Learning and Nonlinear Models最新文献_第5页

Aerial Image Instance Segmentation Through Synthetic Data Using Deep Learning 基于深度学习的航空图像实例分割

Learning and Nonlinear Models Pub Date : 2020-09-01 DOI: 10.21528/lnlm-vol18-no1-art3

Felipe X. Viana, Gabriel Araújo, M. Pinto, J. Colares, D. B. Haddad

{"title":"Aerial Image Instance Segmentation Through Synthetic Data Using Deep Learning","authors":"Felipe X. Viana, Gabriel Araújo, M. Pinto, J. Colares, D. B. Haddad","doi":"10.21528/lnlm-vol18-no1-art3","DOIUrl":"https://doi.org/10.21528/lnlm-vol18-no1-art3","url":null,"abstract":"In the last decades, current trends in autonomous navigation have demonstrated an increased use of computational vision over traditional techniques. This relies on the fact that most of the spaces are designed for human navigation. As a result, they are filled with visual cues. In this sense, visual recognition is an essential ability to avoid obstacles when an autonomous vehicle interacts with the real world. Data collection using Unmanned Aerial Vehicles (UAVs) navigating in a real-world scenario is a high-cost and time-expensive activity. For this reason, one of the most valuable assets of technology companies is a database containing locations and interactions. One solution to this problem is the adoption of a photo-realistic 3D simulator as a data source. Using this resource, it is possible to gather a significant amount of data. Therefore, this research creates a dataset for instance segmentation using images from a frontal UAV camera navigating in a 3D simulator. This work applies a state-of-the-art deep learning technique, the Mask-RCNN. The architecture takes an image input and predicts per-pixel instance segmentation. Experimental results showed that Mask RCNN has superior performance in our dataset when refining a model trained using COCO dataset. Besides, the proposed methodology presents a good generalization capability due to the promising results in real-world data.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117005850","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 3

Multi-objective evolutionary algorithms for the truck dispatch problem in open-pit mining operations 露天矿卡车调度问题的多目标进化算法

Learning and Nonlinear Models Pub Date : 2019-11-25 DOI: 10.21528/lmln-vol17-no2-art5

R. F. Alexandre, F. Campelo, J. Vasconcelos

{"title":"Multi-objective evolutionary algorithms for the truck dispatch problem in open-pit mining operations","authors":"R. F. Alexandre, F. Campelo, J. Vasconcelos","doi":"10.21528/lmln-vol17-no2-art5","DOIUrl":"https://doi.org/10.21528/lmln-vol17-no2-art5","url":null,"abstract":"This work is concerned with the efficient allocation of trucks to shovels in operation at open-pit mines. As this problem involves high-value assets, namely mining trucks and shovels, any improvement obtained in terms of operational efficiency can result in considerable financial savings. Thus, this work presents multi-objective strategies for solving the problem of dynamically allocating a heterogeneous fleet of trucks in an open-pit mining operation, aiming at maximizing production and minimizing costs, subject to a set of operational and physical constraints. Two Multi-objective Genetic Algorithms (MOGAs) were specially developed to address this problem: the first uses specialized crossover and mutation operators, while the second employs Path-Relinking as its main variation engine. Four test instances were constructed based on real open-pit mining scenarios, and used to validate the proposed methods. The two MOGAs were compared to each other and against a Greedy Heuristic (GH), suggesting of of the MOGAs as a potential strategy for solving the multi-objective truck dispatch problem for open-pit mining operations.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130829774","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 7

Hybrid differential evolution with the topographical heuristic 地形启发式混合差分进化

Learning and Nonlinear Models Pub Date : 2019-11-25 DOI: 10.21528/lmln-vol17-no2-art4

Ane Élida Nogueira Frauches Almoaia, Wagner F. Sacco, A. Neto

引用次数: 1

Bayesian partition crossover for pseudo-Boolean optimization 伪布尔优化的贝叶斯分割交叉

Learning and Nonlinear Models Pub Date : 2019-11-25 DOI: 10.21528/lmln-vol17-no2-art2

