WearNet Model Fashion MNIST Classifier |
Author(s): |
| Bhoomika Madhukar Keni , BNMIT; Ananya HC, BNMIT |
Keywords: |
| Deep Neural Networks, Fashion MNIST, Image Classification, Optimizers, Regularization, Transfer Learning, Gradient Descent |
Abstract |
|
Deep neural networks (DNNs) are powerful artificial intelligence devices that replicate the hierarchical learning structure of the human brain. This work explores the usage of DNNs to classify grayscale garments images by applying the Fashion MNIST data set. This research is interested in the training stages of networks, vanishing/ exploding gradients, transfer learning, and optimizers. Implementation is compared among different optimizer like Adam, RMSprop, Adagrad, and SGD with and without momentum, and regularization techniques like L1, L2, and L1_L2 to prevent over fitting. The Adam optimizer performs optimum with a test accuracy of around 93.39%. The project shows the practical efficiency of DNNs for image classification. provide more detailed information. |
Other Details |
|
Paper ID: IJSRDV13I70036 Published in: Volume : 13, Issue : 7 Publication Date: 01/10/2025 Page(s): 35-40 |
Article Preview |
|
|
|
|
