Semi-supervised learning using varitional autoencoders results

less than 1 minute read

Published:

The model trained on data size of 5714 and 36 classes didn’t perform that good. I tried with approaches based on unlabeled data as well as fully labeled data, however, the accuracy still remained low. The graphs are attached as follows :

enter image description here

enter image description here X-axis denotes number of epochs while y-axis denotes accuracy and loss respectively.

The code for the above can be found at : https://www.dropbox.com/s/8aiphylf1kgt1q1/code.py?dl=0

The architecture of the model is broadly :

enter image description here The papers and references that were followed for this :

  • https://arxiv.org/pdf/1806.00081.pdf
  • https://medium.com/@arpanlosalka/resisting-adversarial-attacks-using-gaussian-mixture-variational-autoencoders-be98e69b5070
  • http://bjlkeng.github.io/posts/semi-supervised-learning-with-variational-autoencoders/