Applied and Computational Engineering

- The Open Access Proceedings Series for Conferences

Volume Info.

  • Title

    Proceedings of the 4th International Conference on Signal Processing and Machine Learning

    Conference Date

    2024-01-15

    Website

    https://www.confspml.org/

    Notes

     

    ISBN

    978-1-83558-331-9 (Print)

    978-1-83558-332-6 (Online)

    Published Date

    2024-03-15

    Editors

    Marwan Omar, Illinois Institute of Technology

Articles

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241016

    An overview of Neural Radiance Fields

    Synthesizing controllable, photo-realistic images and videos is one of the fundamental goals of computer graphics. Neural rendering is a rapidly emerging field in image synthesis that allows a compact representation of scenes, and by utilizing neural networks, rendering can be learned from existing observations. Neural Radiance Fields (NeRF) implement an effective combination of Neural Fields and the graphics component Volume rendering. It achieves the first photo-level view synthesis effect using an implicit representation. Unlike previous approaches, NeRF chooses Volume as an intermediate representation to reconstruct an implicit Volume. Although the advantages of NeRF are apparent, there are many drawbacks in the original version of NeRF: it is slow to train and render, requires a large number of perspectives, can only represent static scenes, and the trained NeRF representation does not generalize to other scenes. This report focuses on optimizing the shortcomings mentioned above of NeRF by scholars in the last three years and analyzes the solutions to the problems of NeRF from several perspectives.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241019

    Assessing the effectiveness of special education services on fifth grade math scores: Using traditional and machine learning methods with ECLS-K data

    The Early Childhood Longitudinal Study, Kindergarten Class of 1998-99 (ECLS-K) is a well-known research endeavor in the field of child development. In this research, some special education services are offered to those students who need supplementary support in some aspects. In this paper, our study aims to estimate the average treatment effect on students’ fifth grade math scores and assesses the effectiveness of these special education services based on the ECLS-K dataset, through both machine learning methods and traditional methods. We introduce Donald Rubin’s causal model and Propensity Score Analysis in the part of traditional methods, and Ordinary Least Squares (OLS), Targeted Maximum Likelihood Estimation (TMLE), Bayesian Additive Regression Trees (BART), Generalized Random Forests (GRF) and Double Machine Learning (DML) in the part of machine learning methods. Finally, we employ Propensity Score Matching, OLS and BART to estimate the ATE. All estimated ATEs are significantly different from zero. The estimated ATEs are found to be minus, suggesting that these special education services may have a negative effect on students’ fifth grade math scores. Obviously, this conclusion is inconsistent with the original intent of these services, which aimed to have a positive impact.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241020

    Flexible lower limb exoskeleton rehabilitation robot: A review

    Today, medical and rehabilitation exoskeletons are chosen by more therapists to treat individuals with lower limb injuries. A flexible lower limb exoskeleton (FLLE) is a new exoskeleton robot for rehabilitation. Compared with rigid lower extremity exoskeleton (RLLE), FLLE has the advantages of lower weight, better compliance, lower energy consumption, and higher safety. This paper reviews the development and innovation of FLLE in recent years from the aspects of driving mode, design requirements and critical technologies. The characteristics of existing FLLE products are analyzed and summarized, and the challenges and future development directions of the research, such as the construction of a general, flexible exoskeleton power system model, motion intention and fast processing of pattern recognition, are discussed.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241021

    Research on virtual reality in the live stream industry

    The live-streaming industry has seen significant growth in recent years, driven by advances in virtual reality technology and the growing popularity of online content consumption. This article explores the potential of virtual reality (VR) to revolutionize the live streaming industry, and analyzes VR's impact on user experience and content creation. After discussing the potential impact of VR technology on broadcasters, content creators and viewers, the article also elaborates on the main challenges and opportunities related to VR and live streaming platforms. The findings of this study provide valuable insights for researchers, practitioners, and industry stakeholders interested in understanding the role of VR in shaping the future of live streaming.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241022

    Development of a smart health monitoring system for elderly care

    This paper describes an intelligent health monitoring system developed for detecting the health of the elderly. This paper introduces an intelligent health monitoring system used to detect the health status of the elderly. With the progress of mobile communication technology and the increasing demand for personal intelligence, wearable devices have become more and more popular products. The system can provide personalized services for the elderly, and also provide more comprehensive and accurate data support for medical staff, which is expected to play an important role in the future care of the elderly.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241023

