added more info about transfer learning experiment

paper/paper.md

@@ -34,6 +34,17 @@ Recent advancements leverage deep learning models to streamline waste sorting an
 
 Integration of machine learning models with mobile devices presents a promising avenue for precise waste management [@narayan_deepwaste:_2021]. The use of optimized deep learning techniques in an app demonstrates potential, achieving an accuracy of 0.881 in waste classification. However, limitations persist, prompting the introduction of Deep Waste, a mobile app employing computer vision to classify waste into ten types. Using transfer learning [@5288526], Deep Waste attains a remarkable 96.41% precision on the test set, functioning both online and offline.
 
+The model was trained with Tesla T4 GPU and uses EfficientNetV2 [@tan2021efficientnetv2] model as a base model with addition of agumentation layer. Adam was used as an optmizer with intital learning rate of 0.01. Which was later optmised using [optuna](https://optuna.org/) to create more accurate optimization parameters. The training and validation loss is shown in \autoref{fig:training_vs_val_loss} whereas \autoref{fig:training_vs_val_accuracy} shows training and validation accuracy on the performed experiment[^2].
+
+![Training and Validation loss at different epochs\label{fig:training_vs_val_loss}](training_vs_val_loss.png){width="48%"}
+
+![Training and Validation accuracy at different epochs\label{fig:training_vs_val_accuracy}](training_vs_val_accuracy.png){width="48%"}
+
+
+The confusion matix of the modle is shown in  \autoref{fig:confusion_matrix}.
+![Confusion Matrix\label{fig:confusion_matrix}](confusion_matrix.png){width="100%"}
+
+[^2]: [https://www.kaggle.com/code/sumn2u/garbage-classification-transfer-learning](https://www.kaggle.com/code/sumn2u/garbage-classification-transfer-learning).
 
 
 # Workflow