arabic summarization systems, offering culturally sensitive and contextually aware solutions. By bridging the gap between advanced AI techniques and Arabic language processing, this work fosters scalable and effective summarization in the Arabic domain.

KEYWORDS

Arabic Text Summarization, Abstractive Summarization, Extractive Summarization, Natural Language Processing (NLP), Fine-Tuning Language Models, Unsupervised Machine Learning, Vectorization Techniques, Pre-trained Language Models, Language Model Adaptation.


Universal Dependencies Annotation of Old English Texts With Spacy and Mobilebert. Evaluation and Perspectives

Javier Martín Arista1, Ana Elvira Ojanguren López1, Sara Domínguez Barragán1 and Mario Barcala Rodríguez2, 1Department of Modern Languages, University of La Rioja, Spain, 2NLPgo, Santiago de Compostela, Spain

ABSTRACT

The aim of this paper is to assess three training corpora of Old English and three configurations and training procedures as to the performance of the task of automatic annotation of Universal Dependencies (UD, Nivre et al., 2016 [1]). The method was aimed to deciding to what extent the size of the corpus improved results and which configuration turned out the best metrics. The training methods included a pipeline with default configuration, pre-training of tok2vec step and a model of language based on transformers. For all training methods, three training corpora with different sizes are tested: 1,000, 5,000, 10,000, and 20,000 words. The training and the evaluation corpora are based on ParCorOEv2 (Martín Arista et al., 2021 [2]). The results can be summarised as follows. The larger training corpora result in improved performance in all the stages of the pipeline, especially in POS tagging and dependency parsing. Pre-training the tok2vec stage yields better results than the default pipeline. It can be concluded that the performance could improve with more training data or by fine-tuning the models. However, even with the limited training data selected for this study, satisfactory results have been obtained for the task of automatically annotating Old English with UD.

KEYWORDS

Natural Language Processing, Universal Dependencies, syntactic annotation, NPL library, Transformers.


An Analysis of Clinical and Sociodemographic Data on Congenital Syphilis Using Gaussian Naive Bayes and Xai Modeling for Interpretability Using Lime

Nishat Nayla, Dr Munima Haque, Md Sabbir Hossain and Annajiat Alim Rasel, Department of Computer Science & Engineering, Brac University, Dhaka, Bangladesh

ABSTRACT

This study delves into the significant public health challenge of congenital syphilis in Pernambuco, Brazil, by leveraging the power of machine learning and advanced data analytics. Utilising a comprehensive dataset from the Mãe Coruja Pernambucana Program, which includes 47,604 records spanning a decade (2013-2022), the research focuses on identifying key risk factors associated with congenital syphilis and developing strategies for ef ective public health interventions. At the core of the analysis is the application of the Gaussian Naive Bayes classification algorithm, renowned for its simplicity and ef ectiveness in handling large datasets. The models performance was evaluated using standard metrics, revealing an impressive accuracy of 97\%, precision of 98\%, recall of 99\%, and an F1 score of 99\%. These metrics indicate the models exceptional ability to correctly identify cases of congenital syphilis and minimise false positives and negatives, which is crucial in a public health context. Moreover, the study employs Explainable Artificial Intelligence (XAI), specifically the Local Interpretable Model-agnostic Explanations (LIME) technique, to provide clarity on the decision-making process of the model. A key finding from the feature importance analysis was the prominence of the "MOTHER\_SYPHILIS\_TEST\_RESULT" variable. This variable emerged as a significant predictor, indicating that the syphilis test results of mothers are crucial in predicting congenital syphilis in newborns. In conclusion, the study shows how machine learning and explainable AI, specifically Gaussian Naive Bayes and LIME, ef ectively identify key predictors of congenital syphilis, leading to improved, actionable public health interventions.

KEYWORDS

congenital syphilis, Gaussian Naive Bayes, LIME, explainable AI.


Forging the Future: Linguistic Engineering and the Birth of a New Paradigm in Digital Humanities

Nisrine El Hannach1 and Ali Boualam2, 1Department of English, Mohamed 1st University, Oujda, Morocco, 2Moulay Ismail University. Meknes. Morocco

ABSTRACT

This article aims to address the problematics of the humanities in the context of the digital revolution by questioning a set of hypotheses primarily revolving around the trajectories of scientific achievements within the realms of cognitive sciences. It considers these realms as pivotal points upon which theoretical and methodological relationships between exact sciences and humanities are built. This is accomplished by elucidating the bridges woven by linguistic research in bridging the gap between various fields of knowledge. In this context, the possibilities of constructing a new guiding model will be presented, a system for this digital transformation and its impacts in shaping epistemological features, establishing a horizon for digital humanities that surpasses the closed structure of humanities and social sciences. The aim is to integrate the technological component into the cultural system by highlighting the contemporary linguistic curve through its computational, visual, and cognitive articulations within the realm commonly referred to as "platform linguistics." This field is essentially an engineering approach to digitizing knowledge, employing mechanisms of artificial intelligence. Thus, the architecture of the intervention will rely on the linguistic engineering framework to highlight the inputs that establish a new conceptual model, serving as a foundation for the concept of digital humanities. This concept will not materialize without the transfer of technology from its technical level to an intellectual system foundational for digital culture.

KEYWORDS

Digital Humanities - New Conceptual Model - Cognitive Sciences - Platform Linguistics - Artificial Intelligence - Digital Culture.


An Interactive and Helpful Program to Help Foreign Language Learners Learn New Languages Through Videos

Shengyi Wang1, Christopher Muller2, 1Margaret’s Episcopal School, 31641 La Novia Ave, San Juan Capistrano, CA 92675, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

ForeignLanguagePro is a foreign language learning app targeted at beginner to intermediate level language learners to help them learn a new language while watching videos they enjoy [1]. This app is created in hopes of solving the problem of not having an enjoyable time learning a foreign language without engaging or interesting activities to look at while learning [2]. In the program of the app, AI is heavily used in order to automatically transcribe the videos, tokenize or select certain types of words in the transcript and generate multiple choice questions from the selected words. Firebase is used as an iCloud database for the program. Challenges such as tokenization not selecting the right words and ChatGPT needing a membership to authenticate the user are fixed by understanding how to use these functions correctly [3]. My application works with basically everyone who wants to watch a video while learning a new language. The most important result that I found is that you can make an effective and simple app that can be expanded extensively in terms of the diversity of videos users can choose from while attempting to learn a new language. I think my idea is very helpful to everyone who wants to learn a new language at the moment and also does not want to bore themselves out from constantly doing repetitive practice to learn that language and ultimately make language learning an enjoyable time rather than a tedious one.

KEYWORDS

Foreign Language, Database, Randomizer, Machine Learning .


Comments Analysis in Social Media

David Ramamonjisoa and Shuma Suzuki, Iwate Prefectural University, Faculty of Software and Information Science, postbox 020-0693

ABSTRACT

This paper presents an online tool that automatically filters comments on YouTube based on agents built with Large Language Models (LLMs). We designed and built agent-based comment filtering system driven by the Langchain framework and LLMs to allow users a better interface for comments reading and discussions. Specialized agents can detect spam, toxic or constructive comments. Agents are constructed as modular Langchain object with memory components and reasoning tools to make a chain of thought by interacting with LLMs. In this paper, we provide a comprehensive detail of the agent architecture and workflow to fulfil its task. Additionally, we compare the generated result by the agents to our previous tool available online. We discuss our expectations regarding the future of the comment system.

KEYWORDS

YouTube comments, Large Language Model, Agents, Filtering and Visualization .


Amosl: Adaptive Modality-wise Structure Learning in Multi-view Graph Neural Networks for Enhanced Unified Representation

Peiyu Liang, Hongchang Gao, and Xubin He, Computer and Information Science, Temple University, Philadelphia, PA, USA, 19122

ABSTRACT

While MVGNNs excel at leveraging diverse modalities for learning object representation, existing methods assume identical local topology structures across modalities that overlook real-world discrepancies. This leads MVGNNs straggles in modality fusion and representations denoising. To address these issues, we propose adaptive modality-wise structure learning (AMoSL). AMoSL captures node correspondences between modalities via optimal transport, and jointly learning with graph embedding. To enable efficient end-to-end training, we employ an efficient solution for the resulting complex bilevel optimization problem. Furthermore, AMoSL adapts to downstream tasks through unsupervised learning on inter-modality distances. The effectiveness of AMoSL is demonstrated by its ability to train more accurate graph classifiers on six benchmark datasets.

KEYWORDS

Multi-view Graph Neural Network, Graph Classification, Graph Mining, Optimal.


A Complexity-aware Web Searching Paradigm to Improve User Productivity Using Natural Language Processing and the Distilbert Transformer Model

Ruiyi Zhang1, Carlos Gonzalez2, 1Troy High School, 2200 Dorothy Ln, Fullerton, CA 92831, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

Search engines (Google search, Bing search, etc.) have had great success over the past decade, promoting productivity in almost every area. Based on user inputs, search engines are able to present users with lists of related contents (links) and previews. More recently, high-level human-like responses combining various searched contents are being made possible due to recent advancements in large language models (LLM). However, oftentimes, users still find it still hard to quickly navigate to the contents they really look for and demand a better searching framework. For example, some users might waste time skimming through lots of technical details when they just hope to have an overview. We examine this user demand and believe a complexity-aware pipeline could greatly help with this inconvenience. More specifically, we propose a searching paradigm that analyzes results from standard search engines by their complexities first, and then present users with complexity-labeled contents through a new user interface design. Through this new searching paradigm, we aim to present users with more customized search results sorted by their complexity labels with consideration to user intent, whether that would be a high-level overview or a detailed technical inspection. This is done through utilizing state-of-the-art transformer models fine-tuned on our custom-made dataset and modified for our intent.

KEYWORDS

Transformer, Natural Language Processing, Complexity-Aware, Web Search.


Fine-tuning of Small/medium Llms for Business Qa on Structured Data

Rasha Ragab Ahmed, University of Leeds, United Kingdom

ABSTRACT

Enabling business users to directly query their data sources is a significant advantage for organisations. Most of the organizational data is stored in databases, necessitating lengthy processes involving intermediate layers for custom report creation. The concept of enabling natural language queries, where a chatbot can interpret user questions into database queries and promptly return results, holds promise for expediting decision-making and enhancing business responsiveness. This approach empowers experienced users to swiftly obtain data-driven insights. The integration of Text-to-SQL and Large Language Model (LLM) capabilities represents a solution to this challenge, offering businesses a powerful tool for query automation. However, security concerns prevent organizations from granting direct database access akin to platforms like OpenAI. To address this limitation, this Paper proposes developing fine-tuned small/medium LLMs tailored to specific domains like retail and supply chain. These models would be trained on domain-specific questions and database table structures to ensure efficacy and security. A pilot study is undertaken to bridge this gap by fine-tuning selected LLMs to handle business-related queries and associated database structures, focusing on sales and supply chain domains. The research endeavours to experiment with zero-shot and fine-tuning techniques to identify the optimal model. Notably, a new dataset is curated for fine-tuning, comprising business-specific questions pertinent to the sales and supply chain sectors. This experimental framework aims to evaluate the readiness of LLMs to meet the demands for business query automation within these specific domains. The study contributes to advancing the field of natural language query processing and database interaction for business intelligence applications.

KEYWORDS

Text-2-SQL, Fine-Tuning, Small/Medium LLM.


A Review of Optimization Algorithms in Deep Learning Models for Improving the Forecasting Accuracy in Sequential Datasets With Application in the South Africanstock Market Index

Sanele Makamo, Benguela Global Fund Managers, South Africa

ABSTRACT

In this paper we review different popular optimization algorithms for machine learning models, we then evaluate the model performance and convergence rates for each optimizer using a multilayer fully connected neural networks. Using sequential dataset of index returns (time-series data) spanning over of 20-years, we demonstrate Adam and RMSprop optimizers can efficiently solve practical deep learning problems dealing with sequential datasets. We use the same parameter initialization when comparing different optimization algorithms. The hyper-parameters, such as learning rate and momentum, are searched over a dense grid and the results are reported using the best hyper-parameter setting.

KEYWORDS

machine learning, deep learning, neural networks, optimization algorithms, loss function.


