In this talk, I'll speak about how CNNs can be used for Text Classification. I'll go into the details of gathering and preprocessing data, the basics of CNNs for NLP, defining a CNN architecture for text classification, and using word embeddings such as word2vec and GloVe, and learning your own word embeddings from scratch, all in Tensorflow.
We'll be building a CNN for text classification using the Rotten Tomatoes movie review dataset. This is how the talk will proceed: - Going from Computer Vision to NLP using CNNs (Representing words as vectors and different channels) - Where and how CNNs are currently being used in NLP (a brief literature review) - Gathering and preprocessing data - Basic CNN architecture we'll be using (Different layers) - Visualising the results - Extensions and different tips and tricks