PART A: CONVOLUTIONAL NEURAL NETWORK
Background
Implement an image classifier using a deep learning network.
Dataset
You are to use the Fashion MNIST dataset.
tf.keras.datasets.fashion_mnist.load_data()
Tasks
1. Write the code to solve the prediction task. You would be using TensorFlow 2.0/Keras, but if you'd prefer to work with some other toolkit such as MXNET or PyTorch, that is fine.
2. Write a Jupyter notebook detailing your implementation, your experiments and analysis. Remember to also save the jupyter notebook as a HTML file after running it.
3. Create a set of slides with the highlights of your Jupyter notebook. Explain the entire deep learning process you went through, data exploration, data cleaning, feature engineering, and model building and evaluation. Write your conclusions.
PART B: CONVOLUTIONAL NEURAL NETWORK 2
Background
Implement an image classifier using a deep learning network. These images are in colour instead of black and white.
Dataset
You are to use the CIFAR10 dataset.
tf.keras.datasets.cifar10.load_data()
Tasks
1. Write the code to solve the prediction task. You would be using TensorFlow 2.0/Keras, but if you'd prefer to work with some other toolkit such as MXNET or PyTorch, that is fine.
2. Write a Jupyter notebook detailing your implementation, your experiments and analysis. Remember to also save the jupyter notebook as a HTML file after running it.
3. Create a set of slides with the highlights of your Jupyter notebook. Explain the entire deep learning process you went through, data exploration, data cleaning, feature engineering, and model building and evaluation. Write your conclusions.
PART C: Technical Paper
This part of the assignment is to be completed individually. This is a challenge task for students who wish to attempt it for higher marks.
Write a technical paper in single column format on any ONE of the following topics.
§ CNN
§ RNN or LSTM
The paper should have the following component:
1. Abstract
2. Introduction
3. Related Works
4. Dataset/Methodology/Experiment
5. Discussion
6. Conclusions
7. References