Enumeration of Pirogues Using Google Earth Images

Olalekan Olaluwoye1, Timothee Brochier2 and Alassane Bah3, 1African Institute for Mathematical Sciences, Mbour, Senegal, 2Institute of Research for Development, Marseille, France, 3Université Cheikh Anta Diop de Dakar, Senegal

ABSTRACT

Satellite image analysis has gained paramount importance across diverse applications, ranging from land cover mapping and disaster management. Within this spectrum, the detection and enumeration of specific objects, such as pirogues (fishing boats), hold significant value. However, object detection is faced with challenges, including varying resolutions, occlusion, and complex backgrounds. This research addresses a critical challenge in obtaining reliable data on the numbers of pirogues at fishing landing sites in Senegal. The aim is to develop an artificial intelligence workflow using Google Earth images to enumerate pirogues at the fishing landing sites. The workflow encompasses data collection, annotation, model development, and robust analysis. The performance of two well-known object detection algorithms, YOLOv5 and YOLOv8, is evaluated. The results reveal the superiority of YOLOv8 in terms of mean average precision. The study further integrates YOLOv8 with multiple object tracking to automate the counting of pirogues at the Kayar fishing landing site.

KEYWORDS

Artisanal Fishing Landing Sites, Google Earth Images, Pirogues, Artificial Intelligence, YOLOv5 &YOLOv8 .


Data Trustworthiness: Quality Scoring

Faouzi Boufarès1, Aicha Ben Salem1, 2, and Adam Boufarès3, 1Northern Paris Computer Science Lab, LIPN, France LaMSN, Sorbonne Paris Nord University, France, 2Laboratory RIADI-La Manouba,Tunisia, 3Transactis Company, France

ABSTRACT

Nowadays, data is very important in organizations and companies. The data quality has a strong influence on decisions and the consequences can be very significant. It is very important to have an idea of how much trust you should place in the data before any process starts. For this reason, it is essential to carry out a checklist of data validity before using them and initiate, as much as possible, a process of data correction and enrichment. In this paper we present a method to visualize errors in a data source. A score is assigned to each error. The database checklist covers existing and missing values homogenizations, functional dependencies, duplicate and similar rows. Several measures are performed to assist the correction of anomalies.

KEYWORDS

Data Quality, Data trustworthiness, Data Anomalies, Constraints, Management Rules Scoring Data, Data Cleaning, Machine Learning, Datasets, Structured or not Structured Data, CSV or JSON files.


Load-aware Smart Consumer for Adaptive Resource Management

Sai Guruju, Abhishek Niranjan, Krishnachaitanya Gogineni, Jithendra Vepa, Observe.AI

ABSTRACT

In modern systems, especially those dealing with multimedia content like audios and videos, managing varying workloads ef iciently is crucial for maintaining performance and cost-ef ectiveness. This paper presents a Load-Aware Smart Consumer system designed to handle the transcription of customer-agent conversations, a process that is often hampered by the wide variance in the duration of audio calls. Our system dynamically adjusts the concurrency of processing based on the current load, thereby ensuring stability and ef icient resource utilization. By monitoring the processing load across instance-workers, the system can make informed decisions about accepting and processing new tasks, leading to improved resilience and cost savings. This approach is not limited to transcription engines but can be applied to any multimedia processing system facing similar challenges of input variability and resource constraints.

KEYWORDS

high latency systems, concurrency control, cost-ef iciency, stability, dynamic resource allocation, multimedia processing .


Arthub: an Interactive Visual Art Gallery to Display Student Works and Support Art Education Using Virtual Reality

Ada Zhou1, Joshua Lai2, 1Portola High School, 1001 Cadence, Irvine, CA 92618, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

The expression of art is incredibly important for young students. Equally important is allowing these students to showcase what they’ve done to peers and others. It helps them gain confidence in their own abilities and gives them motivation to practice their skills. The best way to do this is typically an in-person art gallery, but this is not always convenient. Those who wish to view the art may not always have the opportunity to travel to a physical location, and there are numerous costs associated with setup and cleanup. To address this issue, we propose ArtHub, a hybrid desktop and virtual reality application that allows anyone to upload their artwork to a virtual gallery that may then be explored by others. It makes use of the Unity game engine and is compatible with the Meta Quest 2 headset. In order to allow different people to view each other’s work, it features a remote database that users may save their artwork to. Galleries are associated with a passcode to prevent different galleries from interfering. Unlike physical galleries ArtHub may be accessed from anywhere, requires little logistical setup, and is infinitely expandable. Our application will help support young artists in schools who otherwise may not have had the chance to show their work to a wider audience.

KEYWORDS

Art, Virtual Reality, Education, Gallery.


