Experimental Evaluation of Growing and Pruning Hyper Basis Function Neural Networks Trained with Extended Information Filter
Апстракт
In this paper we test Extended Information Filter (EIF) for sequential training of Hyper Basis Function Neural Networks with growing and pruning ability (HBF-GP). The HBF neuron allows different scaling of input dimensions to provide better generalization property when dealing with complex nonlinear problems in engineering practice. The main intuition behind HBF is in generalization of Gaussian type of neuron that applies Mahalanobis-like distance as a distance metrics between input training sample and prototype vector. We exploit concept of neuron’s significance and allow growing and pruning of HBF neurons during sequential learning process. From engineer’s perspective, EIF is attractive for training of neural networks because it allows a designer to have scarce initial knowledge of the system/problem. Extensive experimental study shows that HBF neural network trained with EIF achieves same prediction error and compactness of network topology when compared to EKF, but without the need... to know initial state uncertainty, which is its main advantage over EKF.
Кључне речи:
Extended Information Filter (EIF) / Hyper Basis Function Neural Networks / Growing and pruning ability / Mahalanobis-like distance / Extended Kalman Filter (EKF) / Gaussian type of neuronИзвор:
Proceedings of the 5th International Conference on Information Society and Technology (ICIST 2015), Kopaonik 8-11. March 2015, 2015, 89-94Издавач:
- Society for Information Systems and Computer Networks
Финансирање / пројекти:
- Иновативни приступ у примени интелигентних технолошких система за производњу делова од лима заснован на еколошким принципима (RS-MESTD-Technological Development (TD or TR)-35004)
Колекције
Институција/група
Mašinski fakultetTY - CONF AU - Vuković, Najdan AU - Mitić, Marko AU - Petrović, Milica AU - Petronijević, Jelena AU - Miljković, Zoran PY - 2015 UR - https://machinery.mas.bg.ac.rs/handle/123456789/4468 AB - In this paper we test Extended Information Filter (EIF) for sequential training of Hyper Basis Function Neural Networks with growing and pruning ability (HBF-GP). The HBF neuron allows different scaling of input dimensions to provide better generalization property when dealing with complex nonlinear problems in engineering practice. The main intuition behind HBF is in generalization of Gaussian type of neuron that applies Mahalanobis-like distance as a distance metrics between input training sample and prototype vector. We exploit concept of neuron’s significance and allow growing and pruning of HBF neurons during sequential learning process. From engineer’s perspective, EIF is attractive for training of neural networks because it allows a designer to have scarce initial knowledge of the system/problem. Extensive experimental study shows that HBF neural network trained with EIF achieves same prediction error and compactness of network topology when compared to EKF, but without the need to know initial state uncertainty, which is its main advantage over EKF. PB - Society for Information Systems and Computer Networks C3 - Proceedings of the 5th International Conference on Information Society and Technology (ICIST 2015), Kopaonik 8-11. March 2015 T1 - Experimental Evaluation of Growing and Pruning Hyper Basis Function Neural Networks Trained with Extended Information Filter EP - 94 SP - 89 UR - https://hdl.handle.net/21.15107/rcub_machinery_4468 ER -
@conference{ author = "Vuković, Najdan and Mitić, Marko and Petrović, Milica and Petronijević, Jelena and Miljković, Zoran", year = "2015", abstract = "In this paper we test Extended Information Filter (EIF) for sequential training of Hyper Basis Function Neural Networks with growing and pruning ability (HBF-GP). The HBF neuron allows different scaling of input dimensions to provide better generalization property when dealing with complex nonlinear problems in engineering practice. The main intuition behind HBF is in generalization of Gaussian type of neuron that applies Mahalanobis-like distance as a distance metrics between input training sample and prototype vector. We exploit concept of neuron’s significance and allow growing and pruning of HBF neurons during sequential learning process. From engineer’s perspective, EIF is attractive for training of neural networks because it allows a designer to have scarce initial knowledge of the system/problem. Extensive experimental study shows that HBF neural network trained with EIF achieves same prediction error and compactness of network topology when compared to EKF, but without the need to know initial state uncertainty, which is its main advantage over EKF.", publisher = "Society for Information Systems and Computer Networks", journal = "Proceedings of the 5th International Conference on Information Society and Technology (ICIST 2015), Kopaonik 8-11. March 2015", title = "Experimental Evaluation of Growing and Pruning Hyper Basis Function Neural Networks Trained with Extended Information Filter", pages = "94-89", url = "https://hdl.handle.net/21.15107/rcub_machinery_4468" }
Vuković, N., Mitić, M., Petrović, M., Petronijević, J.,& Miljković, Z.. (2015). Experimental Evaluation of Growing and Pruning Hyper Basis Function Neural Networks Trained with Extended Information Filter. in Proceedings of the 5th International Conference on Information Society and Technology (ICIST 2015), Kopaonik 8-11. March 2015 Society for Information Systems and Computer Networks., 89-94. https://hdl.handle.net/21.15107/rcub_machinery_4468
Vuković N, Mitić M, Petrović M, Petronijević J, Miljković Z. Experimental Evaluation of Growing and Pruning Hyper Basis Function Neural Networks Trained with Extended Information Filter. in Proceedings of the 5th International Conference on Information Society and Technology (ICIST 2015), Kopaonik 8-11. March 2015. 2015;:89-94. https://hdl.handle.net/21.15107/rcub_machinery_4468 .
Vuković, Najdan, Mitić, Marko, Petrović, Milica, Petronijević, Jelena, Miljković, Zoran, "Experimental Evaluation of Growing and Pruning Hyper Basis Function Neural Networks Trained with Extended Information Filter" in Proceedings of the 5th International Conference on Information Society and Technology (ICIST 2015), Kopaonik 8-11. March 2015 (2015):89-94, https://hdl.handle.net/21.15107/rcub_machinery_4468 .