Diogenes Laertius, R. Tinós

{"title":"Bayesian partition crossover for pseudo-Boolean optimization","authors":"Diogenes Laertius, R. Tinós","doi":"10.21528/lmln-vol17-no2-art2","DOIUrl":"https://doi.org/10.21528/lmln-vol17-no2-art2","url":null,"abstract":"– The recombination of solutions is important for population metaheuristics and other optimization algorithms. Re-cently, an efﬁcient recombination operator that preserve the interaction between the decision variables was proposed for pseudo-Boolean optimization. Partition Crossover (PX) groups decision variables in order to allow the decomposition of the evaluation function. PX allows to ﬁnd, with computational cost proportional to the cost of evaluating one solution of the problem, the best solution among a number of offspring solutions that grows exponentially with the number of recombining components found by the operator. PX has been so far used only in problems where the information about the linkage between the decision variables is known a priori . This information is stored in a graph, know as variable interaction graph. We propose a new PX for pseudo-Boolean optimization problems that can be used when the variable interaction graph is not known a priori . For this purpose, it is necessary to estimate the linkage between the decision variables by using procedures generally employed in estimation of distribution algorithms. The experimental results show that the new recombination operator generally improves the number of offspring that are better than their parents when compared to traditional recombination operators. However, generating better offspring does not necessarily imply in better performance for the evolutionary algorithm.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126467562","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A face recognition framework using self-adaptive differential evolution 基于自适应差分进化的人脸识别框架

Learning and Nonlinear Models Pub Date : 2019-11-25 DOI: 10.21528/lmln-vol17-no2-art1

G. Plichoski, Chidambaram Chidambaram, R. S. Parpinelli

{"title":"A face recognition framework using self-adaptive differential evolution","authors":"G. Plichoski, Chidambaram Chidambaram, R. S. Parpinelli","doi":"10.21528/lmln-vol17-no2-art1","DOIUrl":"https://doi.org/10.21528/lmln-vol17-no2-art1","url":null,"abstract":"It is well known that the development of face recognition (FR) systems is challenging under uncontrolled conditions often related to the variation of pose, illumination, expression, and occlusion. Also, to collect the necessary amount of images is hard to guarantee in many situations, e.g. ID cards, drivers licenses or visas, leading to the one sample per person (OSPP) problem. This work addresses the OSPP problem along with illumination and pose variation using an FR framework composed of a self-adaptive Differential Evolution, named FRjDE. The main feature of the framework stands on the use of the optimization algorithm for choosing which preprocessing and feature extraction strategies to use, as well as tunning their parameters. Also, by using the jDE algorithm, F and CR control parameters are self-adapted. Experiments are made using two well-known databases, named CMU-PIE and FERET. Results obtained from the FRjDE approach are compared against the FR framework using the standard DE algorithm and against results found in the literature. Results suggest that the proposed approach is highly competitive and well suited for face recognition.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127700692","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

GA-TCTN: a framework for hyper-parameter optimization and text classification using transductive semi-supervised learning through term networks GA-TCTN:一个通过术语网络使用换向半监督学习的超参数优化和文本分类框架

Learning and Nonlinear Models Pub Date : 2019-11-25 DOI: 10.21528/lmln-vol17-no2-art3

F. P. Coutinho, S. O. Rezende, R. G. Rossi

{"title":"GA-TCTN: a framework for hyper-parameter optimization and text classification using transductive semi-supervised learning through term networks","authors":"F. P. Coutinho, S. O. Rezende, R. G. Rossi","doi":"10.21528/lmln-vol17-no2-art3","DOIUrl":"https://doi.org/10.21528/lmln-vol17-no2-art3","url":null,"abstract":"Transductive Classification through Term Network (TCTN) is an interesting and accurate approach to perform semi-supervised learning based on term networks for text classification. TCTN can surpass the accuracies obtained by transductive classification approach considering texts represented in other types of networks or vector space model. Also, TCTN can surpass the accuracies obtained by inductive supervised learning algorithms. Besides, the term networks in TCTN can have their size decreased while still keeps its classification performance. This implies a less computational cost than other semi-supervised learning approaches based on networks. Originally, TCTN considered just manually defined hyper-parameters. However, even better results can be achieved with a more carefully chosen hyper-parameters values. Thus, in this article, we present a genetic algorithm that (GA) can be used for finding better hyper-parameter values for TCTN. The proposed approach is called GATCTN. Our approach is applied in 25 text collections, and results demonstrate that a GA can be useful together with TCTN for semi-supervised text classification. Besides this contribution, comparisons among hyper-parameters distributions are performed to identify some pattern in its structure. The results indicate that TCTN and GA-TCTN tend to generate a similar set of hyper-parameters. However, GA-TCTN still allows the use of more specific hyper-parameters values being more flexible and practical than TCTN with manually defined parameters. Besides, GA-TCTN obtained better results than TCTN with statistically significant differences.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124686499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Rotulação automática de clusters baseada em análise de filogramas 基于纤维图分析的自动聚类标记