    A survey of text generation models

    In this article, I propose four model classifications to summarize the characteristics and analyze the advantages and disadvantages of text generation models that have emerged in recent years, so as to give researchers an overall overview. The models based on the decoder only use the decoder for text extraction, and its output only depends on the previous output. The models based on the encoder-decoder, on the other hand, refer to both the encoder's output and the previous prediction. I've deliberately categorized prefix models and ensemble models to highlight their differences. I also present the current state of the text generation field and compare the advantages and disadvantages of several of these models. Finally, I summarize the difficulties encountered in the field of text generation and provide a research direction for the field. In the module Challenges, I focused on the problem of scarcity regarding datasets. The current solutions are given, as well as the efforts made by relevant workers on domain-specific datasets.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241024

    Personalized medical recommendation system supported by medical data

    A personalized medical recommendation system is an intelligent system that utilizes medical data to provide targeted medical advice and services to individuals. With the lack of accumulation and development of medical data, personalized medical recommendation systems have great potential in improving medical effectiveness and saving medical resources. This article aims to explore the principles, methods, and applications of personalized medical recommendation systems based on medical data.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241025

    Artificial intelligence and its impact on the study abroad industry

    Artificial intelligence has played an important role in the overseas study industry. It provides students with more convenient ways to inquire and apply for overseas study information, and helps students to better choose their study abroad goals and schools through intelligent recommendation systems and personalized counseling services. At the same time, artificial intelligence is also applied to language learning and document writing, providing more efficient learning tools and writing AIDS. However, AI also presents some challenges, such as information security and privacy protection issues, as well as vicarious impacts on human professional advisors.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241026

    Remote sensing image scene classification based on ResNeSt

    The study used the deep learning method to achieve the natural scene classification of remote sensing images, which were taken by the satellite Tiangong-2. Because of the diversity of remote sensing images, a Convolutional Neural Network (CNN) model that can complete the task of classifying natural scenes of remote sensing images was constructed using the variant ResNeSt based on the Residual Neural Network (ResNet). The NaSC-TG2 remote sensing image dataset released by the Space Application Engineering and Technology Center of the Chinese Academy of Sciences was used in this work. The dataset consists of 20,000 photos that are grouped into ten scene groups on average, with 2,000 images per scene category. And nine models including ResNet50, ResNet101, ResNet200, SE-ResNet50, SE-ResNeXt50, SE-ResNeXt101, SE-ResNeXt152, ResNeSt50, ResNeSt101 and ResNeSt200 were compared and tested on the NaSC-TG2 dataset. After training and testing on the dataset, ResNeSt101 achieved better results than other comparative models in the end, with the highest accuracy of 98.52% on the testing sets. This study offers a technique for categorizing remote sensing picture scenes and has made some significant contributions to space geoscience and application research.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241027

    Optimal control of traffic light signals using stochastic simulation

    Traffic congestion is one of the serious problems facing modern cities, posing a huge challenge to people’s travel and urban development. And it has been proved that traffic lights should be responsible for the congestion instead of too many cars on the road occasionally. So it is very important to find the optimal control of the traffic light signal. The purpose of this experiment is to explore ways to optimally control traffic light signals in order to reduce the average waiting time for people and vehicles at intersections. This paper used a stochastic simulation approach, based on an assumed Poisson process and Gamma distribution, to simulate the specific time for vehicles and pedestrians to arrive at the intersection over the course of an hour, and used this to calculate the average waiting time. We investigated the effect of the duration of red and green lights on the average waiting time and wrote the corresponding code for simulation and calculation through MATLAB. The experimental results show that shorter red light duration and moderate green light duration can achieve optimal results when there are more cars. On the contrary, shorter green light duration and moderate red light duration can achieve optimal results when there are more pedestrians. We reached this conclusion through experiments and evaluated the hypotheses in the experiments.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241028

    Research on path planning of 4-link robot based on optimized Q-learning algorithm

    In Q-learning algorithm, the balance between exploration and exploitation is crucial for the success of the entire algorithm. If the algorithm is too biased towards exploration, it may waste too much time on unnecessary paths, while if the algorithm is too biased towards exploitation, it may miss better paths. Ensuring the balance between exploration and exploitation in Q-learning algorithms requires comprehensive consideration of various aspects of the algorithm, including learning rate, state space representation, reward function design, policy selection, and addition of random noise. This paper proposes and verifies the construction of the state and action sets for a three-degree- of-freedom robotic arm within a specified space, where the reward values are reasonably set, and Q-learning with an introduced exploration factor is used as an improvement to prevent the robotic arm from insufficient exploration in action selection, which can lead to the algorithm failing to find a better strategy. The results of experiment illustrate that introducing the exploration factor can significantly enhance the convergence speed of the algorithm.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241029

    A bibliometric review of technical applications of deep learning in computer vision

    Computer vision is a field of Artificial Intelligence (AI), which is the field of technology that allows computers and systems to capture meaningful information from visual inputs such as images, videos, etc. through external cameras, data, and internal algorithms, and to take action or provide recommendations based on the captured information. Common application scenarios of computer vision are image recognition, image classification, object detection, pose detection, image segmentation, etc. In the era of artificial intelligence, computer vision plays the role of a "perceptron", which is the "eye" of the artificial intelligence era, and provides the "planning" and "decision-making" for the artificial intelligence. "Decision-making" provides an effective source of information and information support, and promotes the breakthrough development of computer vision technology while realising the iterative update of AI technology. By 2022, the market value of this field reaches $48.6 billion. In this paper, a visual knowledge graph data analysis of relevant literature on Web of Science (WOS) and China National Knowledge Infrastructure(CNKI) was conducted through CiteSpace. Among them, it focuses on analysing the development trend of image restoration technology based on diffusion model, which provides some help for scholars to carry out research.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241030

    Research on recommendation algorithm based on user sentiment analysis

    With the development of Internet technology, the recommendation system is becoming an essential part of major e-commerce platforms, social media platforms and other application fields. The main purpose of the recommendation algorithm is to provide users with personalized and accurate recommendations of goods, services and information. Traditional recommendation algorithms are mainly based on information such as users’ historical behavior to recommend similar items to users. However, only considering the historical behavior of users cannot fully reflect the individual needs of users because the emotions and interests of users are dynamic. Therefore, introducing user emotion is an important direction of a personalized recommendation system. Based on user emotional analysis,a recommendation algorithm aims to understand user preferences and interests by analyzing their emotional reactions. It utilizes emotional data to generate personalized recommendations, considering the sentimental factors that influence consumer purchasing decisions. By incorporating emotional intelligence into the recommendation process, this algorithm aims to improve the accuracy and effectiveness of product suggestions, ultimately enhancing user satisfaction. This paper will discuss the introduction of the user emotion recommendation algorithm and explore its implementation and application scenarios.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241031

    Research on predicting Ames housing price based on forward selection regression and principal component regression

    This study aimed to predict home prices in Ames, Iowa, using two regression techniques: forward selection regression and principal component regression. The research began with exploratory data analysis and several pre-processing steps. Features are classified, overlapping features are merged, missing values are processed appropriately, and predictors of zero or near-zero variance are removed. The data is then scaled to solve the problem of data fragmentation. The results indicated that the forward selection regression method had higher prediction accuracy and a lower root mean square error (RMSE) value, suggesting its superiority in predicting home prices. The findings of this study have practical implications for various stakeholders in the real estate market. By focusing on the identified areas that contribute significantly to the value of real estate, stakeholders can make informed decisions to enhance their investments. This research provides valuable guidance for homeowners, real estate agents, and property developers in understanding the factors influencing home prices in Ames, Iowa.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241032

    Detection of malicious encrypted communication based on feature engineering and machine learning

    With the advancement of network technology, malicious encrypted communication has become a covert network threat and has drawn significant attention in the field of cybersecurity. Network threats are increasingly severe, and traditional detection methods struggle to cope with the intricate changes in malicious encrypted communication. Therefore, in order to find effective approaches to detect malicious encrypted communication, this research focuses on exploring the detection methods of malicious encrypted communication based on feature engineering and machine learning. The crucial role of feature engineering and the application of machine learning methods in this context are extensively discussed. The research conclusions indicate that the proper design of feature engineering and method selection can improve the detection accuracy and efficiency. However, practical applications still face challenges such as data scarcity and limited computational resources. Therefore, future research directions are proposed, including further optimizing feature engineering methods, developing feature representations more suitable for detecting malicious encrypted communication, and exploring more efficient deep learning models. The significance of this research lies in providing theoretical guidance and practical application advice for professionals and researchers in the field of cybersecurity, contributing actively to building a safer and more stable network environment, and jointly safeguarding the security and stability of the digital world.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241033

    Application and development trends of flexible pressure sensors

    Flexible pressure sensors have emerged as a significant sensor technology with promising and broad applicability in various sectors, such as electronic skin and health monitoring, in recent years. This study evaluates the principles, classifications, uses, and development trends of flexible pressure sensors. It reviews the research progress in terms of preparation, design, and materials, discusses technical challenges and limitations, and proposes potential solutions and future development directions. The wide range of applications and projected advancements for flexible pressure sensors in different fields are summarized. With the continuous advancement of technology and society, the utilization of flexible pressure sensors is expected to become increasingly prevalent. In the future, these sensors will play a pivotal role in diverse fields, including health monitoring, human-computer interaction, and environmental monitoring. This paper aims to aid future researchers in comprehending the fundamental principles of flexible pressure sensors, various types of classifications, different application fields, and future development trends more quickly and thoroughly.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241034

    The application of Rasa framework in the interaction between players and game NPCs

    In the current era of thriving electronic games, some players have articulated a novel requirement for electronic games – the need for greater freedom and autonomy within the gaming experience. In traditional electronic games’ interaction methods, there often exist issues of low engagement and insufficient immersion due to the limited narrative approach. At the same time, Rasa technology has made significant progress in the field of open-source dialogue systems. Rasa has the ability to understand and process multiple rounds of conversations, supports multilingual and multi-channel applications, and provides rich tools and components to enable developers to quickly build intelligent dialogue systems which is likely to be applied within the game industry. In the future, Rasa will continue to develop and provide people with a more intelligent and personalized conversation experience. This paper will explore a new application of Rasa platform for players to interact with non-player characters (NPCs) in electronic games which contributes to solving the issues.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241037

    A review of research on the control of quadrotor UAVs based on deep learning PID algorithm

    As a flexible and reliable flying platform, the Quadrotor Unmanned Aerial Vehicle (UAV) has the advantages of convenient operation, simple structure and strong maneuverability. It is widely used in different fields and has great military and civilian value. In today’s increasingly complex application environment and situations, it is more difficult to achieve precise control of quadrotor UAVs. The conventional Proportional-Integral-Derivative (PID) algorithm is often unsatisfactory for external disturbances and system nonlinearities during flight. The PID algorithm based on deep learning has better control accuracy, dynamic performance and stability for quadrotor UAVs. Good control effect, so it can be better applied to the current situation. The article mainly summarizes the research status of the control of quadrotor drones based on deep learning PID algorithm, aiming to provide useful reference for researchers, developers and industry stakeholders. In terms of content, the three parts of deep learning quadrotor UAV control, PID algorithm optimization research, and quadrotor UAV control system design are reviewed and summarized, and finally possible improvement schemes and future research directions are proposed.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241038

    AI-based text-to-image synthesis: A review

    The traditional methods of art generation, such as texture synthesis and texture mapping, have been instrumental in crafting digital art for decades. They are used as artistic tools to design and map textures onto 3D models, thereby generating 2D images or animations. However, they can only generate simple, repetitive images. Thanks to the rapid development of deep learning and artificial intelligence, today’s text-to-image synthesis (T2IS) models can generate high-quality, realistic images matching the textual description given by the users. This review paper aims to present a comprehensive exploration of groundbreaking AI-based T2IS models in history. We start with an in-depth analysis of the fundamental concepts that underpin T2IS models, followed by an introduction to the primary, or vanilla, models that have served as the foundation for the fields’ development. Then, we delve into the examination of several groundbreaking AI-based T2IS applications, from GAN-based to Diffusion-based models, demonstrating their ability to produce high-quality, contextually accurate images from textual descriptions, along with their strengths and weaknesses. In the end, we will discuss the current challenges and potential future directions in the realm of T2IS.

  • Open Access | Article 2024-03-15 Doi: 10.54254/2755-2721/45/20241041

    Analysis of the current development and future prospect of autonomous driving

    Autonomous driving technology, a rapidly advancing field, holds great potential to transform the way people commute and travel. This technology enables vehicles to operate without human intervention through the integration of sensors, cameras, and sophisticated algorithms. The race to perfect autonomous driving is well underway with major automobile manufacturers like Tesla, Ford, and General Motors heavily invested in research and development. This paper mainly discusses the current development status of autonomous driving, its advantages and challenges. The key benefit of autonomous driving lies in its potential to significantly enhance safety on the roads. Moreover, autonomous driving can mitigate traffic congestion issues and enhance fuel efficiency, ultimately leading to a more sustainable and eco-friendly transportation system. However, this technological advancement does not come without its challenges. The lack of a robust regulatory framework poses a hurdle to adopting autonomous vehicles. Additionally, the high cost associated with developing and implementing autonomous driving technology has been a barrier to its accessibility. Although autonomous driving technology is still in its early stages, it holds immense promise for the future. The potential benefits of autonomous driving, such as improved safety, reduced traffic congestion, and enhanced fuel efficiency, make it an exciting prospect for the future of transportation. Nonetheless, overcoming challenges related to regulation, implementation costs, and security remains crucial for the widespread integration of this technology. As research and development efforts in autonomous driving continue, it can be anticipated that a more sustainable and efficient transportation system that could fundamentally reshape people’s daily lives.

Copyright © 2023 EWA Publishing. Unless Otherwise Stated