G. Notte, P. Chilibroste, M. Pedemonte, Héctor Cancela
{"title":"Evolutionary multi-objective algorithms for feed resource allocation in dairy systems","authors":"G. Notte, P. Chilibroste, M. Pedemonte, Héctor Cancela","doi":"10.1109/LA-CCI48322.2021.9769787","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769787","url":null,"abstract":"In grassland-based dairy systems, determining how to rotate the cows among fields for grazing, how much concentrate to supply and the correct stocking rate to be used are important decisions that impact on the efficiency of the system. Considering the presence of conflictive objectives, a multi-objective approach is therefore the natural way of facing the problem. Due to the computational difficulty of finding the full solution set (the Pareto front) of multi-objective models, it is usually necessary to employ algorithms giving a good approximation of this set. In particular, a number of multi-objective evolutionary algorithms with different characteristics have been proposed in the general optimization literature; but there is no current study of which is the most appropriate one for feed resource allocation in dairy systems. In this work, we present the performance evaluation of four multi-objective evolutionary algorithms to generate an approximation of the Pareto front of the feed resource allocation problem in dairy systems. Two classical genetic algorithms (NSGA-II and SPEA-2) and two differential evolution (DE) algorithms (GDE-3 and a Pareto-based DE) were used. To evaluate the algorithms, two experiments based on scenarios constructed from real data were performed. The comparison took into account running times, objective function values attained, Pareto front comparisons, and approximation quality measures based on four different metrics. From the results we conclude that the SPEA-2 is the algorithm that obtains the best quality performance for the problem under study, but also the slowest one, opening a future work opportunity of improving its computational performance.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115499703","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}
{"title":"Towards best default configuration settings for NMPSO in multi-objective optimization","authors":"Rodrigo Marinao-Rivas, M. Zambrano-Bigiarini","doi":"10.1109/LA-CCI48322.2021.9769844","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769844","url":null,"abstract":"In this work we tested different configuration settings for the NMPSO algorithm, aiming at solving multi-objective optimization problems with a small number of function evaluations, which is an important aspect that must be addressed in real-world optimization problems. Sixteen different configurations were tested for NMPSO, with different combinations of: i) the swarm size, ii) the maximum number of particles in the external archive, and iii) the maximum amount of genetic operations in the external archive. Three DTLZ problems were used to select the best configuration, which was then evaluated against other state-of-the-art multi-objective optimization algorithms (MMOPSO, NSGA-II, NSGA-III). Our results showed that the fastest convergence towards the true Pareto-optimal front is provided by the configuration with a swarm size of 10, a maximum number of particles allowed in the external archive of 100, and a limit of genetic operations per iteration given by 50% of the maximum number of particles allowed in the external archive. The selected configuration was also very competitive or even superior against NSGA-II and NSGA-III, in terms of the number of function evaluations required to start having an HV larger than zero, but also in the HV values achieved after stabilization of the Pareto-optimal front.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117077324","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}
R. Lira, M. Macedo, H. Siqueira, R. Menezes, C. J. A. B. Filho
{"title":"Modelling the Social Interactions in Grey Wolf Optimizer","authors":"R. Lira, M. Macedo, H. Siqueira, R. Menezes, C. J. A. B. Filho","doi":"10.1109/LA-CCI48322.2021.9769781","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769781","url":null,"abstract":"Swarm Intelligence has been successfully used for solving high-dimensional and multimodal optimization problems. However, the wide range of swarm-based techniques, operators, and parameters requires prior knowledge before applying them to real-world problems. Because of this, we have been studying the meso-level characteristics that emerge from the social interactions within the swarm to understand each swarm-based technique’s unique characteristics. In this paper, we model and study the interaction network of the Grey Wolf Optimizer (GWO) to capture its social behaviour. We used Portrait divergence to compare the similarity between network structures over experiments, simulations and iterations of the GWO. We also used Kullback divergence to compare the probability distributions of the network flows varying over experiments, simulations and iterations of the GWO. Furthermore, we discovered we could identify the GWO convergence using the interaction network approach. Comparing different simulations, we found that the wolves communicate using a stable network structure but not necessarily a stable network flow indicating variance in the number of highly influential wolves. We also point out patterns found in GWO that appears to be similar to other swarm-based algorithms (GPSO and FSS).","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130958864","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}
