Review
Radiology, Nuclear Medicine & Medical Imaging
Fakhriddin Pirlepesov, Lydia Wilson, Vadim P. Moskvin, Alex Breuer, Franz Parkins, John T. Lucas, Thomas E. Merchant, Austin M. Faught
Summary: Challenges in proton therapy include patient selection, plan consistency, and biological uncertainties. Knowledge-based planning may help address these challenges. The study developed a three-dimensional dose and LETD-prediction model for cranial proton therapy, showing promising results in dose accuracy and LETD agreement.
Article
Biochemical Research Methods
Oezlem Muslu, Charles Tapley Hoyt, Mauricio Lacerda, Martin Hofmann-Apitius, Holger Froehlich
Summary: The study proposes a novel approach, GuiltyTargets, for prioritization of putative targets using attributed network representation learning and positive-unlabeled learning. The evaluation on multiple disease datasets demonstrates its superiority over previous methods and its potential for target repositioning across related diseases.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2022)
Article
Geosciences, Multidisciplinary
Moritz Feigl, Katharina Lebiedzinski, Mathew Herrnegger, Karsten Schulz
Summary: This study evaluates the performance of six different machine-learning models for daily water temperature prediction in 10 Austrian catchments, with FNNs and XGBoost performing best in majority of the catchments. The importance of hyperparameter optimization is highlighted, showing varied performance for different models.
HYDROLOGY AND EARTH SYSTEM SCIENCES
(2021)
Article
Engineering, Multidisciplinary
Alexander Henkes, Henning Wessels
Summary: This study introduces a generative adversarial network tailored towards three-dimensional microstructure generation, which can learn the underlying properties of the material from a single mu CT-scan without the need of explicit descriptors. During prediction time, the network can generate unique three-dimensional microstructures with the same properties of the original data in seconds and consistently high quality.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2022)
Article
Mathematics
Adnan Bashir, Muhammad Ahmed Shehzad, Aamna Khan, Ayesha Niaz, Muhammad Nabeel Asghar, Ramy Aldallal, Mutua Kilai
Summary: This study investigates a new hybrid model, the wavelet bootstrap quadratic response surface, for accurate streamflow prediction. The results show that this model provides the most efficient results.
JOURNAL OF MATHEMATICS
(2023)
Article
Engineering, Mechanical
Zekun Xu, Jun Chen, Jiaxu Shen, Mengjie Xiang
Summary: Urban seismic damage assessment is an emerging research topic due to global urbanization trend. Traditional methods may suffer from accuracy or efficiency issues, while machine learning methods lack scalability and real datasets. To tackle this issue, this paper proposes an artificial neural network framework for simultaneously predicting nonlinear seismic responses of all buildings in a cluster. The framework aggregates information from historical response records and physical characteristics to improve performance. Experimental results demonstrate high computational efficiency and accuracy.
ENGINEERING FAILURE ANALYSIS
(2023)
Article
Engineering, Mechanical
Long Bai, Fei Xu, Xiao Chen, Xin Su, Fuyao Lai, Jianfeng Xu
Summary: The use of artificial intelligence to predict the dimensional accuracy of machined parts is important in manufacturing. In this study, a predictive model for precision milling of thin-walled structural components is developed. The model classifies different features of structural components based on their dimensional errors and achieves higher accuracy compared to traditional machine learning methods.
FRONTIERS OF MECHANICAL ENGINEERING
(2022)
Article
Construction & Building Technology
Quang-Viet Vu, Sawekchai Tangaramvong, Thu Huynh Van, George Papazafeiropoulos
Summary: The paper proposes two hybrid metaheuristic optimization and artificial neural network (ANN) methods, GA-ANN and PSO-ANN, for predicting the ultimate axial compressive capacity of CFDST columns. These methods optimize the ANN model's architecture by dynamically adjusting the number and sizes of hidden layers, as well as the weights and biases of the neurons, using genetic algorithm (GA) and particle swarm optimization (PSO) approaches. With Bayesian regularization, these techniques enhance the optimization process and outperform standard ANNs in predicting the capacity of CFDST columns.
STEEL AND COMPOSITE STRUCTURES
(2023)
Article
Multidisciplinary Sciences
Zhengcai Li, Xinmin Hu, Chun Chen, Chenyang Liu, Yalu Han, Yuanfeng Yu, Lizhi Du
Summary: This paper investigates the optimization algorithms based on machine learning for settlement prediction. By comparing the performance of different algorithms, the study finds that Sparrow Search Algorithm (SSA) significantly improves the optimization effect of the gradient descent model and enhances its stability to a certain degree.
SCIENTIFIC REPORTS
(2022)
Article
Computer Science, Artificial Intelligence
Xu Huang, Bowen Zhang, Yunming Ye, Shanshan Feng, Xutao Li
Summary: This paper proposes a novel neural network-based method for tackling 3D multi-plane spatiotemporal prediction tasks. By separating the information interactions into three types, i.e., inter-level, intra-level, and unknown residual interactions, a new spatiotemporal module is designed for prediction. Experimental results show that the proposed method achieves a significant improvement and state-of-the-art performance on a dataset of atmospheric temperature changes.
APPLIED INTELLIGENCE
(2023)
Article
Engineering, Multidisciplinary
Hisham Alghamdi, Chika Maduabuchi, Abdullah Albaker, Ibrahim Alatawi, Theyab R. Alsenani, Ahmed S. Alsafran, Abdulaziz Almalaq, Mohammed AlAqil, Mostafa A. H. Abdelmohimen, Mohammad Alkhedher
Summary: This research aims to find the best surrogate performance prediction model for a solar PV-TE module with different semiconductor materials. Several surrogate machine learning models, including recurrent, time delay, and regular artificial neural networks, were trained using expensive finite element generated data. The results show that the optimal machine learning model for solar PV-TE performance modelling is an ANN with two hidden layers and five neurons per layer. Lithium nitride oxide PV-TE outperforms bismuth telluride PV-TE by 65% under 25 Suns. The surrogate ANN also significantly outperforms conventional numerical simulations.
ENGINEERING SCIENCE AND TECHNOLOGY-AN INTERNATIONAL JOURNAL-JESTECH
(2023)
Article
Engineering, Aerospace
Berkant Konakoglu, Funda Kutlu Onay, Salih Berkan Aydemir
Summary: This study evaluates the prediction capability of a novel hybrid artificial neural network-GTO (Gorilla Troops Optimizer) for predicting zenith wet delay (ZWD), which has not been applied in previous literature. The developed ANN-GTO is compared with two training algorithms, Levenberg-Marquardt and Bayesian Regularization, as well as improved ANN methods to demonstrate its efficiency. The results show that all ANN models enhanced with GTO outperform the classic ANN and other hybrid ANN models in predicting ZWD.
ADVANCES IN SPACE RESEARCH
(2023)
Article
Biology
Dorota Lis-Studniarska, Marta Lipnicka, Marcin Studniarski, Robert Irzmanski
Summary: The aim of this study was to determine the impact of patient's potential risk factors on the occurrence of low-energy fractures and their hierarchy. Various methods were used, including artificial neural networks, to predict the risk of fractures. The most important risk factors were found to be age, chronic kidney disease, neck T-score, and serum phosphate level. These findings contribute to the early identification and treatment of patients at risk of low-energy fractures.
Article
Environmental Sciences
Hamid Darabi, Omid Rahmati, Seyed Amir Naghibi, Farnoush Mohammadi, Ebrahim Ahmadisharaf, Zahra Kalantari, Ali Torabi Haghighi, Seyed Masoud Soleimanpour, John P. Tiefenbacher, Dieu Tien Bui
Summary: A new hybrid machine learning algorithm called MultiB-MLPNN was developed for urban flood susceptibility mapping and tested in Amol City, Iran, showing the best predictive performance. The model is useful for generating realistic flood susceptibility maps in data-scarce urban areas, aiding in developing risk-reduction measures to protect urban areas from devastating floods.
