Research Article
BibTex RIS Cite

Normalizasyona ve Prototip Vektörlerin Başlangıç Değerlerine Göre Öğrenmeli Vektör Kuantalama Metotlarının İncelenmesi

Year 2022, , 8 - 13, 31.12.2022
https://doi.org/10.31590/ejosat.1222296

Abstract

Öğrenmeli Vektör Kuantalama, prototip tabanlı bir yapay sinir ağıdır. Öğrenmeli Vektör Kuantalama ile sınıflandırma, veri seti sınıfları, prototip vektörleri ile temsil edilerek gerçekleştirilir. Bu çalışmada, Öğrenmeli Vektör Kuantalama’nın LVQ1, LVQ2.1, LVQ3, LVQX ve OLVQ1 gibi bazı LVQ varyantları kullanılarak sistemler tasarlanmış, gerçekleştirilmiştir. Oluşturulan sistemler, veri setlerine ve prototip vektörlerinin başlangıç değerlerine göre incelenmiştir. Her veri seti eğitim ve test veri setlerine bölünmüştür. LVQ ağları destekleyici öğrenme stratejisi ile eğitim veri setini kullanarak eğitilir. Sistemlerin başarısını test etmek için her ağ için modeller oluşturulmuştur. Ayrıca sistemler, z-skoru ve doğrusal ölçekleme gibi bazı belirgin normalizasyon teknikleri kullanılarak birbirleriyle karşılaştırılır. Başlangıç değeri atamalarında, tüm prototip vektörleri için rastgele değerler seçilebilir ve tüm prototip vektörlerinin değerleri sıfıra atanabilir. Geliştirilen sistemler, doğruluk ve f-ölçüsü metrikleri ile değerlendirilmiştir ve başarı oranları ile karşılaştırılmıştır.

References

  • Günel, K., Aşlıyan, R. and İclal, G. (2016). A Geometrical Modification of Learning Vector Quantization Method for Solving Classification Problems. Suleyman Demirel University Journal of Natural and Applied Sciences, vol. 20(3), pp. 414-420.
  • Hammer, B. and Villmann, T. (2002). Generalized relevance learning vector quantization. Neural Networks, vol. 15(8-9), pp. 1059-1068.
  • Iris data set. (2022). Website [Online]. Available: https://archive.ics.uci.edu/ml/datasets/iris
  • Katagiri, S. and Lee, C.H. (1993). A new hybrid algorithm for speech recognition based on HMM segmentation and learning vector quantization. IEEE Transactions on Speech and Audio Processing, vol. 1(4), pp. 421-430.
  • Kohonen, T. (1986). Learning vector quantization for pattern recognition. Report TKK-F-A601, Helsinki University of Technology, Espoo, Finland.
  • Kohonen, T., Barna, G. and Chrisley, R. (1988). Statistical pattern recognition with neural networks: Benchmarking studies. In Proc. of the International Conference on Neural Networks (ICNN), vol. I, Los Alamitos, CA. IEEE Computer Soc. Press, p. 61-68.
  • Kohonen, T. (1990). Improved versions of learning vector quantization. In Pro. of the International Joint Conference on Neural Networks (IJCNN), vol. 1, pages 545-550, San Diego, California.
  • Kohonen, T. (1992). New developments of learning vector quantization and self-organizing map. In Proc. Symposium on Neural Networks, Alliances and Perspectives in Senri, Osaka, Japan.
  • Kohonen, T. (1995). Self-Organizing Maps. Springer, Berlin, Germany.
  • Kohonen, T., Hynninen, J., Kangas, J., Laaksonen, J. and Torkkola, K. (1996). LVQ_PAK: the learning vector quantization programming package. Report A30, Helsinki University of Technology, Laboratory of Computer and Information Science, Espoo, Finland.
  • Makino, S., Endo, M., Sone, T. and Kido, K. (1992). Recognition of phonemes in continuous speech using a modified LVQ2 method. J. Acoustical Society of Japan, vol. 13(6) pp. 351-360.
  • McDermott, E. (1990). LVQ3 for phoneme recognition. In Proc. Spring Meet. Acoust. Soc. Jpn., p. 151-152.
  • Öztemel, E. (2012). Yapay Sinir Ağları, Ezgi Kitapevi, Bursa.
  • Pham, D.T. and Oztemel, E. (1993). Control Chart Pattern Recognition Using Combinations of Multilayer Perceptrons and Learning Vector Quantization Neural Networks. Proc. Instn. Mech. Engrs. Vol. 207, pp. 113-118.
  • Pham, D.T. and Oztemel, E. (1994). Control Chart Pattern Recognition Using Combinations of Multilayer Perceptrons and Learning Vector Quantization Neural Networks. International Journal of Production Research, vol. 32, 721-729.
  • Sato, A. and Yamada, K. (1995). Generalized Learning Vector Quantization”, NIPS.
  • Wine data set. (2022). Website [Online]. Available: https://archive.ics.uci.edu/ml/datasets/wine

