CINXE.COM

<!DOCTYPE html> <html> <head>  <link href="https://fonts.googleapis.com/icon?family=Material+Icons" rel="stylesheet"> <link href="https://fonts.googleapis.com/css?family=Roboto+Condensed" rel="stylesheet">  <link type="text/css" rel="stylesheet" href="css/materialize.min.css" media="screen,projection" /> <link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.0.13/css/all.css" integrity="sha384-DNOHZ68U8hZfKXOrtjWvjxusGo9WQnrNx2sqG0tfsghAvtVlRW3tvkXWZh58N9jp" crossorigin="anonymous"> <link type="text/css" rel="stylesheet" href="css/main.css" /> <meta charset="UTF-8">  <meta name="viewport" content="width=device-width, initial-scale=1.0" /> <title>15th International Conference on Computer Science, Engineering and Information Technology (CCSEIT 2025)</title> <link rel="icon" type="image/ico" href="img/logo.ico"> </head> <body>  <header class="main-header"> <nav class="transparent"> <div class="container"> <div class="nav-wrapper"> <a href="#" class="brand-logo">CCSEIT</a> <a href="#" data-activates="mobile-nav" class="button-collapse"> <i class="fa fa-bars"></i> </a> <ul class="right hide-on-med-and-down"> <li> <a href="index">HOME</a> </li> <li> <a href="papersubmission">PAPER SUBMISSION</a> </li> <li> <a href="committee">PROGRAM COMMITTEE</a> </li> <li> <a class="active-link" href="#">ACCEPTED PAPERS</a> </li> <li> <a href="contact">CONTACT US</a> </li> <li> <a href="venue">VENUE</a> </li> </ul> <ul class="side-nav grey darken-1 white-text" id="mobile-nav"> <h4 class="center">CCSEIT 2025</h4> <li> <div class="divider"></div> </li> <li> <a href="index"> <i class="fa fa-home white-text"></i>Home </a> </li> <li> <a href="papersubmission"> <i class="fa fa-user white-text"></i>Paper Submission </a> </li> <li> <a href="committee"> <i class="fa fa-user white-text"></i>Program Committee </a> </li> <li> <a class="active-link" href="papers"> <i class="fa fa-newspaper white-text"></i>Accepted Papers </a> </li> <li> <a href="contact"> <i class="fa fa-phone white-text"></i>Contact Us </a> </li> <li> <a href="venue"> <i class="fa fa-phone white-text"></i>Venue </a> </li> <li> <div class="divider"></div> </li> <li> <a href="/submission/index.php" target="blank" class="btn grey waves-effect waves-light">Paper Submission</a> </li> </ul> </div> </div> </nav>  <div class="showcase container"> <div class="row"> <div class="col s12 m10 offset-m1 center grey-text text-darken-3"> <h5>Welcome to CCSEIT 2025</h5> <h2>15<sup>th</sup> International Conference on Computer Science, Engineering and Information Technology (CCSEIT 2025)</h2> <p>January 25 ~ 26, 2025, Copenhagen, Denmark</p> <br> <br> </div> </div> </div> </header> <section class="section section-icons "> <div class="container"> <div class="row"> <div class="col s12 m12"> <div class="card-panel grey darken-2 z-depth-3 white-text center"> <i class="fa fa-paper-plane fa-3x"></i> <h5>Accepted Papers</h5> </div> </div> <div class="col s12 m12"> <div class="card-panel white z-depth-3 ">  <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Grounding Large Language Models in Knowledge and Reason </b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Ram Eshwar Kaundinya, Drexel University, Philadelphia PA 19104, USA </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Large language models (LLMs) have led to a leap in generative AI capability with human-like language production across vast domains. While this has been a stunning success in some respects, it has highlighted many of the limitations of a purely connectionist approach to AI. LLMs are not sufficiently grounded in a knowledge base exacerbating problems with reasoning and planning. Fine-tuning and RAG (Retrieval Augmented Generation), are not robust and are limited. This paper takes a symbolic approach to online learning within a trading game environment. I introduce two novel ideas - a memory module based on ideas from cognitive architectures such as ACT-R and a symbolic knowledge graph. This allows for online learning within a dynamic game environment. The novel architecture is generalizable and allows for grounding LLMs in a desired domain without extensive fine-tuning or RAG, enabling the creation of personalized LLM systems. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Large Language Models, Cognitive Architecture, Neurosymbolic Computing. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>From Rigid Robo-advisors to Human-like Interactions: Revolutionizing Financial Assistance with Llm-powered Solutions</b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Hamza Landolsi<sup>1</sup>, Ines Abdeljaoued-Tej<sup>1, 2</sup>, <sup>1</sup>Engineering School of Statistics and Information Analysis, University of Carthage, Ariana, Tunisia,<sup>2</sup>Laboratory of BioInformatics bioMathematics, and bioStatistics (LR24IPT09), Institut Pasteur de Tunis, University of Tunis El Manar, 13, place Pasteur, B.P. 74, Belv´ed`ere, 1002, Tunis, Tunisia </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Generative Artificial Intelligence (GenAI) is revolutionizing the business world by increasing availability, efficiency, cost reduction, and innovation. This paper explores the application of Large Language Models (LLMs) and GenAI to finance. It proposes a novel framework on how we can imagine robo-advisory systems, from a traditional rigid platform to a more humanized solution that further engages the investor in a hand-picking asset selection process and better understands their goals and profile using LLMs. We designed an end-to-end solution to overcome many limitations such as lack of flexibility in robo-advisors, lack of possible asset types (usually only equities) and the problem of real-time access to high quality data. The solution architecture includes dynamic client profiling, risk aversion estimation and portfolio optimization. It tailored asset selector agent using robust data pipelines to curate the latest market information. Through iterative development, we employed prompt engineering and multi-agent workflows to enhance user interactions and deliver meaningful insights. By developing an innovative chatbot platform, we demonstrate the potential of LLMs to transform customer service, increase engagement, and provide strategic financial advice. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Generative AI, Large Language Models (LLM), Big Data, Practical Applications, Agentic Design Patterns, Finance, Investment analysis, Portfolio Optimization </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Survey: Understand the Challenges of Machine Learning Experts using Named Entity Recognition Tools </b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Florian Freund , Philippe Tamla , and Matthias Hemmje University of Hagen, Faculty of Mathematics and Computer Science 58097 Hagen, Germany </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Comparison and selection of Named Entity Recognition (NER) tools and frameworks is a critical step in leveraging NER for Information Retrieval to support the development of Clinical Practice Guidelines. This paper presents a survey based on Kasunic’s survey research methodology to identify the criteria used by Machine Learning (ML) experts to evaluate NER tools and their significance in the selection process. In addition, it examines the main challenges faced by ML experts when choosing suitable NER tools and frameworks. Using Nunamaker’s methodology, the article begins with an introduction to the topic, contextualizes the research, reviews the state of the art in science and technology, and identifies challenges for an expert survey on NER tools and frameworks. This is followed by a description of the survey’s design and implementation. The paper concludes with an evaluation of the survey results and the insights gained, ending with a summary and conclusions. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Expert Survey, Natural Language Processing, Named Entity Recognition, Machine Learning, Cloud Computing. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Figurative Style Classification in Arabic Texts Using MT5-based Pre-trained Language Models </b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Zouheir BANOU, Sanaa EL FILALI, El Habib BENLAHMAR, Laila ELJIANI, and Fatima-Zahra ALAOUI, Faculty of Sciences Ben M’Sik – Hassan II University, Bd Commandant Driss Al Harti, 7955, Casablanca, Morocco </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Figurative language detection is a challenging task in natural language processing (NLP), especially for morphologically rich languages like Arabic. This study investigates the effectiveness of pre-trained language models (PLMs) for detecting hyperbole and metaphor in Arabic, comparing general-purpose models of varying sizes (mT5-Small, mT5-Base, and mT5-Large) with a specialized, fine-tuned model (MMFLD) trained specifically for figurative language tasks. Results indicate that while larger models such as mT5-Large excel in capturing complex figurative expressions, the task-specific MMFLD model achieves competitive performance, especially in metaphor detection. This highlights the benefits of both model size and specialized training in figurative language tasks. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Figurative Language Detection, Arabic NLP, Pre-trained Language Models (PLMs). </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Acceleration Through Fusion of Avgpool2d and Silu Kernels </b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Andreas Falkenberg, Falkenberg Technology Consulting Inc Escondido, California 92026 </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The need to accelerate LLM (large language models) requires the use of always advancing compiler technologies. Operator fusion is one of the promising techniques to considerably improve the throughput of LLMs. This paper discusses the impact of operator fusion on the direct operator performance. The paper compares throughputs between pure CPU implementation, versus two kernel implementations versus a fused single kernel solution for AvgPool2D fused with Silu. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">AvgPool2D, Silu, Kernel, AI, LLM, GPU, CPU. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b> Tqwere: Transformer-based Sql Query Executor </b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Nir Regev, Asaf Shabtai, Ben Gurion University Of the Negev, Beer Sheva, Israel </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Recent developments in large language models (LLMs) trained on large-scale unstructured textual data have produced high-achieving models. However, it remains a challenge to pre-trained an LLM on structured tabular data where the model needs to learn how tabular columns are related and describe a row entity. This is mainly due to the fact that generative pre-trained transformers (GPTs) are trained on textualdata and designed to predict the next textual token given a context of textual tokens. We propose a novel method - TQwerE, for approximating SQL aggregated queries’ results over large data sets for the focused goal of data exploration done by data scientists usually prior to building a ML model. Our main concern was to reduce query latency and incurred costs that are often pose a limitation when data scientists and analysts explore large data set. Moreover, since we focus on large data sets, majority of models that scan raw data are not applicable. Instead, our method fine tunes Jurassic-2 to learn the relations between aggregated SQL queries and their results without referring directly to the underlying raw data. We demonstrate TQwerE’s ability to approximate aggregated queries with state-of-the-art accuracy and speed. TQwerE addresses this task by first constructing a diverse SQL queries training set (referred as fine tuning set) and then performing a fine-tuning task. We evaluated TQwerE on twelve datasets, and our results demonstrated its superiority to both the state-of-the-art approximate query processing (AQP) method and Tapas - an LLM developed by Google for question answering tasks over tabular data. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>PolyIPA - Multilingual Phoneme-to-Grapheme Conversion Model </b></h6> <p style="color:black;text-align:justify;font-size: 15px;"><p style="color:black;text-align:justify;font-size: 15px;">Davor Lauc, Faculty of Humanities and Social Sciences, Croatia (Hrvatska) </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">This paper presents PolyIPA, a novel multilingual phoneme-to-grapheme conversion model designed for multilingual name transliteration, onomastic research, and information retrieval. The model leverages two helper models developed for data augmentation: IPA2vec for finding soundalikes across languages, and similarIPA for handling phonetic notation variations. Evaluated on a test set that spans multiple languages and writing systems, the model achieves a mean Character Error Rate of 0.055 and a character-level BLEU score of 0.914, with particularly strong performance on languages with shallow orthographies. The implementation of beam search further improves practical utility, with top-3 candidates reducing the effective error rate by 52.7% (to CER: 0.026), demonstrating the model’s effectiveness for cross-linguistic applications </p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>The Impact of Iot on the Modern World a Review and Evaluation Study </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Heidrich Vicci, College of Business, Florida International University, USA </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The IoT provides users with dynamic and rich context-aware services that are highly responsive to the user needs. The users can remotely monitor and control the environment. The IoT will allow more direct integration of the physical world into computer-based systems, resulting in improved accuracy, efficiency, and economic benefit in addition to reduced human intervention. The basic premise is to have objects or things working for humans rather than humans working for them. As we can see, the usage of the term follows a similar trend in both economics and search engine results, which could be a good indicator of the existence of a correlation between the two. More than 16 years after its inception, technology remains a hot topic both in the business world and in academia. (Jagarlamudi et al.2022)(Pradeep et al.2021)(Pradeep et al.2021) </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Internet of Things (IoT), computer-based systems, human intervention. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Leveraging Pervasiveness of Internet of Things in Ameliorating Military Operations </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Avnish Singh and Rachit Ahluwalia, Milit, Pune, India </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Internet of Things (IoT) has revolutionized the manner businesses interact and run. Unlike Internet which found its genesis in a military environment, credits for IOT goes to the civil industry, academia and researchers. IoT is a disruptive technology, and will change the world in many ways in the near future. Its genesis has paved opportunities which have reaped tremendous benefits for various sectors managing multitude of assets and engaged in coordination of complex and intricated processes. Armed forces have been a step behind in adopting this revolutionary technology, and there by reaping its tremendous benefits. Hence, with this paper we intend to bring out the opportunities that are waiting to be grabbed in the field of IOT, by defence forces. As the penetration of IOT devices grow, modern military will also gracefully embrace this technology or fear the risk of getting left behind in the race. This paper analyses the plethora of ways in which IoT can be used in a modern military and provide benefits similar to those prevailing in the industry, and much more. Wide variety of literature was studied and analyzed, and it was observed that no comprehensive paper exists which mentions about the possible diverse applications of IOT in a modern armed force. The paper intends to present a collation of existing applications as well as many novel concepts using IOT and related technologies. Also, we talk about how IOT and related technologies can provide the cutting edge to any military force enabling it to trump over its competitors and adversaries. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">IoT, IoT Architecture, OEM, ANE, RFID, Biometric, cloud computing, edge computing, drones, sensors, actuators, LoRaWAN, LTE-M </p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>An Intelligent Tracking System to Analyze Shooting Angles Compared to Nba Players using AI and Machine Learning </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Zhixiang Zhang<sup>1</sup>, Ang Li<sup>2</sup>, <sup>1</sup>Ruben s Ayala High School, 14255 Peyton Dr, Chino Hills, CA 91709, <sup>2</sup>California State University, Long Beach, 1250 Bellflower Blvd, Long Beach, CA 90840 </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">This project aims to solve the problem of providing real-time, personalized feedback on basketball shooting form using machine learning (ML). By comparing a user’s body angles during their shot to those of professional players, the program delivers tailored suggestions for improvement. The core technologies used include pose detection through computer vision and a machine learning model that analyzes and compares joint angles. Challenges included fine-tuning the model’s confidence score to ensure accurate comparisons between users and pros, handling image quality issues, and providing clear feedback to users of different skill levels. The experiments showed that when professional players were compared to themselves, the system returned very high similarity scores, confirming the model’s accuracy. The project stands out because of its personalized feedback feature, helping both beginner and advanced users improve their shooting form. By addressing common limitations such as image quality and skill variability, this tool offers a unique solution for athletes looking to refine their performance. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Basketball Analyze, Machine Learning Comparison, Computer Vision. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Optimizing Deep Learning Models for Osteoporosis Detection: a Case Study on Knee X-ray Images Using Transfer Learning </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Zahraa Shams Alden<sup>1, 2</sup> and Oguz Ata<sup>3</sup>, <sup>1</sup>University of Altinbas, Electrical and Computer Engineering, Turkey, <sup>2</sup>University of Kerbala, Tourism Science, Iraq, <sup>3</sup>University of Altinbas, Information Technology, Turkey </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The analysis of medical images is a very risen area of study and the speed and precision necessary in medical image analysis. Deep learning may aid in resolving medical image processing issues including labelled datasets by experts to learn effectively. This can be difficult to achieve in the medical field, where access to large amounts of labeled data may be limited. Another challenge is the complexity of medical data. Therefore, this study proposed a deep neural network-based model for medical imaging to detect osteoporosis using transfer learning with MobileNetV2. Class weights are used to alleviate class imbalance, and the learning rate schedule improves model adaptability. The model was created in two variants: one with a learning rate schedule and class weights with an accuracy of 96%, and the second model with only a learning rate schedule with an accuracy of 94%. The anticipated experimental results should illustrate the efficiency of the proposed framework for the future designing of deep learning models for predicting bone fracture and speeding up medical data analysis and interpretation. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Medical image analysis, Machine learning, CNN, Transfer Learning, Osteoporosis, Deep learning, MobileNetV2. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Performance Evaluation of Mobility in Non-terrestrial Networks </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Li-Sheng Chen<sup>1</sup> and Shu-Han Liao<sup>2</sup>, <sup>1</sup>Department of Computer Science and Information Engineering, National Ilan University, Ilan, 260007, R.O.C., <sup>2</sup>Department of Electrical Engineering, Tamkang University, New Taipei, 251301 , R.O.C. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Low Earth Orbit (LEO) satellites exhibit high mobility, leading to frequent handover challenges. Addressing these handover issues is crucial for maintaining seamless and stable service connections. In this paper, we tackle the handover problems in LEO by utilizing the D1 events, as discussed in the 3rd Generation Partnership Project (3GPP). Unlike terrestrial networks, the difference between the reference signal received power (RSRP) at the cell edge and the cell center is minimal in non-terrestrial networks (NTN). Therefore, 3GPP has been exploring location-based handover methods using absolute thresholds instead of comparing the RSRP of serving and neighboring cells in handover events. We introduce the D1 event as a handover trigger and explore handover parameters in conjunction with the UE’s position (referred to as enhanced D1) to ensure reliable handover for NTN. Simulation results show that enhanced D1 handover outperforms traditional D1 handover, particularly in reducing ping-pong effects and handover failures (HOF). </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Low earth orbit (LEO), Non-terrestrial networks (NTN), Mobility, Satellite communication, Handover. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Enhancing Distributed Computing with Artificial Intelligence: a Framework for Scalable, Resilient, and Autonomous Systems </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Vincent Froom, VST, Canada </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">This paper explores the integration of Articial Intelligence (AI) into distributed computing systems to enhance scalability, resilience, and efciency. It highlights AI’s role in dynamic resource management, fault tolerance, and real-time analytics through techniques like machine learning and reinforcement learning. Key contributions include an AI-driven framework for self-healing architectures, predictive performance monitoring, and AIquantum hybrid models. Applications in industrial automation, healthcare, and smart cities demonstrate AI’s transformative impact, while future directions address ethics, sustainability, and emerging technologies. This study lays the foundation for adaptive and intelligent distributed systems capable of meeting modern computing challenges. </p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Comparison of Training for Hand Gesture Recognition for Synthetic and Real Datasets </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Pranav Vaidik Dhulipala, Samuel Oncken, Steven Claypool, and Stavros Kalafatis, Department of Electrical and Computer Engineering, Texas A&M University, College Station, Texas-77845, USA </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Human gesture recognition is often implemented in many HRI applications. Building datasets that involve human subjects, when aiming to capture comprehensive diversity and all possible edge cases is often both challenging and labor-intensive. While applying the concept of domain randomization to build synthetic datasets helps address the problem, an innate reality gap always exists that needs to be mitigated. In this paper, We present and discuss a comprehensive performance comparison of our synth datasets with real ones and demonstrate the results in this paper. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Acceleration of Cuda Kernels Through Fusion Measurements for Avgpool2d and Relu </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Andreas Falkenberg, Dr Falkenberg Technology Consulting Inc., Escondido, California 92026 </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The need to accelerate LLM (large language models) requires the use of always advancing compiler technologies. Operator fusion is one of the promising techniques to considerably improve the throughput of LLMs. This paper discusses the impact of operator fusion on the direct operator performance. The paper compares throughputs between pure CPU implementation, versus two kernel implementations versus a fused single kernel solution for AvgPool2D fused with ReLU. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">AvgPool2D, ReLU, Kernel, AI, LLM, GPU, CPU. </p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Cracking the WTP Metric for Enhancing Pricing Framework and Margin Optimization using Reasoning Llms, Agents and Langchain Framework </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Chirag Soni<sup>1</sup> and Swati Shah<sup>2</sup>, <sup>1</sup>Senior Product Manager, Data and Insights, PayPal, Bangalore India, <sup>2</sup>Swati Shah, Head – Margin Optimization Product, PayPal, SJ USA </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The advancements in llms and agentic universe has enabled businesses to incorporate intelligent automation across business applications. With this paper we present a methodology to enhance b2b sales process where langchain framework and agents could be used to fetch critical information from the internet in real time to indirectly calculate a business’s willingness to pay in the form of a score generated by reasoning llms, and rank order these businesses to filter out the ones who at the given point in time have the highest wtp, which could be used by sales teams to prioritize, negotiate a better pricing and identify sales opportunities currently active in their respective time zones. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Generative AI, LLM, LangChain, AI Agents, Reasoning models, Risk, Sales.</p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Separated Inter/Intra-modal Fusion Prompts for Compositional Zero-Shot Learning </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Sua Jung </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Compositional Zero-Shot Learning (CZSL) aims to recognize subtle differences in meaning or the combination of states and objects through the use of known and unknown concepts during training. Existing methods either focused on prompt configuration or on using prompts to tune the pre-trained VisionLanguage model. However, these methods faced challenges in accurately identifying subtle differences in meaning or combining states with objects. To jointly eradicate the above issues and construct an efficient and effective CZSL technique, we suggest a method to improve attribute recognition performance by utilizing diverse Prompt Learning with an Inter/Intra-Modality Fusion Synthesizer in scene understanding involving subtle semantic differences and multiple objects. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Attribute recognition, Prompt-based Learning, Zero-shot Learning.</p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Evolving Quantum Neural Network Operations With Data Re-uploading, Entanglement, and Consciousness Based on Orch or Theory </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Thomas McIver </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">This paper introduces an approach to quantum neural networks that combines the principles of data re-uploading and entanglement. Based on the Orchestrated Objective Reduction (Orch OR) theory proposed by Roger Penrose and Stuart Hameroff, the study explores how quantum mechanical processes can improve neural network capabilities. By reuploading classical data at different stages of computation and utilizing quantum entanglement, the proposed network aims to achieve advanced information processing and learning abilities. This approach not only enhances the network’s performance but also provides insights into the potential quantum basis of consciousness. The incorporation of these quantum operations within a feedback loop further enhances the learning process, potentially resulting in emergent behaviours reminiscent of consciousness. </p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>AIOT-based smart traffic management system </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Ahmed Mahmoud Elbasha and Mohammad M. Abdellatif, Electrical Engineering Department, Faculty of Engineering, The British University in Egypt, Cairo, Egypt </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">This paper presents a novel AI-based smart traffic management system de-signed to optimize traffic flow and reduce congestion in urban environments. By analysing live footage from existing CCTV cameras, this approach eliminates the need for additional hardware, thereby minimizing both deployment costs and ongoing maintenance expenses. The AI model processes live video feeds to accurately count vehicles and assess traffic density, allowing for adaptive signal control that prioritizes directions with higher traffic volumes. This real-time adaptability ensures smoother traffic flow, reduces congestion, and minimizes waiting times for drivers. Additionally, the proposed system is simulated using PyGame to evaluate its performance under various traffic conditions. The simulation results demonstrate that the AI-based system out-performs traditional static traffic light systems by 34%, leading to significant improvements in traffic flow efficiency. The use of AI to optimize traffic signals can play a crucial role in addressing urban traffic challenges, offering a cost-effective, scalable, and efficient solution for modern cities. This innovative system represents a key advancement in the field of smart city infra-structure and intelligent transportation systems. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;"> AI, ITS,IoT, Traffic Management</p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Fair-anonymity: a Novel Fairness Notion for Cryptocurrency </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Taishi Higuchi and Akira Otsuka, Institute of Information Security (IISEC), Kanagawa, Japan </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">In recent years, there has been a growing demand for using tokens of public blockchains like Bitcoin for legitimate transactions. However, the lack of authoritative guarantees on these tokens raises concerns about their potential misuse in criminal activities. Conversely, the introduction of full transparency regulation may stifle the highly innovative cryptocurrency community. This paper introduces a novel concept of fairness, termed Fair-Anonymity, which allows regulatory authorities to probabilistically trace the payer’s ID with the pre-agreed probability based solely on the total amount of the transaction, even when divided into smaller transactions. The Fair-Anonymity protocol can be applied to many blockchains by adding proof to the transaction, in which public verifiers can verify the result. Our scheme cryptographically enforces the revealing probability using k-out-of-n Committed Oblivious Transfer, ensuring that neither the sender nor the receiver can manipulate the probability or alter the committed values, thus disincentivizing illegal high-value transactions. Conversely, enterprises accepting only tokens with Fair-Anonymity proofs can externally demonstrate their commitment to lawful operations. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;"> Blockchain, Security, Electronic-cash, Cryptocurrency, Fairness, Anonymity, Traceability, Oblivious transfer.</p> <br> <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>A Chronological Review of Deepfake Detection: Techniques and Evolutions </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">JINGJING RAO<sup>1</sup> and TETSUTARO UEHARA<sup>2</sup>, <sup>1</sup>Graduate School of Information Science and Engineering. Ritsumeikan University, Japan, <sup>2</sup>College of Information Science and Engineering Ritsumeikan University, Japan </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Since the development of deep learning technology, various new technologies have emerged one after another, greatly facilitating our daily lives. However, the development of these technologies has also brought some troubles, among which Deepfake technology is a typical example. Deepfake technology is mainly used to generate false pictures and videos, or modify real pictures and videos to achieve the purpose of deception. In the early days of this technology, people could often distinguish the authenticity with the naked eye. However, as the technology matures, the generated pictures and videos become more and more realistic, and many criminals have begun to use this technology to commit economic fraud, produce illegal pornographic content, distort political facts and other illegal acts. In order to better understand the importance of Deepfake detection and its related technologies, this article sorts out the main Deepfake detection technologies from 2018 to 2024. We briefly explain the various methods mentioned in the work and organize them into a table form. At the same time, we also set up a series of Q&A sessions, the purpose of which is to comprehensively introduce Deepfake technology and its detection methods from multiple perspectives, so as to help readers fully understand the latest developments and challenges in this field. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Deepfake, Detection, State-of-the-Art,GANs, Deeplearning, Dataset, Traditional method.</p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Automatic Lung Nodule Segmentation in Ct Images Based on U-net Architectures </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Alejandro Jer´onimo<sup>1</sup>, Ignacio Rojas<sup>1</sup>, and Olga Valenzuela<sup>2</sup>, <sup>1</sup>Computer Engineering, Automatics and Robotics Department, University of Granada, 18071 Granada, Spain, <sup>2</sup>Department of Applied Mathematics, University of Granada, 18071 Granada, Spain </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">Lung cancer is the most common type of cancer worldwide, with 2.5 million new cases reported in recent years, according to the World Health Organization. It also has the highest mortality rate, and early diagnosis is crucial for reducing mortality. Deep Learning techniques, particularly computer-aided diagnosis (CAD) systems, have advanced automatic detection of pulmonary diseases. While many studies propose pipelines with complex architectures, the nnU-Net model provides a robust, automatic framework for segmentation across various medical imaging modalities. This work evaluates nnU-Net’s performance in semantic segmentation of nodules of varying sizes by integrating various preprocessing techniques. Results show improved Dice Score and IoU metrics, especially for large nodules. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;"> Deep learning, lung nodules, semantic segmentation, U-Net, nnU-Net.</p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Predicting Type 2 Diabetes Among Social Media Users in Saudi Arabia using Machine Learning </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Saleha Masood & Mousa Ahmad Al-Bashrawi, IRC for Finance and Digital Economy, Saudi Arabia </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The steadily rising cases of Type 2 Diabetes in Saudi Arabia further stress the quest for new early detection strategies. The paper discusses how Instagram posts can be used to develop a prediction model of Type 2 Diabetes among users in Saudi Arabia. We crawled more than 5,266 posts related to health, conducted extensive preprocessing, and performed topic modeling using Latent Dirichlet Allocation on various dimensions such as life style habits, health status, exercising habits, and eating behaviors. Emotion tone and subjectivity were analyzed using TextBlob and further included in features of the predictive model. With TF-IDF vectorization and aggregation at the user level, the Random Forest classifier optimized with GridSearchCV reached an accuracy of 90% with an AUC of 0.92. This study therefore reinforces the possibility of social media for type 2 Diabetes risk prediction. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Type 2 Diabetes, Machine Learning, Machine Learning, Instagram, Predictive Modeling, Sentiment Analysis, TF-IDF Vectorization.</p> <br>   <h6 style="color:black;font-family:classic wide,sans-cserif;font-size:20px"><b>Artificial Intelligence Based Transformation Projects-the Role of Data Sciences (RDS) </b></h6> <p style="color:black;text-align:justify;font-size: 15px;">Antoine Trad, IBISTM, Courbevoie, France </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>ABSTRACT</b></h6> <p style="color:black;text-align:justify;">The Applied Polymathical/Holistic Mathematical Model for Integrating Data Sciences (AHMM4IDS) supports Enterprise’s transformation projects (simply Project). The AHMM4IDS uses various Mathematical Models (MM), that abstract, incorporate, and integrate Data Sciences (DS), AI-Subdomains, Information Communication System (ICS) components with Project’s transformed resources. Transformed resources can be services (and artefacts), success factors (or calibration-variables), business processes (and scenarios), mixed-methods, AI-Models, and adequate Enterprise Architecture (EA) Models (EAM). MMs, mixed-methods’ based services, artefacts, and EAMs can be used to establish set of DS Patterns (DSP) that include DS technics/capabilities, data-platforms’ access (and management), algorithms-functions, mapping-concepts, unbundled services; to model and implement Decision Making Processes’ (DMP) related infrastructure, data-storage(s), components-models, and end-users’ integration. The integration of DSPs enforces and automated DMPs, Project’s validity-checking, and Gap Analysis (GAPA); which all need adapted interfaces to access Enterprise, Project, Data-storage(s), ICS, EAMs, pool(s) of Artificial Intelligence (AI) services, and other types of resources. On the other-hand, DSPs communicate with other, by using Project’s and AI components; and can use also various medias-types formats, like the eXtensible Markup Language (XML) format, and many others. Imported (or exported) DSs’ contents and structures are combined with other Project’s artefacts and components, to deliver DSPs for various AI-Subdomains. </p> <h6 style="color:black;font-family:classic wide,sans-serif;"><b>Keywords</b></h6> <p style="color:black;text-align:justify;">Data Sciences, AI-Subdomains, Polymathical mathematical models, Business and common transformation projects, Enterprise architecture, Artificial intelligence, Qualitative and quantitative research, and Critical success factors/areas.</p> <br>  </div> </div> </div> </div> </section>   <footer class="page-footer grey lighten-1"> <div class="container"> <div class="row"> <div class="col s12 m6"> <h5 class="grey-text lighten-3"> <font color="#FFF">Contact Us</font> </h5> <a href="mailto:ccseit@ccseit2025.org" style="color:#000">ccseit@ccseit2025.org</a> </div> </div> </div> <div class="footer-copyright grey darken-2"> <div class="container center"> Copyright © CCSEIT 2025 </div> </div> </footer>  <script type="text/javascript" src="https://code.jquery.com/jquery-3.2.1.min.js"></script> <script type="text/javascript" src="js/materialize.min.js"></script> <script> $(document).ready(function() { // Custom JS & jQuery here $('.button-collapse').sideNav(); }); </script> </body> </html>