Machine Learning: Enhancing Intelligent Search and Information Discovery

Nikhil Ghadge, Software Architect, Identity Cloud, Okta.Inc

ABSTRACT

Machine learning algorithms are revolutionizing intelligent search and information discovery capabilities. By incorporating techniques like supervised learning, unsupervised learning, reinforcement learning, and deep learning, systems can automatically extract insights and patterns from vast data repositories. Natural language processing enables deeper comprehension of text, while image recognition unlocks knowledge from visual data. Machine learning powers personalized recommendation engines and accurate sentiment analysis. Integrating knowledge graphs enriches machine learning models with background knowledge for enhanced accuracy and explainability. Applications span voice search, anomaly detection, predictive analytics, text mining, and data clustering. However, interpretable AI models are crucial for enabling transparency and trustworthiness. Key challenges include limited training data, complex domain knowledge requirements, and ethical considerations around bias and privacy. Continued research bridging machine learning, knowledge representation, and human-centered design principles will drive future advancements in intelligent search and discovery. The synergy of artificial and human intelligence promises transformative information access and knowledge acquisition capabilities.

KEYWORDS

Machine Learning, Artificial Intelligence, Search Engine, Data Retrieval, Natural Language Processing, Data mining.


Development of an Attendance Monitoring System Utilizing Face Recognition Libraries in Python

Yves Spencer Catuday Mark1, Jerald De Torres2 and Godwin Emmanuel Tayas3, 1Department of Electronics Engineering, Batangas State University, Batangas City, Philippines, 2Department of Electronics Engineering, Batangas State University, Batangas City, Philippines, 3Department of Electronics Engineering, Batangas State University, Batangas City, Philippines

ABSTRACT

This study presents the development of an attendance monitoring system that utilizes face recognition technology. The system is built using Python libraries and aims to provide an efficient method for tracking student attendance in educational institutions. The study discusses the rapid advancements in face recognition technology and its growing application in various fields, including security, authentication, and identification. Traditional attendance methods are often tedious and time-consuming, leading to the exploration of automated systems like the one proposed in this study. The system works by initializing a web camera and detecting student s faces in real time. Once a face is recognized, the system marks the student s attendance. The system has been tested with a dataset of 25 student images, achieving a recognition rate of 92% and an overall accuracy of 84%. Despite some challenges, such as the complexity of installing Python libraries and factors affecting recognition accuracy, the system demonstrates the potential for real-world application. The study concludes that face recognition libraries in Python can successfully locate and identify faces from a database, making them suitable for attendance monitoring scenarios. For future research, the study suggests adding features to adjust video quality based on surrounding conditions and incorporating a stabilizer to improve the accuracy and stability of the recognition phase. The researchers believe these enhancements could improve system performance and broader applicability. This study contributes to the growing body of research on the practical applications of face recognition technology and offers a novel approach to attendance monitoring in educational settings.

KEYWORDS

Face Recognition, Python, Attendance Monitoring System, Face Recognition Library.


Airwatch: a Real-time and Fine-granularity Air Quality Monitoring and Analytical System Using Machine Learning and Drone Technology

Shuaishuai Guo1, Jonathan Sahagun2, 1Crean Lutheran High School, 12500 Sand Canyon Ave, Irvine, CA 92618, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

This paper addresses the critical environmental challenge of air quality degradation, exacerbated by industrial emissions, vehicular pollutants, and agricultural activities [1]. Our proposed solution, a Real-Time and FineGranularity Air Quality Monitoring and Analytical System, leverages machine learning and drone technology to dynamically monitor and analyze air quality across diverse locations and altitudes. By integrating drone-mounted sensors, advanced machine learning algorithms, and a user-friendly interface, the system offers unprecedented spatial and temporal resolution in air quality assessment. The study navigated through limitations such as data transmission reliability and the complexity of real-time data analysis, employing robust communication protocols and enhanced analytical models for improved accuracy [2]. Experimentation across various urban and rural settings demonstrated the systems effectiveness in identifying pollution hotspots and predicting air quality trends, with significant improvements over traditional stationary monitoring methods. Our findings highlight the potential of combining drone mobility with machine learning efficiency to revolutionize air quality monitoring, making it an indispensable tool for environmental management and public health protection [3].

KEYWORDS

Air Quality Monitoring, Drone Technology, Machine Learning, Real-time Data Analysis.