Examining Variants of Learning Vector Quantizations According to Normalization and Initialization of Vector Positions

Year 2022, , 8 - 13, 31.12.2022
https://doi.org/10.31590/ejosat.1222296

Abstract

Learning Vector Quantization is a prototype-based artificial neural network. The classification is performed by representing the data set with the prototype vectors of the classes. In this study, using some variants of Learning Vector Quantization such as LVQ1, LVQ2.1, LVQ3, LVQX, and OLVQ1, the systems are designed and implemented, and they are examined according to initializations of prototype vectors and data sets. Every data set is divided into training and testing data sets. With the training data set, all LVQ networks are trained in a reinforcement learning strategy, and the models for each network are generated to test the success of the systems. In addition, the systems are compared with each other using some distinct normalization techniques such as z-score and linear scaling. In initial conditions, all prototype vectors can be randomly selected, and the values of all prototype vectors can be assigned to zero. The generated systems are evaluated by accuracy and f-measure benchmark measures and compared by their success rates.

References

  • Günel, K., Aşlıyan, R. and İclal, G. (2016). A Geometrical Modification of Learning Vector Quantization Method for Solving Classification Problems. Suleyman Demirel University Journal of Natural and Applied Sciences, vol. 20(3), pp. 414-420.
  • Hammer, B. and Villmann, T. (2002). Generalized relevance learning vector quantization. Neural Networks, vol. 15(8-9), pp. 1059-1068.
  • Iris data set. (2022). Website [Online]. Available: https://archive.ics.uci.edu/ml/datasets/iris
  • Katagiri, S. and Lee, C.H. (1993). A new hybrid algorithm for speech recognition based on HMM segmentation and learning vector quantization. IEEE Transactions on Speech and Audio Processing, vol. 1(4), pp. 421-430.
  • Kohonen, T. (1986). Learning vector quantization for pattern recognition. Report TKK-F-A601, Helsinki University of Technology, Espoo, Finland.
  • Kohonen, T., Barna, G. and Chrisley, R. (1988). Statistical pattern recognition with neural networks: Benchmarking studies. In Proc. of the International Conference on Neural Networks (ICNN), vol. I, Los Alamitos, CA. IEEE Computer Soc. Press, p. 61-68.
  • Kohonen, T. (1990). Improved versions of learning vector quantization. In Pro. of the International Joint Conference on Neural Networks (IJCNN), vol. 1, pages 545-550, San Diego, California.
  • Kohonen, T. (1992). New developments of learning vector quantization and self-organizing map. In Proc. Symposium on Neural Networks, Alliances and Perspectives in Senri, Osaka, Japan.
  • Kohonen, T. (1995). Self-Organizing Maps. Springer, Berlin, Germany.
  • Kohonen, T., Hynninen, J., Kangas, J., Laaksonen, J. and Torkkola, K. (1996). LVQ_PAK: the learning vector quantization programming package. Report A30, Helsinki University of Technology, Laboratory of Computer and Information Science, Espoo, Finland.
  • Makino, S., Endo, M., Sone, T. and Kido, K. (1992). Recognition of phonemes in continuous speech using a modified LVQ2 method. J. Acoustical Society of Japan, vol. 13(6) pp. 351-360.
  • McDermott, E. (1990). LVQ3 for phoneme recognition. In Proc. Spring Meet. Acoust. Soc. Jpn., p. 151-152.
  • Öztemel, E. (2012). Yapay Sinir Ağları, Ezgi Kitapevi, Bursa.
  • Pham, D.T. and Oztemel, E. (1993). Control Chart Pattern Recognition Using Combinations of Multilayer Perceptrons and Learning Vector Quantization Neural Networks. Proc. Instn. Mech. Engrs. Vol. 207, pp. 113-118.
  • Pham, D.T. and Oztemel, E. (1994). Control Chart Pattern Recognition Using Combinations of Multilayer Perceptrons and Learning Vector Quantization Neural Networks. International Journal of Production Research, vol. 32, 721-729.
  • Sato, A. and Yamada, K. (1995). Generalized Learning Vector Quantization”, NIPS.
  • Wine data set. (2022). Website [Online]. Available: https://archive.ics.uci.edu/ml/datasets/wine
There are 17 citations in total.

Details

Primary Language English
Subjects Engineering
Journal Section Articles
Authors

Rıfat Aşlıyan 0000-0003-1495-713X

Publication Date December 31, 2022
Published in Issue Year 2022

Cite

APA Aşlıyan, R. (2022). Examining Variants of Learning Vector Quantizations According to Normalization and Initialization of Vector Positions. Avrupa Bilim Ve Teknoloji Dergisi(45), 8-13. https://doi.org/10.31590/ejosat.1222296