{"title":"Sentiment Analysis on Brazilian Portuguese User Reviews","authors":"F. Souza, Joao Filho","doi":"10.1109/LA-CCI48322.2021.9769838","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769838","url":null,"abstract":"Sentiment Analysis is one of the most classical and primarily studied natural language processing tasks. This problem had a notable advance with the proposition of more complex and scalable machine learning models. Despite this progress, the Brazilian Portuguese language still disposes only of limited linguistic resources, such as datasets dedicated to sentiment classification, especially when considering the existence of predefined partitions in training, testing, and validation sets that would allow a more fair comparison of different algorithm alternatives. Motivated by these issues, this work analyzes the predictive performance of a range of document embedding strategies, assuming the polarity as the system outcome. This analysis includes five sentiment analysis datasets in Brazilian Portuguese, unified in a single dataset, and a reference partitioning in training, testing, and validation sets, both made publicly available through a digital repository. A cross-evaluation of dataset-specific models over different contexts is conducted to evaluate their generalization capabilities and the feasibility of adopting a unique model for addressing all scenarios.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124140461","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}
Eduardo Sperle Honorato, J. B. O. S. Filho, Victor Hugo da Silva Muniz
{"title":"A Hierarchical Ensemble of LSTM-based Autoencoders for Novelty Detection in Passive Sonar Systems","authors":"Eduardo Sperle Honorato, J. B. O. S. Filho, Victor Hugo da Silva Muniz","doi":"10.1109/LA-CCI48322.2021.9769821","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769821","url":null,"abstract":"Sonar operators represent a vital workforce for identifying potential threats to submarines (referred to as contacts) by analysing underwater acoustic signatures acquired by their passive sonar systems. Automatic contact classification models may alleviate the sonar operator task but require additional tools for identifying any class of contact not considered during the system development. This paper proposes a hierarchical detector of unknown contact classes for passive sonar based on modelling signal spectra using Long Short-Term Memory Autoencoders networks. Considering the radiated noise of 28 ships belonging to 8 classes acquired in the Brazilian Navy acoustic range, the system achieved an expressive average value for the area under the detection operation curve (0.946) in a simulated novelty detection scenario involving five known and three unknown classes, surpassing the state-of-the-art.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"164 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127427430","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}
Ricardo Xavier Llugsi Cañar, S. Yacoubi, Allyx Fontaine, P. Lupera
{"title":"A novel Encoder-Decoder structure for Time Series analysis based on Bayesian Uncertainty reduction","authors":"Ricardo Xavier Llugsi Cañar, S. Yacoubi, Allyx Fontaine, P. Lupera","doi":"10.1109/LA-CCI48322.2021.9769850","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769850","url":null,"abstract":"In the present work, a novel Convolutional LSTM Encoder-Decoder structure for the implementation of Weather Forecast for the Andean city of Quito is presented. Aside from the above, the Encoder-Decoder structure uses a Walk-Forward validation, an adjustment of the Bayesian posterior predictive distribution and the ADAMW optimizer to carry out the forecast. The aforementioned stages are combined to obtain 4 error metrics per hour. The prediction is done in base of acquired data from a network of Automatic Weather Stations. The results show that the Convolutional Encoder-Decoder structure with a dropout probability of 0.05 and a model precision equal to 0.1 performs better than a LSTM model, LSTM Stacked model or ARIMA models reaching a maximum error of 1.03 °C. Finally, the methodology could be applied as an effective option to implement the post-processing stage for the physical model of a Weather Forecast System.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126866759","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}
