A STUDY ON OUTLIER DETECTION USING HYBRID CONVOLUTION NEURAL NETWORK AND NOVEL BI-DIRECTIONAL GATED RECURRENT UNIT (CNN-BIGRU)

Authors

  • G.Manoharan
  • K. Sivakumar

Keywords:

Outlier Detection, Convolution Neural Network, Bi-directional Gated Recurrent Unit (Bi-GRU), F-score, AUC.

Abstract

In this paper a study onConvolution Neural Network and novel Bi-Gated Recurrent Unit (BiGRU) with extraction stacked auto encoders are discussed, further the data using outliers and inliers have been classified by Probabilistic Neural Networks (PNNs).This kind of system shows better performance compared with the existing technologies. More hidden layers/depth usage in deep learning framework resulted in better performance. However too much of depth may be avoided for better accuracy and hence this current work can be extended by CNNBiGRU and evaluated in terms of the efficiency, F-score, AUC for the outlier detection.

Downloads

Downloads

Published

2020-11-28 — Updated on 2020-11-28

How to Cite

G.Manoharan, & K. Sivakumar. (2020). A STUDY ON OUTLIER DETECTION USING HYBRID CONVOLUTION NEURAL NETWORK AND NOVEL BI-DIRECTIONAL GATED RECURRENT UNIT (CNN-BIGRU). PalArch’s Journal of Archaeology of Egypt / Egyptology, 17(7), 4802-4808. Retrieved from https://archives.palarch.nl/index.php/jae/article/view/2598