Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367723
T. Djukić, D. Cvetković, Milos D. Radovic, M. Zivanovic, N. Filipovic
One of the approaches that could be used for cancer treatment is electroporation. This is a relatively new technique and thus its effect on various cancer cell types should be analyzed in detail. In this paper numerical simulations are used, in order to model the behavior of cells after electroporation. Fitting procedure was used for estimation of the parameters of the computer model. This model enables continuous tracking of changes in cell viability and provides some quantitative information about the effect of electric field (change in proliferation and death rate, oxygen consumption etc.). The accuracy of the model is validated using experimental data.
{"title":"Numerical modeling of behavior of cancer cells after electroporation","authors":"T. Djukić, D. Cvetković, Milos D. Radovic, M. Zivanovic, N. Filipovic","doi":"10.1109/BIBE.2015.7367723","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367723","url":null,"abstract":"One of the approaches that could be used for cancer treatment is electroporation. This is a relatively new technique and thus its effect on various cancer cell types should be analyzed in detail. In this paper numerical simulations are used, in order to model the behavior of cells after electroporation. Fitting procedure was used for estimation of the parameters of the computer model. This model enables continuous tracking of changes in cell viability and provides some quantitative information about the effect of electric field (change in proliferation and death rate, oxygen consumption etc.). The accuracy of the model is validated using experimental data.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"48 24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124218584","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367637
Ioannis N. Kouris, C. Tsirbas, T. Tagaris, E. Vellidou, P. Vartholomeos, Stamatia Rizou, D. Koutsouris
The independence and the maintenance of the autonomy of the elderly promotes the self-care and the quality of living of the long-lived population. With the aid of Information and Communication Technologies (ICT), KINOPTIM project develops a tele-monitoring solution to be used by the elderly in their home environment in order to reduce the risk of falls by remotely monitoring the mobility of the persons through a series of interactive virtual reality games defined by clinicians. Data acquired by optical and motion sensors are fused to evaluate the mobility status of the senior and are processed by the Medical Business Intelligence (MBI) module of KINOPTIM platform, to assist the identification of symptomatic functional features, that decrease person's mobility status over time, and help to decide for proactive or active interventions. This paper focuses on the progress of the development of MBI module and the integration with the KINOPTIM platform.
{"title":"KINOPTIM: The medical business intelligence module for fall prevention of the elderly","authors":"Ioannis N. Kouris, C. Tsirbas, T. Tagaris, E. Vellidou, P. Vartholomeos, Stamatia Rizou, D. Koutsouris","doi":"10.1109/BIBE.2015.7367637","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367637","url":null,"abstract":"The independence and the maintenance of the autonomy of the elderly promotes the self-care and the quality of living of the long-lived population. With the aid of Information and Communication Technologies (ICT), KINOPTIM project develops a tele-monitoring solution to be used by the elderly in their home environment in order to reduce the risk of falls by remotely monitoring the mobility of the persons through a series of interactive virtual reality games defined by clinicians. Data acquired by optical and motion sensors are fused to evaluate the mobility status of the senior and are processed by the Medical Business Intelligence (MBI) module of KINOPTIM platform, to assist the identification of symptomatic functional features, that decrease person's mobility status over time, and help to decide for proactive or active interventions. This paper focuses on the progress of the development of MBI module and the integration with the KINOPTIM platform.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124592267","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367705
N. Bayramoglu, Juho Kannala, J. Heikkilä
Automated cell classification in Indirect Immunofluorescence (IIF) images has potential to be an important tool in clinical practice and research. This paper presents a framework for classification of Human Epithelial Type 2 cell IIF images using convolutional neural networks (CNNs). Previuos state-of-the-art methods show classification accuracy of 75.6% on a benchmark dataset. We conduct an exploration of different strategies for enhancing, augmenting and processing training data in a CNN framework for image classification. Our proposed strategy for training data and pre-training and fine-tuning the CNN network led to a significant increase in the performance over other approaches that have been used until now. Specifically, our method achieves a 80.25% classification accuracy. Source code and models to reproduce the experiments in the paper is made publicly available.