J. P. Vega, E. Sánchez, A. Loukianov, Larbi Djilali
{"title":"Neural Sliding Mode Block Control of Single-Phase Induction Motors","authors":"J. P. Vega, E. Sánchez, A. Loukianov, Larbi Djilali","doi":"10.1109/LA-CCI48322.2021.9769817","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769817","url":null,"abstract":"Single-phase induction motors are widely used for small-scale industries and domestic applications due to their availability, low price, and low maintenance cost. To achieve desired performances of these motors, this paper presents the development of a neural identifier based on a Recurrent High Order Neural Networks (RHONN) on-line trained by an Ex-tended Kalman Filter (EKF) complemented with a discrete-time sliding mode block control strategy to control a Single-Phase Induction Motor (SPIM) mechanical speed and flux. Simulation results illustrate effectiveness of the proposed control scheme to ensure time-varying references adequate tracking. In addition, robustness in presence of parameter variations is achieved.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131945248","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}
Tathiana M. Barchi, L. F. P. Costa, E. Puchta, M. Martins, M. L. Andrade, P. S. D. M. Neto, H. Siqueira
{"title":"A Hybrid Model With Error Correction for Wind Speed Forecasting","authors":"Tathiana M. Barchi, L. F. P. Costa, E. Puchta, M. Martins, M. L. Andrade, P. S. D. M. Neto, H. Siqueira","doi":"10.1109/LA-CCI48322.2021.9769818","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769818","url":null,"abstract":"In recent times wind energy generation has stood out due its integration with traditional electricity grids. Many investigations addressed wind speed forecasting since it presents high volatile and intermittent behavior. Due to this, such a source shows accuracy challenges in relation to its prediction. In this work, a hybrid model based on error correction is proposed, combining the linear Autoregressive and Moving average (ARMA) model and the Multilayer Perceptron (MLP). The approaches was applied in two databases referring to the Brazilian northeast -a prominent region in wind energy. The results reveal that the proposed hybrid model showed good results in comparison to linear and neural-based methods.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134081563","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}
Carlos A. Galindo Meza, Juan A. del Hoyo Ontiveros, P. López-Meyer
{"title":"Classification of Respiration Sounds Using Deep Pre-trained Audio Embeddings","authors":"Carlos A. Galindo Meza, Juan A. del Hoyo Ontiveros, P. López-Meyer","doi":"10.1109/LA-CCI48322.2021.9769831","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769831","url":null,"abstract":"In this work we present the use of an end-to-end deep learning based pre-trained audio embeddings generator, and apply it to the purpose of classification of respiration sounds. With this approach, there is no need to pre-compute spectral representations, e.g. MFCC or filterbanks, since the classification model uses raw audio as the input. Transfer learning was used to train an audio classifier for sounds of respiratory cycles as defined in the ICBHI 2017 challenge. The results on this dataset show that this end-to-end model represents a viable alternative to more common spectral-based classifiers, while achieving state-of-the-art performance.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124741808","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}
{"title":"Using Deep Learning Predictors and Latent Dirichlet Allocation to Identify Key Issues Affecting Clients in Chilean Restaurants","authors":"A. Ferreira, Walter Gómez, Ronald Kliebs","doi":"10.1109/LA-CCI48322.2021.9769840","DOIUrl":"https://doi.org/10.1109/LA-CCI48322.2021.9769840","url":null,"abstract":"In this work, we use a general sentiment analysis methodology to train different Deep Learning predictors so that a proper quantitative valuation can be obtained based on qualitative information given by customers on social media (comments). We use Convolutional Neural Network and Long Short-Term Memory combined with different inputs including the body of the comments and its title. With the trained predictors we classify a large set of comments regarding negative positive customer experiences. Finally we use Latent Dirichlet Allocation algorithm to identify the specific issues appearing on the comments related to negative customer experience ratings.","PeriodicalId":431041,"journal":{"name":"2021 IEEE Latin American Conference on Computational Intelligence (LA-CCI)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115297993","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}