Evaluating the performance of a thresholding filter on handwritten images classification task

Authors

  • Vidal Wyatt M. Lopez National Institute of Physics, University of the Philippines Diliman
  • Francis N. C. Paraan National Institute of Physics, University of the Philippines Diliman

Abstract

In this paper we evaluate the effect of a thresholding filter on the accuracy and training times of a deep neural network. The filter increases the brightness of each pixel in the input image and then applies a threshold condition that zeroes out values exceeding a preset value. Although the filter is lossy, we demonstrate improved learning performance under some use cases.

Downloads

Published

2018-05-29

How to Cite

[1]
VWM Lopez and FNC Paraan, Evaluating the performance of a thresholding filter on handwritten images classification task, Proceedings of the Samahang Pisika ng Pilipinas 36, SPP-2018-PB-47 (2018). URL: https://proceedings.spp-online.org/article/view/SPP-2018-PB-47.

Issue

Section

Poster Session B (Complex Systems, Simulations, and Theoretical Physics)