Overview
Throughout Fall 2019 and Winter 2020, our team worked on replicating the structure of a text-adaptive generative adversarial network (TAGAN). Proposed in 2018, TAGAN differs because of its text-adaptive discriminator that creates word-level local discriminators according to input text to classify fine-grained attributes independently. We implemented the network using the materials from the original research paper, and with the help of our advisor, Anna Rafferty. Our code was written in python and we used the PyTorch library for our machine learning framework. We trained our model using a GPU on Google Colab.
Our Paper
In this study, we replicate the text-adaptive generative adversarial network proposed by the paper "Text-Adaptive Generative Adversarial Networks: Manipulating Images with Natural Language", NeurIPS 2018. We do this for two reasons: first, image modification with natural language is an interesting and challenging problem, and second, replicating studies can provide additional information about the vailidity and generalizability of the original authors' claims.
Download Paper