Add assignment_1.ipynb file #40
Conversation
|
Hello, thank you for your contribution. If you are a participant, please close this pull request and open it in your own forked repository instead of here. Please read the instructions on your onboarding Assignment Submission Guide more carefully. If you are not a participant, please give us up to 72 hours to review your PR. Alternatively, you can reach out to us directly to expedite the review process. |
|
Hello, thank you for your contribution. If you are a participant, please close this pull request and open it in your own forked repository instead of here. Please read the instructions on your onboarding Assignment Submission Guide more carefully. If you are not a participant, please give us up to 72 hours to review your PR. Alternatively, you can reach out to us directly to expedite the review process. |
What changes are you trying to make?
I have implemented a Convolutional Neural Network (CNN) model to classify the CIFAR-100 dataset. The model was designed with multiple convolutional layers, max-pooling layers, and dropout regularization to prevent overfitting. I also performed the following steps:
Normalized the images.
One-hot encoded the labels.
Split the training set into training and validation sets.
Built the CNN architecture, compiled the model, and trained it on the CIFAR-100 dataset.
Evaluated the model’s performance and visualized the accuracy over epochs.
Experimented with an enhanced model architecture that includes additional dropout layers.
What did you learn from the changes you have made?
I have learned how to build, train, and evaluate a CNN for a multi-class classification problem. I also gained experience in using regularization techniques like dropout to improve the generalization of the model. Additionally, I observed the model's performance on the CIFAR-100 dataset and the effect of dropout on model accuracy and overfitting.
Was there another approach you were thinking about making? If so, what approach(es) were you thinking of?
Another approach I was considering was experimenting with different types of regularization, such as L2 regularization, to compare its performance with dropout. I also thought about testing other architectures, such as deeper networks or using pre-trained models like ResNet or VGG for transfer learning, which could potentially improve accuracy further.