An Ai-powered System to Increase User’s Productivity Using Computer Vision and Machine Learning

Lida Song, The York School, 1320 Yonge Street, Toronto, Ontario, M4T1X2, Canada

ABSTRACT

DistraXcel represents an innovative AI-powered system designed to enhance productivity by using computer vision and machine learning to mitigate distractions from digital applications and websites [1]. The internet, likened to a "giant hypodermic" filled with distracting "psychoactive drugs," significantly impedes focus [2]. DistraXcel addresses this problem by identifying and closing distracting digital content, leveraging Python, Tkinter for the frontend, Firebase for backend operations, and Roboflow for building an image recognition model. However, challenges such as auto-login hassles, AIs specificity to certain screens leading to false negatives, and cross-platform GUI inconsistencies were encountered [3]. Through methodical experiments, the AI demonstrated strong identification capabilities but also revealed an overfitting issue. A user perception survey highlighted a moderate improvement in perceived productivity post-use, indicating DistraXcels effectiveness yet suggesting room for addressing broader distraction factors. DistraXcel advances beyond current methodologies by offering a more positive feedback mechanism, extensive customization, and prioritizing privacy [4]. It evolves as a user-centric tool aimed at fostering a focused and productive digital environment, marking a significant step forward in addressing the pervasive issue of digital distractions.

KEYWORDS

Focus Assistance, Computer Vision, Machine Learning, Python.


Designing Fault-tolerant Modern Data Engineering Solutions With Reliability Theory as the Driving Force

Ankit Virmani1 and Manoj Kuppam2, 1Department of Artificial Intelligence, Virufy, California, USA, 2IEEE Senior Member, Dallas, Texas, USA

ABSTRACT

This study has been undertaken to amalgamate the principles of Site Reliability Engineering and Data Engineering to effectively measure, monitor and manage the reliability of petabyte-scale data engineering process from collection at source to processing, analysing, and distributing the data for appropriate decision making to improve business outcomes and system performance. Modern data architectures increasingly leverage cloud platforms, low-code systems, and serverless technologies to enable scalable data engineering. However, these innovations also introduce new complexities regarding reliability assurance. As these failure-prone yet business-critical data infrastructures continue rapid adoption, it is vital to elucidate architectural paradigms, quantified benchmarks, and procedural methodologies tailored to safeguarding dependability across heterogeneous, distributed data ecosystems. This paper will equip end users with a reusable framework ingrained with best practices to develop a blueprint for data reliability across business units of an organisation.

KEYWORDS

Big Data Engineering, Site Reliability Engineering, Data Operations, Platform Management, Cloud Computing.


Unlocking Access to Healthcare: Evaluating the Efficacy of AI-driven Diagnosis in a Mobile Application

Mylyn Zheng1, Adam Grant2, 1Portola High School, 1001 Cadence, Irvine, CA 92618, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768

ABSTRACT

In todays fast-paced world, accessing quality healthcare can often feel like an uphill battle. Whether hindered by geographical barriers, long waiting times, or financial constraints, many individuals struggle to receive timely diagnoses for their health concerns. Recognizing this pressing need, we ve developed an innovative solution in the form of our cutting-edge app [7]. Harnessing the power of artificial intelligence, our app offers users a seamless pathway to diagnosis based on their reported symptoms. Our journey began with meticulous research to identify the most reliable AI technology, leading us to integrate ChatGPT, a state-of-the-art Large Language Model renowned for its ability to mimic professional medical responses accurately [8]. To ensure seamless functionality, we complemented this AI prowess with a robust Firebase database, facilitating efficient data storage and retrieval [9]. Following rigorous development and testing phases, we re thrilled to announce that our app delivers unparalleled accuracy in diagnosis. Our trials revealed an impressive average accuracy rating of 1.8 out of 2 for common illnesses, underscoring the reliability and efficacy of our diagnostic algorithms. While initial evaluations highlighted room for improvement in symptom diversity across a spectrum of 21 illnesses, our focus on refining the apps capabilities for prevalent conditions has yielded remarkable results. Ultimately, our app stands as a beacon of accessibility in the realm of basic healthcare. By bridging the gap between individuals and vital medical expertise, were empowering users to take proactive control of their well-being, irrespective of their circumstances. With our unwavering commitment to innovation and excellence, were proud to revolutionize healthcare delivery and make a meaningful difference in peoples lives.

KEYWORDS

Artificial Intelligence, Flutter, App Development, Health Tracking.


menu
Reach Us

emailicaita@icaita2024.org


emailicaitaconf@yahoo.com

close