{"title":"Human Epithelial Type 2 cell classification with convolutional neural networks","authors":"N. Bayramoglu, Juho Kannala, J. Heikkilä","doi":"10.1109/BIBE.2015.7367705","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367705","url":null,"abstract":"Automated cell classification in Indirect Immunofluorescence (IIF) images has potential to be an important tool in clinical practice and research. This paper presents a framework for classification of Human Epithelial Type 2 cell IIF images using convolutional neural networks (CNNs). Previuos state-of-the-art methods show classification accuracy of 75.6% on a benchmark dataset. We conduct an exploration of different strategies for enhancing, augmenting and processing training data in a CNN framework for image classification. Our proposed strategy for training data and pre-training and fine-tuning the CNN network led to a significant increase in the performance over other approaches that have been used until now. Specifically, our method achieves a 80.25% classification accuracy. Source code and models to reproduce the experiments in the paper is made publicly available.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130044379","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367736
I. Mporas, Anastasia Efstathiou, V. Megalooikonomou
In this article we present a fusion architecture for the automatic classification of sleep stages. The architecture relies on time and frequency domain features which are processed by dissimilar classifiers. The initial predictions of each classifier are refined by using fusion of the prediction estimations together with temporal contextual information of the electroencephalographic signal. The experimental results showed that the proposed architecture achieved approximately 95% sleep stage classification accuracy, which corresponds to an improvement of 5% comparing to the best performing single classifier.
{"title":"Improving sleep stage classification from electroencephalographic signals by fusion of contextual information","authors":"I. Mporas, Anastasia Efstathiou, V. Megalooikonomou","doi":"10.1109/BIBE.2015.7367736","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367736","url":null,"abstract":"In this article we present a fusion architecture for the automatic classification of sleep stages. The architecture relies on time and frequency domain features which are processed by dissimilar classifiers. The initial predictions of each classifier are refined by using fusion of the prediction estimations together with temporal contextual information of the electroencephalographic signal. The experimental results showed that the proposed architecture achieved approximately 95% sleep stage classification accuracy, which corresponds to an improvement of 5% comparing to the best performing single classifier.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130135786","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367691
J. Wulffen, Patrick C. F. Buchholz, O. Sawodny, R. Feuer
In bioindustrial large scale fermenters microorganisms are exposed to conditions of unsteady nutrient supply which occur only rarely on small lab-scale fermenters and lead to economic losses. In aerobic processes cells face different availabilities of oxygen, nitrogen and carbon sources along various directions inside a fermenter. The adaptation of the central metabolism in the facultative anaerobic bacterium Escherichia coli to changing oxygen concentrations will be investigated. Flux balance analysis (FBA) is an often used method to calculate reaction fluxes under given environmental conditions. FBA is based on a stoichiometric model with possible reaction fluxes which are limited by constraints. One sort of constraints are the lower and upper flux bounds. Existing methods of FBA describe metabolic adaptations to changing environments not in sufficient detail. This work develops a variant of FBA in order to close this gap. Balance equations for important gene transcripts and gene products are formulated and flux bounds are calculated continuously. The described variant of FBA is applied to a model of E. coli central metabolism. A transition between anaerobic and aerobic environment is simulated. The results are compared with a conventional FBA approach and regulatory FBA (rFBA). The FBA method described in this study shows possible targets for experimental validation.
{"title":"Modeling the metabolism of escherichia coli under oxygen gradients with dynamically changing flux bounds","authors":"J. Wulffen, Patrick C. F. Buchholz, O. Sawodny, R. Feuer","doi":"10.1109/BIBE.2015.7367691","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367691","url":null,"abstract":"In bioindustrial large scale fermenters microorganisms are exposed to conditions of unsteady nutrient supply which occur only rarely on small lab-scale fermenters and lead to economic losses. In aerobic processes cells face different availabilities of oxygen, nitrogen and carbon sources along various directions inside a fermenter. The adaptation of the central metabolism in the facultative anaerobic bacterium Escherichia coli to changing oxygen concentrations will be investigated. Flux balance analysis (FBA) is an often used method to calculate reaction fluxes under given environmental conditions. FBA is based on a stoichiometric model with possible reaction fluxes which are limited by constraints. One sort of constraints are the lower and upper flux bounds. Existing methods of FBA describe metabolic adaptations to changing environments not in sufficient detail. This work develops a variant of FBA in order to close this gap. Balance equations for important gene transcripts and gene products are formulated and flux bounds are calculated continuously. The described variant of FBA is applied to a model of E. coli central metabolism. A transition between anaerobic and aerobic environment is simulated. The results are compared with a conventional FBA approach and regulatory FBA (rFBA). The FBA method described in this study shows possible targets for experimental validation.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"77 2","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114096860","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367709
Evangelia Pippa, I. Mporas, V. Megalooikonomou
In this paper, we propose a computationally efficient method to estimate the optimal order of the autoregressive (AR) modeling of electroencephalographic (EEG) signals in order to use the AR coefficients as features for the analysis of EEG signals and the automatic detection of epileptic seizures. The estimation of the optimal AR-order is made using regression analysis of statistical features extracted from the samples of the EEG signals. The proposed method was evaluated in both background and ictal EEG segments using recordings from 10 epileptic patients. The experimental evaluation showed that the mean absolute error of the estimated optimal AR order is approximately 4 units.
{"title":"Automatic estimation of the optimal AR order for epilepsy analysis using EEG signals","authors":"Evangelia Pippa, I. Mporas, V. Megalooikonomou","doi":"10.1109/BIBE.2015.7367709","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367709","url":null,"abstract":"In this paper, we propose a computationally efficient method to estimate the optimal order of the autoregressive (AR) modeling of electroencephalographic (EEG) signals in order to use the AR coefficients as features for the analysis of EEG signals and the automatic detection of epileptic seizures. The estimation of the optimal AR-order is made using regression analysis of statistical features extracted from the samples of the EEG signals. The proposed method was evaluated in both background and ictal EEG segments using recordings from 10 epileptic patients. The experimental evaluation showed that the mean absolute error of the estimated optimal AR order is approximately 4 units.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115971098","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367713
M. Blagojevic, Milos D. Radovic, M. Radovic, N. Filipovic
This paper describes the use of artificial neural networks in predicting value and position maximal wall shear stress in aneurysm. For the purpose of neural network training, back propagation algorithm was used. Input data in the network are geometric parameters of aneurysm model. Obtained results indicate the possibility of a successful application of neural networks in the problems of predicting certain parameters of arteries. Future work relates to the creation of a web-based application that allows users to display the results.
{"title":"Neural network based approach for predicting maximal wall shear stress in the artery","authors":"M. Blagojevic, Milos D. Radovic, M. Radovic, N. Filipovic","doi":"10.1109/BIBE.2015.7367713","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367713","url":null,"abstract":"This paper describes the use of artificial neural networks in predicting value and position maximal wall shear stress in aneurysm. For the purpose of neural network training, back propagation algorithm was used. Input data in the network are geometric parameters of aneurysm model. Obtained results indicate the possibility of a successful application of neural networks in the problems of predicting certain parameters of arteries. Future work relates to the creation of a web-based application that allows users to display the results.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131487890","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367652
M. Topalovic, M. Blagojevic, A. Nikolic, M. Zivkovic, N. Filipovic
Smoothed Particle Hydrodynamics is meshless numerical method which is based on continuum mechanics approach, capable of analyzing stresses in both solids and fluids as well as stresses that are result of solid-fluid interaction, with very versatile applications, and yet it' is not sufficiently implemented in biomechanics due to difficulties of node grid generation from complex shape objects. This paper presents multiblock procedure for generation of pseudo-particles which are used in Smoothed Particle Hydrodynamics for representation of discretized parts of analyzed continua. This procedure enables creation of evenly sized pseudo-particles even for the very irregular shaped object such are organs, bones or blood vessels which are analyzed in biomechanics.
{"title":"Application of smoothed particle hydrodynamics in biomechanics: Advanced procedure for discretization of complex biological shapes into pseudo-particles","authors":"M. Topalovic, M. Blagojevic, A. Nikolic, M. Zivkovic, N. Filipovic","doi":"10.1109/BIBE.2015.7367652","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367652","url":null,"abstract":"Smoothed Particle Hydrodynamics is meshless numerical method which is based on continuum mechanics approach, capable of analyzing stresses in both solids and fluids as well as stresses that are result of solid-fluid interaction, with very versatile applications, and yet it' is not sufficiently implemented in biomechanics due to difficulties of node grid generation from complex shape objects. This paper presents multiblock procedure for generation of pseudo-particles which are used in Smoothed Particle Hydrodynamics for representation of discretized parts of analyzed continua. This procedure enables creation of evenly sized pseudo-particles even for the very irregular shaped object such are organs, bones or blood vessels which are analyzed in biomechanics.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"111 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131881048","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367686
M. A. Abdalla, H. Seker, Richard Jiang
Eimeria is an internal animal parasite that causes serious diseases and animal death, and reduces animal productivities. Eimeria has more than one species for every single genus of animals. An early diagnosis of Eimeria infection is usually achieved by examining fecal microscopy images. As Eimeria oocysts vary in terms of shapes, sizes and textures, they can be detected by measuring differences in their shapes, sizes and textural features. As these differences can be driven by analyzing pixel information in microscopic images, this paper therefore presents pixel-based features rather than using the oocysts morphological characteristics. This approach is then applied for the diagnosis of seven different species of Eimeria in chickens as a case study. The pixel-based features are the mean of pixel values over columns and rows of oocyst image matrices in grey-scaled images. The features have been extracted after detecting the oocyst edges by using Moore-Neighbor Tracing Algorithm. For the classification phase, K-Nearest Neighbor classifier was utilized. For its statistical validation, a 5-fold cross validation was adapted and run for 100 times. This proposed approach has yielded an average accuracy of 82% ± 0.54% This is a promising result that is potentially expected to lead fully automated portable parasite detection system.
