I came across Sebastian Ruder’s Requests for Research post. I am very much interested in the Task-independent data augmentation for NLP topic. However, I have zero research experience. I will be glad to get any collaborators on this to keep the energy going.
Ps: Here’s a link to the full post. Please reach out if you’re interested.
3 Likes
Sounds interesting. I’ll be happy to collaborate.
Hi, I’d suggest you to focus on points 3 and 4 :
Learn the augmentation , Learn a word embedding space for data augmentation
They seem the most interesting directions to me. Read and study the papers Sebastian cited, it should be a good start. But remember point 3 is far harder than 4 due to possible semantic shifts. Feel free to reach out to me if you have any further questions.