Your assessment for neural network will take a different form compared to prior IPs. Neural networks have a lot more depth and require a lot more tinkering, so you will be given the entire project week to work on a neural network IP.

This helps us achieve several objectives:

• Spend more time practicing the newly introduced tensor flow library.
• Give you more time to create a thorough neural network
• Give you an opportunity to practice working together
• Give you an opportunity to present a long term project ahead of the project period.
• The rest of this page will share resources that will help you in this endeavour, as well as instructions for your work.

• You can work on this project in teams of no more than 3 students.
• There is no set objective. Feel free to explore problems that your team is interested by. Your neural net can be a classifier or a regressor.
• You can use previously used datasets, but you have to draw new insight / ask new questions of them.
• You will be assessed during your project week presentations next Friday. Your collab must be submitted prior to the start of presentations.

• video url : (https://youtu.be/DFKHh7_zzJc)
• You are expected to use Tensor Flow for this project, We have seen some cases of using the tool already, but this is an opportunity to go deeper. Start by going through this thorough tutorial over Friday and the week end, and make sure to use the official documentation (https://www.tensorflow.org/guide/effective_tf2) as a reference

Here are two guides to cool Neural network projects. They will be useful for inspiration, and to see how people have built this from scratch. The aim is not to replicate these projects, but to draw inspiration from them. You are strongly encouraged to go through at least one of these.

• Recognizing handwriting using neural networks step by step (http://neuralnetworksanddeeplearning.com/chap1.html)
• Sentiment analysis of movie reviews step by step (https://stackabuse.com/python-for-nlp-movie-sentiment-analysis-using-deep-learning-in-keras/)