GEOCARTO INTERNATIONAL
(2022)
Article
Water Resources
Jayashree Pal, Dibakar Chakrabarty
Summary: Missing data is common in hydrology and poses challenges in developing data-driven models. This paper evaluates the effectiveness of four imputation techniques in developing groundwater pollution prediction models and suggests that imputation techniques can be effective in such circumstances.
HYDROLOGICAL SCIENCES JOURNAL
(2023)
Article
Computer Science, Information Systems
Bruno Iochins Grisci, Mathias J. Krause, Marcio Dorn
Summary: The study introduces a relevance aggregation algorithm that combines the relevance computed from multiple samples by a neural network to generate scores for each input feature. Two visualization methods for learned patterns were presented to enhance model comprehension. The method accurately identifies the most important features for network predictions.
INFORMATION SCIENCES
(2021)
Article
Chemistry, Multidisciplinary
Ederson Sales Moreira Pinto, Bruno Cesar Feltes, Conrado Pedebos, Marcio Dorn
Summary: Bioremediation is highlighted as a cost-effective and eco-friendly alternative to reverse the damage caused by oil pollution, but more efforts are needed to design enzymes for the process.
JOURNAL OF COMPUTATIONAL CHEMISTRY
(2021)
Article
Computer Science, Artificial Intelligence
Marcio Dorn, Bruno Iochins Grisci, Pedro Henrique Narloch, Bruno Cesar Feltes, Eduardo Avila, Alessandro Kahmann, Clarice Sampaio Alho
Summary: The article highlights the significant impact of the COVID-19 pandemic on health and economies, particularly in countries with limited financial resources such as Brazil. Machine learning techniques have been extensively used in analyzing clinical data including CBC tests. The study reviews various machine learning classifiers and sampling methods for clinical data analysis, particularly focusing on their performance on Brazilian COVID-19 CBC datasets.
PEERJ COMPUTER SCIENCE
(2021)
Article
Computer Science, Interdisciplinary Applications
Itamar Jose Guimaraes Nunes, Bruno Cesar Feltes, Murilo Zanini David, Marcio Dorn
Summary: This study presents a new R package called GEVA, which can analyze multiple gene expression datasets and identify genes with consistent differential expression across experiments. GEVA takes multiple differential expression analysis results as input and performs weighted summarization, partitioning, and clustering to find genes that vary less across experiments. The package also allows grouping of experimental conditions into factors, and performs additional statistical tests to identify genes specifically or dependently expressed in response to these factors. Results from the package were validated using knockout studies and were found to be consistent with published experimental studies. GEVA provides a robust alternative for multiple comparison analyses.
JOURNAL OF BIOMEDICAL INFORMATICS
(2022)
Article
Genetics & Heredity
Gabriela Flores Goncalves, Joice de Faria Poloni, Marcio Dorn
Summary: This study investigated the association between lncRNAs and host response to C. albicans infection. By analyzing a public RNA-Seq dataset, the researchers identified lncRNAs related to biological processes involved in immune response and reactive species production. These findings provide new insights into the functions of lncRNAs in the immune response.
Review
Genetics & Heredity
Regis Antonioli Junior, Joice de Faria Poloni, Ederson Sales Moreira Pinto, Marcio Dorn
Summary: Biosurfactants are amphipathic molecules that can lower interfacial and superficial tensions. They are biodegradable and have a wide diversity of categories, with lipopeptides being the most abundant and well-known. Protein-containing biosurfactants are less studied but could be a valuable alternative. Understanding the harsh conditions that target organisms can sustain is crucial for successful implementation. This article explores the biotechnological applications of lipopeptide and protein-containing biosurfactants, as well as their natural role and the potential research possibilities using meta-omics and machine learning.
Article
Computer Science, Artificial Intelligence
Manuel Villalobos-Cid, Marcio Dorn, Angela Contreras, Mario Inostroza-Ponta
Summary: Phylogenetic networks are used to capture evolutionary phenomena that cannot be described by phylogenetic trees, and different inference criteria may result in conflicting network topologies. Multi-objective optimization can handle conflicting objectives and reduce dependency on specific criteria. In this study, a multi-objective evolutionary algorithm called MO-PhyNet is proposed to infer phylogenetic networks by minimizing three criteria: hardwired parsimony, softwired parsimony, and the number of reticulations. The formalization of phylogenetic inference as a multi-objective optimization problem allows for the identification of different reticulated topologies representing distinct evolutionary hypotheses.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Artur d'Avila Garcez, Luis C. Lamb
Summary: Current advances in AI and Machine Learning have had a significant impact on research communities and industry. However, concerns about trust, safety, interpretability, and accountability have been raised. Neurosymbolic computing combines neural network-based learning with symbolic knowledge representation and reasoning to address these concerns. This paper reviews recent research in neurosymbolic AI, identifies its important components, and proposes promising directions and challenges for the next decade of AI research.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Software Engineering
Adrian Kummerlaender, Marcio Dorn, Martin Frank, Mathias J. Krause
Summary: This article revisits and extends the work on implicit propagation on directly addressed grids by considering them as transformations of the underlying space filling curve. A new periodic shift (PS) pattern is proposed that provides consistent performance across a range of targets. Benchmark results for PS and shift-swap-streaming (SSS) on SIMD CPUs and Nvidia GPUs are provided, and the application of PS as the propagation pattern of the open source LBM framework OpenLB is summarized.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Chemistry, Multidisciplinary
Bruno Cesar Feltes, Ederson Sales Moreira Pinto, Arthur Tonietto Mangini, Marcio Dorn
Summary: The NIAS server is a tool that analyzes the conformational preferences of amino acids in proteins. It is based on the Angle Probability List and integrates structural information from the Protein Data Bank. The updated NIAS server includes data from various experimental techniques and provides examples of its application in structural biology research, as well as discussions on its limitations.
JOURNAL OF COMPUTATIONAL CHEMISTRY
(2023)
Article
Biochemistry & Molecular Biology
Regis Antonioli Junior, Joice de Faria Poloni, Manuel Riveros A. Escalona, Marcio Dorn
Summary: Crude oil contamination poses a significant threat to the environment and biodiversity. However, certain microorganisms in contaminated areas can utilize hydrocarbons as a source of nutrients, highlighting the importance of understanding local community dynamics in these environments. Through the analysis of genetic and functional data, we identified the prevalence of hydrocarbon-degrading functionality in contaminated sediments and potential targets for bioremediation.
Article
Biochemistry & Molecular Biology
Manuel Adrian Riveros Escalona, Joice de Faria Poloni, Mathias J. Krause, Marcio Dorn
Summary: Colorectal cancer is commonly associated with changes in the gut microbiota, and recent studies have provided a better understanding of the specific species related to its development. However, the importance of certain species and their interactions across different datasets still needs further exploration.
Review
Computer Science, Artificial Intelligence
Bruno I. Grisci, Bruno Cesar Feltes, Joice de Faria Poloni, Pedro H. Narloch, Marcio Dorn
Summary: A review of over 1200 publications on feature selection and gene expression between 2010 and 2020 found that 57% of the publications used outdated datasets, 23% used only outdated data, and 32% did not cite data sources. Problems such as referencing unavailable databases, slow adoption of RNA-seq datasets, and bias towards human cancer data were also identified. These issues can result in inaccurate and misleading biological results.
WILEY INTERDISCIPLINARY REVIEWS-DATA MINING AND KNOWLEDGE DISCOVERY
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Mateus Boiani, Rafael Stubs Parpinelli, Marcio Dorn
Summary: Population diversity plays a crucial role in determining the quality of solutions in Evolutionary Algorithms. This paper proposes a new migration policy for the BRKGA and compares its performance with traditional strategies in continuous search spaces. The results show that the proposal can improve the optimization capability of the BRKGA.
INTELLIGENT SYSTEMS, PT I
(2022)
Article
Mathematics, Interdisciplinary Applications
Robin Trunk, Colin Bret, Gudrun Thaeter, Hermann Nirschl, Marcio Dorn, Mathias J. Krause
Summary: A constructive and data-driven approach is used to obtain new drag correlations related to the influence of particle shape on settling behavior. By extending the list of considered shape parameters, models describing the drag coefficient and settling velocity were found.