{"title":"Automated identification of chicken eimeria species from microscopic images","authors":"M. A. Abdalla, H. Seker, Richard Jiang","doi":"10.1109/BIBE.2015.7367686","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367686","url":null,"abstract":"Eimeria is an internal animal parasite that causes serious diseases and animal death, and reduces animal productivities. Eimeria has more than one species for every single genus of animals. An early diagnosis of Eimeria infection is usually achieved by examining fecal microscopy images. As Eimeria oocysts vary in terms of shapes, sizes and textures, they can be detected by measuring differences in their shapes, sizes and textural features. As these differences can be driven by analyzing pixel information in microscopic images, this paper therefore presents pixel-based features rather than using the oocysts morphological characteristics. This approach is then applied for the diagnosis of seven different species of Eimeria in chickens as a case study. The pixel-based features are the mean of pixel values over columns and rows of oocyst image matrices in grey-scaled images. The features have been extracted after detecting the oocyst edges by using Moore-Neighbor Tracing Algorithm. For the classification phase, K-Nearest Neighbor classifier was utilized. For its statistical validation, a 5-fold cross validation was adapted and run for 100 times. This proposed approach has yielded an average accuracy of 82% ± 0.54% This is a promising result that is potentially expected to lead fully automated portable parasite detection system.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133882719","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}
Pub Date : 2015-11-02DOI: 10.1109/BIBE.2015.7367651
J. C. Nievola, E. Paraiso, A. Freitas
This paper introduces the use of a modified feedforward neural network to cope with the problem of predicting protein functions. Since this kind of classification task is inherently hierarchical, this work proposes the use of two different architectures for the modified feedforward neural network, both mimicking the hierarchical nature of the classes (protein functions) to be predicted. The first approach consists of four feed-forward neural networks in cascade, each one taking as input the classification obtained by the previous network, which means, the input to a network is the classes that could be assigned to the protein at the immediately higher (parent) level in the class hierarchy. The second approach is an extension of the first one, which also adds as input to each sub-network the attributes of the protein being classified. In both situations, it was used two kinds of feed-forward architectures: an Adaline network, which is composed of a single layer of adjustable weights, and a MLP ("Multi-Layer Perceptron"), composed by two layers of adjustable weights. Both approaches were compared with a baseline consisting of a single MLP that maps the input attributes to the classes of the lowest level in the hierarchy. The MLP was built with the input layer, plus one hidden layer and one output layer. The three approaches were compared on eight datasets, the first four involving the prediction of GPCR (G-Protein Coupled Receptor) functions and the second four datasets involving the prediction of enzymes functions. The results show that a big-bang hierarchical neural network, based on the MLP paradigm, using a top-down evaluation for new instances has better behavior in hierarchical problems, when compared to its flat version.
{"title":"A hierarchical neural network for predicting protein functions","authors":"J. C. Nievola, E. Paraiso, A. Freitas","doi":"10.1109/BIBE.2015.7367651","DOIUrl":"https://doi.org/10.1109/BIBE.2015.7367651","url":null,"abstract":"This paper introduces the use of a modified feedforward neural network to cope with the problem of predicting protein functions. Since this kind of classification task is inherently hierarchical, this work proposes the use of two different architectures for the modified feedforward neural network, both mimicking the hierarchical nature of the classes (protein functions) to be predicted. The first approach consists of four feed-forward neural networks in cascade, each one taking as input the classification obtained by the previous network, which means, the input to a network is the classes that could be assigned to the protein at the immediately higher (parent) level in the class hierarchy. The second approach is an extension of the first one, which also adds as input to each sub-network the attributes of the protein being classified. In both situations, it was used two kinds of feed-forward architectures: an Adaline network, which is composed of a single layer of adjustable weights, and a MLP (\"Multi-Layer Perceptron\"), composed by two layers of adjustable weights. Both approaches were compared with a baseline consisting of a single MLP that maps the input attributes to the classes of the lowest level in the hierarchy. The MLP was built with the input layer, plus one hidden layer and one output layer. The three approaches were compared on eight datasets, the first four involving the prediction of GPCR (G-Protein Coupled Receptor) functions and the second four datasets involving the prediction of enzymes functions. The results show that a big-bang hierarchical neural network, based on the MLP paradigm, using a top-down evaluation for new instances has better behavior in hierarchical problems, when compared to its flat version.","PeriodicalId":422807,"journal":{"name":"2015 IEEE 15th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123355720","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：