These course notes were developed using lectures/material from the Neural Networks for NLP (Lecture 1) course by Graham Neubig. The visuals and equations are adopted from the original slides. I simply hope that the notes serve as accompanying study material. Find all my notes for this course in the ML Course Notes repo.
To better understand why we need neural networks to solve problems in natural language processing, we first need to understand why it’s really difficult to handle language.
To start off, we first look at the concept of acceptability judgement.
Acceptability judgement refers to judgements about whether a particular sentence is acceptable or not. For instance, judge whether a sentence is natural, semantically correct, or grammatically correct, etc.
To understand the concept better, let’s look at the following sentences: