Article
Computer Science, Artificial Intelligence
Jeova F. S. Rocha Neto, Pedro Felzenszwalb, Marilyn Vazquez
Summary: This paper proposes a new approach to estimate appearance models directly from images without considering individual pixels. It introduces algebraic expressions that relate local image statistics to spatially coherent regions. Two algorithms, one using a least squares formulation and the other based on eigenvector computation, are presented for estimating appearance models. Experimental results demonstrate the effectiveness of these methods for image segmentation.
SIAM JOURNAL ON IMAGING SCIENCES
(2022)
Article
Mathematics
Jing Su, Hongyu Wang, Bing Yao
Summary: The study introduces two new labeling methods and combines them to form matching-type image labeling with multiple restrictions. The research starts from set-ordered graceful labeling of trees, providing several generation methods and relationships.
Article
Computer Science, Interdisciplinary Applications
Hugo Gangloff, Katherine Morales, Yohan Petetin Samovar
Summary: Hidden Markov models are probabilistic graphical models used for classification tasks in time series applications. This paper explores the extensions of these models, such as pairwise and triplet Markov models, which relax the assumptions and introduce new challenges. The paper proposes solutions to address these challenges, including the use of deep neural networks and continuous latent processes.
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2023)
Article
Geochemistry & Geophysics
Patrick Bogaert, Celine Lamarche, Pierre Defourny
Summary: This article investigates the use of hidden Markov models in the automated processing of classified satellite images for land cover and land-use change. It explores the estimation of emission and transition probabilities to filter out errors and recover the actual sequence of changes. The methodology is illustrated using annual time series of classified images from Brazil, China, and Mali, considering missing observations caused by clouds.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Mathematics
Manuel L. Esquivel, Nadezhda P. Krasii, Gracinda R. Guerreiro
Summary: This study addresses the problem of finding a natural continuous time Markov type process in open populations using information provided by discrete time open Markov chains. Two main approaches are proposed: calibrating a continuous time Markov process using a discrete time transition matrix and directly extending discrete time theory to continuous time theory using semi-Markov processes and open Markov schemes.
Article
Mathematics
Brecht Verbeken, Marie-Anne Guerry
Summary: Discrete time Markov models and semi-Markov models are widely used in manpower planning, but semi-Markov models have more flexible sojourn time distributions to accommodate duration of stay effects. Hybrid semi-Markov models aim to reduce model complexity by only considering duration of stay effects for applicable transitions.
Article
Mathematics
Marta Osca-Guadalajara, Javier Diaz-Carnicero, Silvia Gonzalez-de-Julian, David Vivas-Consuelo
Summary: A cost-utility analysis using Markov chains was conducted to evaluate the cost-effectiveness of four main drugs for osteoporosis treatment in Spain. The study found that teriparatide is the most cost-effective option for treating osteoporosis, especially in patients with fractures from the age of 50.
Article
Computer Science, Artificial Intelligence
Shota Harada, Ryoma Bise, Hideaki Hayashi, Kiyohito Tanaka, Seiichi Uchida
Summary: Utilizing group-based labeling and constrained clustering methods in medical image classification tasks can reduce labeling cost and improve clustering purity. However, challenges arise from inappropriate constraints and extra effort needed. To address these challenges, novel soft-constrained clustering and self-constrained clustering methods were proposed, achieving higher clustering purity in experiments with endoscopic image datasets.
MEDICAL IMAGE ANALYSIS
(2021)
Article
Automation & Control Systems
Guilherme Ost, Daniel Y. Takahashi
Summary: Finite-order Markov models are rarely applied in empirical work when the order is large relative to the sample size due to the exponential growth in the number of parameters and required sample size, as well as the difficulty in interpretation. This paper proposes a subclass of Markov models called Mixture of Transition Distribution models, which can effectively recover the lags and estimate the transition probabilities of high-dimensional MTD models when the set of relevant lags is sparse. The estimated model also allows straightforward interpretation. The key innovation is a recursive procedure for a priori selection of the relevant lags.
JOURNAL OF MACHINE LEARNING RESEARCH
(2023)
Article
Environmental Sciences
Liming Xing, Diogo Bolster, Haifei Liu, Thomas Sherman, David H. Richter, Kyle Rocha-Brownell, Zhiming Ru
Summary: This study investigates the transport of microplastics in open-channel flows by implementing three Markov models. The models are validated using numerical simulations and laboratory experiments, demonstrating their effectiveness and high efficiency. The research provides new insights into preventing and reducing the environmental hazards of microplastics.
WATER RESOURCES RESEARCH
(2022)
Article
Computer Science, Hardware & Architecture
Giulio Masetti, Leonardo Robol, Silvano Chiaradonna, Felicita Di Giandomenico
Summary: A new methodology is proposed for the effective definition and efficient evaluation of dependability-related properties in systems composed of a large number of components. The focus is on component models that can be mapped to stochastic automata, and the new reward structure defined on each component's model is expressed through a newly introduced measure.
IEEE TRANSACTIONS ON RELIABILITY
(2023)
Article
Computer Science, Interdisciplinary Applications
Erhan Karakaya, Alexander Vinel, Alice E. Smith
Summary: This paper proposes a family of Markov models to characterize the distribution of the number of relocations per retrieval in a container depot, focusing on empty container yards. The research expands on existing models by considering different material handling equipment types and allowing for container arrivals during the retrieval process. The results of the models can be used for planning yard layouts, selecting equipment, and determining staffing levels, contributing to both the literature and practice of container depots.
COMPUTERS & INDUSTRIAL ENGINEERING
(2021)
Article
Mathematical & Computational Biology
Rosario Barone, Andrea Tancredi
Summary: Multi-state models are commonly used to represent processes that evolve through a discrete set of states. Two important classes of such models are semi-Markov models, where state transitions may depend on the time spent in the current state, and inhomogeneous Markov models, where transitions are dependent on the time elapsed from the start of the process. Inference for these models becomes computationally challenging when the process is observed only at discrete time points without additional information about state transitions. To address this, a Metropolis-Hastings algorithm is used to reconstruct the unobserved trajectories conditioned on the observed points. The resulting Bayesian inference is illustrated using simulation studies and analysis of benchmark datasets for multi-state models.
STATISTICS IN MEDICINE
(2022)
Article
Economics
Leopoldo Catania, Roberto Di Mari
Summary: A new flexible dynamic model for multivariate nonnegative integer-valued time-series is proposed in the article, utilizing two unobserved integer-valued stochastic variables to control the time and cross-dependence of data. An Expectation-Maximization algorithm is derived for maximum likelihood estimation of model parameters, and a Monte Carlo experiment investigates the finite sample properties of the estimated parameters. The methodology is illustrated through an application with a crime data set, showing superior performance in describing the conditional distribution of crime records.
JOURNAL OF ECONOMETRICS
(2021)
Article
Computer Science, Information Systems
Marcos Lupion, Vicente Gonzalez-Ruiz, Javier Medina-Quero, Juan F. Sanjuan, Pilar M. Ortigosa
Summary: In this study, THPoseLite, a convolutional neural network (CNN) based on MobileNetV2, is proposed to extract poses from thermal images (TIs) by pre-processing and utilizing Blazepose. The integration of THPoseLite into an IoT device with an edge tensor processing unit (TPU) accelerator allows real-time processing of TIs, achieving accurate pose estimation with low energy consumption.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Mathematics, Applied
Mohammad Izadi, Suayip Yuzbasi, Carlo Cattani
Summary: This paper presents a practical and effective scheme to approximate the solutions of the Bagley-Torvik equations using Bessel polynomials as approximation basis functions. The equations are reduced into an algebraic form and a fast algorithm with linear complexity is introduced to compute the fractional derivatives. The proposed approximation algorithm shows high accuracy and efficiency for long-time computations.
