Was interested in learning about quantum machine learning / deep learning and saw this tutorial page in Tensorflow about MNIST classification:

What it does

Demonstrates quantum object recognition, classifying automobiles and deer by training and testing the quantum neural network model with CIFAR-10 database

How we built it

By modifying the tutorial code above that uses MNIST database to use CIFAR-10 database instead.

Challenges we ran into

The shapes of MNIST dataset and CIFAR-10 dataset are different, with MNIST being 28x28x1 and CIFAR-10 being 32x32x3. So we encountered a lot of errors and had to edit almost every part of the code.

Accomplishments that we're proud of

It works! And it achieved a high accuracy of 90%.

What we learned

How to construct a quantum neural network, and how to train and test it to classify images

What's next?

Train and test a classical model, and compare it to the quantum model that we constructed. We can also try other data encoding methods and see if it enhances our model’s accuracy. Furthermore, we can try training and testing with more types of data, since right now we’ve only used two types.

Built With

  • cirq
  • jupyter-notebook
  • python
  • tensorflow
Share this project: