So, one day a friend and I were wondering what we could possibly do to use machine learning in an interesting way. We happened to have a box of GrapeNuts cereal before us and it was then that we realized that we didn't know what GrapeNuts really are. Who does?! Machine learning was a viable solution.
Thus, we dub our project GrapeNuts Classifier after the cereal which we couldn't classify, but that we loved. Our resulting classifier leverages knowledge of nutritional content (minerals, vitamins, etc.).
Although we do provide the source code here, we do provide the final report. Our best accuracies (thus far) are around 79%, which for a classification having upwards of 25 classes is far above baseline.