Pup or Pastry

Inspiration

We were interested in testing the limits of convolutional neural networks given computational space and time constraints. We were also inspired by the creative evolution of internet culture.

What it does

Using Caffe, a deep learning framework based on convolutional neural networks, we performed binary classification of canines and baked goods, given highly similar images.

How we built it

Team members fed images of specifically cropped chihuahuas and a variety of muffins into a neural network inspired by the architectural design of proven machine learning methods for image identification using Imagenet.

Challenges we ran into

Our team had minimal exposure to neural networks, but maximum exposure to memes.

Accomplishments that we're proud of

We created a functional 8-layer convolutional neural network, inspiring a new generation to learn about machine learning by applying it to modern culture.

What we learned

As a machine learning method as complex as a convolutional neural network had difficulty distinguishing chihuahuas from muffins, we realized that there are still strides to be made in deep learning.

What's next for Pup or Pastry

We plan to tackle other such major issues as Kiwi or Shrew, as well as Labradoodle or Fried Chicken, with more thorough parameter tuning to test the true limits of deep learning algorithms.

Built With

• c
• cmake
• machine-learning
• protocol-buffer
• shell