Open Access Open Access  Restricted Access Subscription or Fee Access

Multilayer Perceptron Using Levenberg Marquardt Algorithm for Imbalanced Data

S. B. Shinde, S. S. Sayyad, S. U. Pawar

Abstract


Supervised Learning algorithms learn from training dataset. If dataset has unequal distribution among number of samples of classes, classifier gives inaccurate results for minor class and correct results for major class. This paper studies the cost sensitive learning of Multilayer Perceptron (MLP) neural network to overcome this problem. This cost sensitive approach is for binary dataset which modifies the objective function. By this change, network produces accurate results for minor class also. Levenberg-Marquardt (LM) weight updation algorithm is used for network learning. It is fast and efficient algorithm for neural network training rather than other weight updation algorithms. Statistical results performed on real time dataset, shows that how fast the training is with LM algorithm. Also the result shows that MLP classifier produces accurate results for both major and minor dataset.


Keywords


Imbalanced Data, Cost Sensitive Learning, Multilayer Perceptron Neural Network, Levenberg-Marquardt Algorithm.

Full Text:

PDF

References


Cristiano L. Castro,Antonio P. Braga, “Novel Cost-Sensitive Approach to Improve the Multilayer Perceptron Performance on Imbalanced Data” IEEE transaction on neural networks and learning systems, Vol. 24, no. 6, June 2013

Foster Provost,“Machine Learning from Imbalanced Data Sets 101” New York University

Nathalie Japkowicz,Shaju Stephen, “The class imbalance problem - A systematic study”Intelligent Data Analysis Volume 6 Issue 5, October 2002

G. M. Weiss,“Mining with rarity: A unifying framework” ACM SIGKDD ExplorationsNewslett., vol. 6, no. 1, pp. 719, Jun. 2004.

Zhi-Hua Zhou,Xu-Ying Liu, “Training Cost-Sensitive Neural Networks with Methods Addressing the Class Imbalance Problem” IEEE transaction on knowledge and data engineering

Cristiano Leite Castro, Antonio Padua Braga, “ Artificial neural networks learning in ROC Space” in Proc. Int. Joint Conf. Comput., 2009, pp. 484489.

R. Alejo, V.Garcia, J.M.Sotoca, R.A.Mollineda and J.S.Sanchez,“Improving the Performance of the RBF Neural Networks Trained with Imbalanced Samples” in Computational and Ambient Intelligence (Lecture Notes in Computer Science), vol. 4507. New York, USA:Springer-Verlag, 2007, pp. 162169.

S.H. Oh, “Error back-propagation algorithm for classification of imbalanced data”, Neurocomputing, vol. 74, no. 6, pp. 10581061, Feb. 2011

Hao Yu Bogdan M. ,Wilamowski, “Levenberg Marquadt Training”,Auburn University

M. T. Hagan ,M. B. Menhaj, “Training feedforward networks with the Marquardt algorithm,IEEE Trans. Neural Netw., vol. 5, no. 6, pp. 989993, Nov. 1994.

Bogdan M. Wilamowski,Hao Yu, “Improved Computation for Levenberg Marquardt Training”, IEEE transaction on neural networks, Vol. 21, No. 6, June 2010

A. Frank and A. Asuncion, UCI machine learning repository, (2010)

[Online]. Available: http://archive.ics.uci.edu/ml

Shinde, S.B.,Sayyad, S.S.:Cost Sensitive Improved Levenberg Marquardt Algorithm for Imbalanced Data,Computational Intelligence and Computing Research (ICCIC), IEEE International Conference on IEEE, pp. 1-4, (2016).

Shinde, S.B., Sayyad S. S. ,Mulla, A.N.:Multilayer Perceptron Using Modified Levenberg Marquardt Algorithm for Imbalanced Data,Inventi Impact: Soft Computing, vol.4, pp. 209-215, (2015)


Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.