tensorflow implementation of the FruitFlyEmbeddings Paper
Some Researches proposed a model that is strongly based on the neural net of a fruit fly. They asked whether it is possible to create meaningful embeddings with this simple network.
I want to implement the proposed architecture. Instead of word embeddings I will use market baskets to build product embeddings for an online retailer. This decision is solely based on the fact, that I do not want so spend ~1 day with training and therefore came up with an use cases that is a lot easier to interpret whiel having the same nature of embeddings.
Download the dataset from kaggle and put it into a directory called "data".
Install all requirements:
pip3 install -r requirements.txt