Natural Language Processing
The latest innovation (2019) is the BERT, based on the BERT, based on the Transformer.
Tackle any nlp problem for Kaggle: https://www.kaggle.com/code/abhishek/approaching-almost-any-nlp-problem-on-kaggle
One of the big challenges in language is the ambiguity, which causes a lot of confusion. The English language, for example, has a temporal element to the meaning to sentences:
- The driver hit the cyclist and drove on.
- The driver drove on and hit the cyclist.
title: Alvin Li
I talked about this with Alvin Li, since there are a lot of parallels between [[Natural Language Processing|NLP]] and [[Computer Vision|CV]].
He said that he found NLP more interesting because there is more flexibility. With CV, the pixels of the images are just fixed. There is only so much you can do. But with language, it is much more abstract and a more interesting field to tackle.