RICERCHE DI MATEMATICA
(2023)
Article
Mathematics, Applied
Liu Meng, Meng Kexin, Xing Ruyi, Shuli Mei, Carlo Cattani
Summary: This paper analyzes a nonlinear fractional Black and Scholes model and proposes a novel numerical method for finding the solution. The method combines Haar wavelet integration method, homotopy perturbation method, and variational iteration method to improve efficiency and calculation precision.
MATHEMATICAL METHODS IN THE APPLIED SCIENCES
(2023)
Article
Mathematics, Applied
Julee Shahni, Randhir Singh, Carlo Cattani
Summary: Two efficient numerical algorithms, Bernoulli uniform collocation method and Bernoulli Chebyshev collocation method, are proposed for solving 3rd-order Lane-Emden-Fowler boundary value problems. The singularity at x = 0 is avoided by transforming the problem into its integral form. By using the Bernoulli collocation method, the resulting integral equation is converted into a system of nonlinear equations to be solved numerically. The high accuracy and efficiency of the proposed method are demonstrated by comparing the results with other known techniques.
APPLIED NUMERICAL MATHEMATICS
(2023)
Article
Physics, Multidisciplinary
Wujin Deng, Yan Gao, Jianxue Chen, Aleksey Kudreyko, Carlo Cattani, Enrico Zio, Wanqing Song
Summary: An adaptive remaining useful life prediction model is proposed in this paper for electric vehicle lithium batteries. The capacity degradation of the batteries is modeled using multi-fractal Weibull motion, while the varying degree of long-range dependence and 1/f characteristics in the frequency domain are analyzed. The derived age and state-dependent degradation model includes adaptive drift and diffusion coefficients, which consider the quantitative relations between them. The unit-to-unit variability is considered a random variable, and the convergence of the RUL prediction model is proven for practical application. The model is shown to be effective in a case study.
Article
Automation & Control Systems
Mianzhao Wang, Fan Shi, Xu Cheng, Meng Zhao, Yao Zhang, Chen Jia, Weiwei Tian, Shengyong Chen
Summary: This article introduces the importance of visual object tracking in the field of computer vision, as well as one of its main challenges. To overcome this challenge, the article proposes a new representation for multiview images, called macro-epipolar plane image (macro-EPI), which highlights the spatial topological and angular information of the target and distractors. The macro-EPI is obtained by slicing and restacking the original multiview images, and mapped into 2-D space. The article also presents a modified autoencoder network for training a macro-EPI feature extractor, and a composite framework based on discriminative correlation filters for object tracking.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Automation & Control Systems
Qin Song, Yu-Jun Zheng, Jun Yang, Yu-Jiao Huang, Wei-Guo Sheng, Sheng-Yong Chen
Summary: The COVID-19 pandemic has increased the demand for medical resources. This study proposes a co-evolutionary transfer learning method to predict the demands of medical materials. The method achieves high prediction accuracy compared to other transfer learning and multitask learning models.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Computer Science, Artificial Intelligence
Zan Gao, Xinglei Cui, Tao Zhuo, Zhiyong Cheng, An-An Liu, Meng Wang, Shenyong Chen
Summary: This paper proposes a novel multitemporal-scale spatial-temporal transformer (MSST) network for temporal action localization, which predicts actions on a feature space of multiple temporal scales. The proposed method outperforms state-of-the-art approaches on the THUMOS14 dataset and achieves comparable performance on the ActivityNet1.3 dataset.
IEEE TRANSACTIONS ON HUMAN-MACHINE SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Chao Zhang, Fan Shi, Xinpeng Zhang, Shengyong Chen
Summary: This article introduces a lightweight network called ICFF-YOLOv5 for detecting birds flying at high altitudes. To address the challenges of inconspicuous bird features and unfriendly deep networks for edge devices, the authors designed a feature fusion module and a double combination convolution technique. Experimental results demonstrate the accurate detection of flying birds and high evaluation metric scores.
