Term Project

Fall 2020 projects

=======================================================

Spring 2020 projects

As the term project you may choose one the following subjects:

1- Using Boston housing price regression dataset from Keras, a random forest regression model will be created. Then you need to optimize your model. The project includes research on the optimization methods and their impact on the accuracy of the prediction.

2-Using MNIST dataset of fashion articles from keras, a NN is used to classify the data. Then, try reducing the number of input features used. The goal is developing a strategy for selecting the most important features form a feature set.

3- Using CIFAR10 small image from Keras, we want to develop a GAN network to fabricate similar data. The network should be able to display orginal and fabricated images.

4- Use MNIST dataset of handwritten digits with an auto-encoder network to compress the data. Discuss the impact of using multiple hidden layers. How can we choose the best number of neurons in the hidden layer(s)?

5- Some machine learning algoeithms are sensitive to noise in data. Use one of the free data sets and add noise with different rates. Then verify the impact on the learning and classification performance of the ML algorithm that you examine. Suggest possible improvements.

6- Deep learning methods extract features from data but sometimes the features are not what we meant. For instance, to classify dogs and cats, if you use imges of dogs taken in shadow, and cats taken at sun light, the brightness will be accepted as a feature. Verify this problem by choosing appropriate data sets. What solutions are there (without chaging your dataset) ? Implement one of the solutions and discuss it. 

====================================================

You are supposed to decide about the details of your project. The designed project should be presented orally, and a report describing the details should be submitted on the dates to be announced.