Meme Sentiment Analysis

BITS, Pilani - Goa | November 2019

This project

Was part of the course Neural Networks and Fuzzy Logic (NNFL). The objective of this project was to predict the sentiment, offensiveness and motivation quotient of the meme given the image and caption of the memes, using a multimodal deep learning model. The motivation label had 2 classes (motivational, not_motivational), the sentiment label had 5 classes (very_positive, positive, neutral, negative, very_negative) and the offensiveness label had 4 classes (not_offensive, slightly_offensive, very_offensive, hateful_offensive).

Project Report