IEEE SENSORS JOURNAL
(2023)
Article
Engineering, Civil
Jianhua Zhang, Rucen Wang, Ruyu Liu, Dongyan Guo, Bo Li, Shengyong Chen
Summary: IoT-based intelligent transportation, specifically traffic video monitoring, requires accurate vehicle and pedestrian detection. Deep learning methods have high accuracy but are computationally expensive for IoT devices. This study proposes optimization tactics for object detection CNN models on digital signal processors and evaluates the performance. Results show that the proposed method achieves faster speed with minimal accuracy loss compared to running the same model on a desktop CPU.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Jiajun Meng, Zhenhua Wang, Kaining Ying, Jianhua Zhang, Dongyan Guo, Zhen Zhang, Javen Qinfeng Shi, Shengyong Chen
Summary: Compared to human activity classification, there has been less progress on human interaction understanding (HIU). This is mainly due to the challenge of the task and the limitations of shallow graphical representations used in recent approaches. In this paper, a deep consistency-aware framework is proposed to tackle the grouping and labeling inconsistencies in HIU. The framework consists of three components: a backbone CNN for image feature extraction, a factor graph network for learning higher-order consistencies, and a consistency-aware reasoning module for enforcing consistencies. Experimental results show that the proposed approach achieves leading performance on three HIU benchmarks, demonstrating its effectiveness.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Interdisciplinary Applications
Julee Shahni, Randhir Singh, Carlo Cattani
Summary: Two robust algorithms based on Bernstein and shifted Chebyshev polynomials coupled with the collocation technique are proposed for solving three-point Lane-Emden-Fowler boundary value problems (LEFBVPs). The algorithms construct equivalent integral equations of the problems and utilize approximation and collocation approach to generate a system of nonlinear equations, which is then solved by Newton's method. Unlike traditional methods, this approach avoids the need to approximate the derivatives u' and u'', resulting in reduced computational time and truncation error. Numerical results demonstrate the efficiency of the proposed techniques with a few collocation points.
MATHEMATICS AND COMPUTERS IN SIMULATION
(2023)
Article
Engineering, Electrical & Electronic
Xinpeng Zhang, Meng Zhao, Yao Zhang, Ji Ao, Hongxia Yang, Congcong Wang, Shengyong Chen
Summary: This article proposes a hierarchical pyramid network with a T structure to detect microaneurysm (MA) in retinal fundus images. The method overcomes the difficulties caused by limited information and different sizes by using adaptive computation in data preparation and generating multisize datasets for training. Experimental results show that the proposed method achieves state-of-the-art performance.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Mathematics, Applied
Weam G. Alharbi, Abdullah F. Shater, Abdelhalim Ebaid, Carlo Cattani, Mounirah Areshi, Mohammed M. Jalal, Mohammed K. Alharbi
Summary: This study investigates a generalized model of the COVID-19 pandemic by introducing two different definitions in fractional calculus. The solutions of both models are derived and applied to predict the behavior of infected cases in eight European countries. The validity of the results reported in the literature is also discussed.
Article
Mathematics, Applied
Ozge Ozalp Guller, Carlo Cattani, Ecem Acar, Sevilay Kirci Serenbay
Summary: In this study, we propose a new type of nonlinear Bernstein-Chlodowsky operators based on q-integers. We firstly define the nonlinear q-Bernstein-Chlodowsky operators of max-product kind. Then, we provide an error estimation for the q-Bernstein Chlodowsky operators of max-product kind using a suitable generalizition of the Shisha-Mond Theorem. Furthermore, we present upper estimates of the approximation error for some subclasses of functions.
Article
Mathematical & Computational Biology
Manoj K. Singh, Brajesh K. Singh, Poonam, Carlo Cattani
Summary: The effects of the strong Allee effect on the dynamics of the modified Leslie-Gower predator-prey model, in the presence of nonlinear prey-harvesting, have been investigated. The study found that the behaviors of the described mathematical model are positive and bounded for all future times. The conditions for the local stability and existence for various distinct equilibrium points have been determined.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2023)