What are some good resources for multi-class text classification using word2vec followed by SVM/ANN / Deep Networks?

Question

I need to implement a multi-class text classifier. I thought of using word2vec, can someone lead me to good papers/resources which talk about this. i would have 4-5 classes and I have loads of data. I have to manually label some of them. It would also be great if someone could throw light on the training size. I plan to use pre-trained word2vec for word embedding.

Answer 1

You can use MLP for this task, check Keras' example for classifying Reuters documents. You can add an embedding layer in the beginning of the model and load it with pretrained model weights.

Keep in mind that there is a multi-class classification which outputs a single class for each input and there is multi-label classification which might output many classes for each input (you didn't say anything about your data character so just letting you know).