Article
Computer Science, Artificial Intelligence
Bin Dong, Guirong Weng, Ri Jin
Summary: An unsupervised active contour model with Self Organizing Maps (SOM) is proposed in this paper, which effectively segments images with intensity inhomogeneity.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Immunology
Edwin Yuan, Magdalena Matusiak, Korsuk Sirinukunwattana, Sushama Varma, Lukasz Kidzinski, Robert West
Summary: The study introduces Seg-SOM, a method for dimensionality reduction of cell morphology in tissue images, which can resolve cellular tissue heterogeneity and reveal complex tissue architecture. By utilizing a self-organizing map (SOM) artificial neural network to group cells based on morphological features, Seg-SOM allows for cell segmentation, systematic classification, and in silico cell labeling.
FRONTIERS IN IMMUNOLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Firat Ismailoglu
Summary: Zero-Shot Learning (ZSL) is a method that enables the classification of new test classes for which there are no labeled images for training. Existing ZSL methods usually learn a projection from the feature space to the semantic space, but this study introduces a novel method called SOMZSL which connects the feature space and attribute space through comparable intermediate layers. SOMZSL performs as well as or better than existing ZSL methods without dealing with a complex optimization problem, and it can mitigate the domain shift problem inherent in ZSL by using unlabeled test images in the construction of SOMs.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Multidisciplinary Sciences
Hamish Turner, Bahman Lahoorpoor, David M. Levinson
Summary: This study uses map digitization technology to create a historic dataset of road opening dates in Sydney. By georeferencing and analyzing historic maps, spatial data is extracted and placed in a collective vector layer. The study finds that about half of the road links in a significant area of Sydney were already open by the start of the twentieth century, and another half opened within a thirty-year period. The project establishes the foundation for a historic road dataset for Sydney and provides methods and procedures to further develop the dataset.
Article
Computer Science, Artificial Intelligence
Haibo Pen, Quan Wang, Zhaoxia Wang
Summary: Image analysis, particularly image restoration using SOMs, is a research focus in this paper. The proposed boundary precedence image inpainting method utilizes SOMs to separate damaged images into layers and calculates the filling order based on boundary precedence. The restoration process focuses on the boundary patches first for effective repair of both textural and structural information.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Chemistry, Analytical
Sureerat Makmuang, Anupun Terdwongworakul, Tirayut Vilaivan, Simon Maher, Sanong Ekgasit, Kanet Wongravee
Summary: Weedy rice is a difficult weed to manage in rice-growing regions, as it is hard to distinguish from cultivated rice. This study proposes a novel classification approach using artificial neural networks and near-infrared hyperspectral imaging to directly discriminate weedy rice. The method showed high accuracy in classifying weedy rice samples.
MICROCHEMICAL JOURNAL
(2023)
Article
Chemistry, Analytical
Lisiane Esther Ekemeyong Awong, Teresa Zielinska
Summary: The objective of this article is to develop a methodology for selecting the appropriate number of clusters to group and identify human postures using neural networks with unsupervised self-organizing maps. The use of quality scores to determine the number of clusters frees the expert to make subjective decisions about the number of postures, enabling the use of unsupervised learning. The findings show that DS offers good quality in posture recognition, effectively following postural transitions and similarities.
Article
Environmental Sciences
Aksel S. Danielsen, Tor Arne Johansen, Joseph L. Garrett
Summary: Hyperspectral remote sensing provides detailed information about the optical response of a scene. Self-Organizing Maps (SOMs) can partition hyperspectral datasets into clusters to enable more analysis on-board imaging platforms and reduce downlink time. The on-board performance of the SOM algorithm is calculated within different satellite operational procedures, and it is found that SOMs can run on target hardware, achieve low quantization error, and classify with high accuracy when class labels are assigned.
Article
Computer Science, Artificial Intelligence
Slavisa Jovanovic, Hiroomi Hikawa
Summary: Self-organizing feature maps (SOMs) are commonly used in clustering and data dimensionality reduction. However, their high computational cost limits their real-time online processing. This article surveys the hardware implementations of SOMs and discusses the challenges and trends for their adoption as hardware accelerators in various applications.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Chemistry, Analytical
Laura Nicolas-Saenz, Agapito Ledezma, Javier Pascau, Arrate Munoz-Barrutia
Summary: Classifying pixels according to color and segmenting areas are necessary steps in computer vision tasks. The challenges in properly classifying pixels based on color lie in the differences between human perception, linguistic terminology, and digital representation. To address this, a novel method combining geometric analysis, color theory, fuzzy color theory, and multi-label systems was proposed. The method shows accuracy in color analysis and provides a standardized alternative for color naming recognizable by both humans and machines.
Article
Geography, Physical
Kristen L. Underwood, Donna M. Rizzo, Mandar M. Dewoolkar, Michael Kline
Summary: Given limited resources for managing erosion hazards and water quality along rivers, stakeholders in water resource management could benefit from tools to identify river reaches prone to sediment loading. The Self-Organizing Map (SOM) is a useful tool for clustering multivariate observations and analyzing complex, nonlinear river systems. Through multiple stages of SOM application, we identified seven sediment regimes in river reaches based on stream geomorphic assessment data.
Article
Chemistry, Analytical
Gilbert A. Angulo-Saucedo, Jersson X. Leon-Medina, Wilman Alonso Pineda-Munoz, Miguel Angel Torres-Arredondo, Diego A. Tibaduiza
Summary: Improvements in computing capacity have enabled the development of machine learning algorithms for structural health monitoring (SHM). This study focuses on configuring a data acquisition system, developing a damage classification methodology, and using machine learning algorithms to detect and classify damages. The results validate the effectiveness of the SKN and XYF networks in damage classification tasks.
Article
Computer Science, Artificial Intelligence
Leonardo A. Dias, Augusto M. P. Damasceno, Elena Gaura, Marcelo A. C. Fernandes
Summary: The study introduces a fully parallel architecture for SOM that significantly improves processing speed and resource efficiency.
Article
Computer Science, Artificial Intelligence
Benoit Audelan, Dimitri Hamzaoui, Sarah Montagne, Raphaele Renard-Penna, Herve Delingette
Summary: This paper introduces a novel approach to jointly estimate a reliable consensus map and assess the presence of outliers and confidence in each rater. The robust approach is based on heavy-tailed distributions, allowing for local estimates of rater performance and the introduction of bias and spatial priors.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Energy & Fuels
Na Xu, Wei Zhu, Ru Wang, Qiang Li, Zhiwei Wang, Robert B. Finkelman
Summary: Understanding the modes of occurrence of elements in coal is crucial for evaluating their environmental and health impacts and recovering critical elements from coal ash. This paper introduces the application of the self-organizing map algorithm in analyzing the modes of occurrence of elements in coal and compares it with the average linkage hierarchical clustering algorithm. The results show that the self-organizing map algorithm provides more consistent results with the geochemical nature and previous investigations.
INTERNATIONAL JOURNAL OF COAL GEOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
David Valle-Cruz, Vanessa Fernandez-Cortez, Asdrubal Lopez-Chau, Rodrigo Sandoval-Almazan
Summary: Investors are constantly monitoring the behavior of stock markets which is influenced by social media reactions and emotions, especially during pandemics. Through financial sentiment analysis of Twitter data and financial indices, it was found that the markets reacted 0 to 10 days after information was shared on Twitter during the COVID-19 pandemic and 0 to 15 days after during the H1N1 pandemic. A lexicon-based approach and correlation analysis using SenticNet were effective in detecting highly shifted correlations. The most influential Twitter accounts during the pandemic were found to have a high correlation between sentiments on Twitter and stock market behavior.
COGNITIVE COMPUTATION
(2022)
Article
Chemistry, Applied
Farid Garcia-Lamont, Matias Alvarado, Asdrubal Lopez-Chau, Jair Cervantes
Summary: In this study, a nucleus segmentation proposal of white blood cells using chromatic features is presented. By analyzing the images and processing hue components, the method is able to accurately locate the nucleus of cells. Experimental results show that the proposed method outperforms current techniques in various metrics.
COLOR RESEARCH AND APPLICATION
(2022)
Article
Computer Science, Information Systems
Felipe Castro-Medina, Lisbeth Rodriguez-Mazahua, Asdrubal Lopez-Chau, Jair Cervantes, Giner Alor-Hernandez, Isaac Machorro-Cano, Mario Leoncio Arrioja-Rodriguez
Summary: The proper storage and management of multimedia data is of great interest to industry and academia. Database fragmentation is crucial in distributed data management environments to reduce costs and improve response time performance. Dynamic fragmentation techniques can adapt to changing multimedia database access patterns and provide better performance through content-based queries.
Article
Mathematics
Diana L. Gonzalez-Baldovinos, Pedro Guevara-Lopez, Jose Luis Cano-Rosas, Jorge Salvador Valdez-Martinez, Asdrubal Lopez-Chau
Summary: The response times of computer tasks depend on the hardware and software, and even in real-time operating systems, the response times can vary as instances evolve. This study proposes a model to reconstruct the dynamics of response times for high-priority task instances, allowing for analysis of their offline behavior under specific working conditions.
Article
Chemistry, Multidisciplinary
Isaac Machorro-Cano, Jose Oscar Olmedo-Aguirre, Giner Alor-Hernandez, Lisbeth Rodriguez-Mazahua, Jose Luis Sanchez-Cervantes, Asdrubal Lopez-Chau
Summary: This paper proposes an SOA model called SCM-IoT for incorporating AI into IoT systems. The model addresses coordination issues in IoT systems by offering a mediator for storage, production, discovery, and notification of relevant data, allowing incremental development from multiple perspectives.
APPLIED SCIENCES-BASEL
(2022)
Article
Chemistry, Multidisciplinary
Nidia Rodriguez-Mazahua, Lisbeth Rodriguez-Mazahua, Asdrubal Lopez-Chau, Giner Alor-Hernandez, Isaac Machorro-Cano
Summary: Data warehousing provides a systematic approach for enterprise administrators to utilize data effectively for strategic decision-making. One of the main challenges faced by data warehouse designers is fragmentation, with FTree being a horizontal fragmentation method that uses decision trees to improve efficiency in data warehouse design. Experimental results confirm the efficacy of this method in improving OLAP query response time and data loading maintenance tasks.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Roberto Carlos Valdes Garcia, Farid Garcia Lamont, Rodolfo Zola Garcia Lozano, Asdrubal Lopez Chau, Rodolfo Fraga, Gonzalo Lastra Medina
Summary: This paper proposes the use of Artificial Neural Networks, Random Forest, Decision Trees, and Support Vector Regression for parameter extraction of a resistive load inverter circuit with a Thin Film Transistor (TFT). The supervised learning methods were found to be useful and achieved a good fit between measurements and parameters. The Neural Networks obtained better results, with an average error rate of 6.04%. The method was also successfully applied to real NMOS measurements with minimum errors of up to 0.43%.
IEEE LATIN AMERICA TRANSACTIONS
(2023)
Article
Engineering, Electrical & Electronic
Roberto C. Valdes, Farid Garcia, Rodolfo Z. Garcia, Asdrubal Lopez, Norberto Hernandez
Summary: This work proposes a supervised learning method as an alternative to analytical and optimization methods for extracting parameters in Indium Gallium Zinc Oxide Thin-Film Transistors with aluminium contacts. The method involves generating a set of I-V curves of the device using Spice software, which serves as input samples for the Artificial Neural Networks to predict parameters like threshold voltage, transconductance, and contact resistance. The results show that ANNs can achieve modeling of physical measurements with error rates below 5% for the first two parameters and between 0.06% and 4.62% for the three parameters. A comparison with analytically extracted parameters is also presented.
JOURNAL OF MATERIALS SCIENCE-MATERIALS IN ELECTRONICS
(2023)
Proceedings Paper
Computer Science, Information Systems
David Valle-Cruz, Vanessa Fernandez-Cortez, Asdrubal Lopez-Chau, Rafael Rojas-Hernandez
Summary: This paper explores the potential of machine learning and synthetic data in public budget simulations, using historical data from the Mexican government. It finds that Random Forest performs the best among the algorithms studied. The study could aid public budget decision-making and simulate scenarios in government.
ELECTRONIC GOVERNANCE WITH EMERGING TECHNOLOGIES, EGETC 2022
(2022)
Proceedings Paper
Computer Science, Theory & Methods
Jared Cervantes, Dalia Luna, Jair Cervantes, Farid Garcia-Lamont
Summary: Vessel segmentation is a crucial task in retinal image analysis. This study presents a novel vessel segmentation algorithm using genetic algorithms to optimize the parameters. Experimental results on two datasets demonstrate the effectiveness of the proposed method.
INTELLIGENT COMPUTING THEORIES AND APPLICATION (ICIC 2022), PT I
(2022)
Article
Computer Science, Information Systems
Alma-Delia Cuevas-Rasgado, Carlos-Omar Gonzalez-Moran, Asdrubal Lopez Chau, Ulrich Broeckl
Summary: This paper describes a telemetry system called Rescue in Collapsed Building (RICB) for rescuing victims in collapsed buildings using Artificial Neural Networks (ANN) and Raspberry Pi technology. The system improves the identification of human patterns by analyzing sensor readings, such as movement, sound, and temperature, and can determine if a victim is alive and enable communication with them.
COMPUTACION Y SISTEMAS
(2022)