Learning and Nonlinear Models Pub Date : 2019-04-02 DOI: 10.21528/LNLM-VOL17-NO1-ART1

F. N. Araújo, A. H. M. Soares, Vinícius Machado, R. Rabelo

引用次数: 0

Análise comparativa do algoritmo Dijkstra com e sem desfragmentação nas redes ópticas elásticas 弹性光网络中有和没有碎片整理的Dijkstra算法的比较分析

Learning and Nonlinear Models Pub Date : 2019-04-02 DOI: 10.21528/LNLM-VOL17-NO1-ART3

Selles Araujo, Enio L. V. Barbosa, José Valdemir dos Reis Junior, A. Soares

引用次数: 0

A tool for creating parallel swarm algorithms automatically on multi-core computers 一个在多核计算机上自动创建并行群算法的工具

Learning and Nonlinear Models Pub Date : 2019-04-02 DOI: 10.21528/LNLM-VOL17-NO1-ART4

João Pedro Augusto Costa, O. Cortes

{"title":"A tool for creating parallel swarm algorithms automatically on multi-core computers","authors":"João Pedro Augusto Costa, O. Cortes","doi":"10.21528/LNLM-VOL17-NO1-ART4","DOIUrl":"https://doi.org/10.21528/LNLM-VOL17-NO1-ART4","url":null,"abstract":"Meta-heuristics are usually bio-inspired algorithms (based on genes or social behaviors) that are used for solving optimization problems in a variety of fields and applications. The basic principle of a meta-heuristic, such as genetic algorithms, differential evolutions, particle swarm optimization, etc., is to simulate the pressure that the environment applies to individuals resulting in the survival of the best ones. Regardless of which meta-heuristic is being used, the more complex the problem, the more time consuming the algorithm. In this context, parallel computing represents an attractive way of tackling the necessity for computational power. On the other hand, parallel computing introduces new issues that the programmers have to deal with, such as synchronization and the proper exploration of parallel algorithms/models. To avoid these problems, and at the same time, to provide a fast development of parallel swarm algorithms, this work presents a tool for creating parallel code using Parallel Particle Swarm Optimization (PSO) Algorithms in Java. The generator considers three models of parallelism: master-slaves, island and hierarchical. Experiments in the created code showed that a speedup of 5.3 could be reached in the Island model with 2000 iterations using Griewank’s function. Moreover, using a cost estimation model (COCOMO) we showed that our tool could save from 4.4 to 14.5 person/month on programming effort.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126000931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Um modelo de otimização multi-objetivo de demand response para programação de carga residencial 住宅负荷规划的多目标需求响应优化模型

Learning and Nonlinear Models Pub Date : 2019-04-02 DOI: 10.21528/LNLM-VOL17-NO1-ART2

I. R. A. Silva, R. Rabelo, J. M. Veras, Plácido Rogério Pinheiro

{"title":"Um modelo de otimização multi-objetivo de demand response para programação de carga residencial","authors":"I. R. A. Silva, R. Rabelo, J. M. Veras, Plácido Rogério Pinheiro","doi":"10.21528/LNLM-VOL17-NO1-ART2","DOIUrl":"https://doi.org/10.21528/LNLM-VOL17-NO1-ART2","url":null,"abstract":"– This paper presents a multi-objective optimization model for residential demand Response based on the price of real-time electricity (RTP) in order to minimize both the cost associated with electricity consumption and the level of inconvenience (dissatisfaction/discomfort) of the ﬁnal consumers. The proposed model was formalized as a nonlinear programming problem subject to a set of constraints associated with electrical energy consumption and the operational aspects related to the different categories of electrical appliances. The DR problem shown in this work was solved computationally by means of the Non-Dominated Sorted Genetic Algorithm II (NSGA-II) in order to determine the new operation schedule of the residential apparatus for any time horizon. The numerical results show a reduction in the cost associated with the consumption of electricity and also in the level of inconvenience (dissatisfaction/discomfort) of the ﬁnal consumers. In addition, the results achieved with the NSGA-II using the proposed model allow the consumer to make a decision on the reduction of the required cost, in order to seek adequacy to the amount of inconvenience tolerated by the consumer.","PeriodicalId":386768,"journal":{"name":"Learning and Nonlinear Models","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127446282","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0