3 ms·
It’s paywalled but the main issue I see is in getting useful labels on data. For example, we might know that dolphins make all sorts of noises but without any i
by janalsncm 4y ago
It’s paywalled but the main issue I see is in getting useful labels on data. For example, we might know that dolphins make all sorts of noises but without any information about what those noises are it will be basically impossible derive any deeper meaning.
- blamestross 4y agoModern translation actually doesn't use labels much, it uses structural alignment. I'm more concerned with how to make a comparable "corpus" to human languages.
- throwaway1851 4y agoWell, there’s an implicit label, isn’t there? Translation models are shown pairs of texts in two languages, so the implicit label is “these two texts are saying the same thing”. Won’t you always need some form of supervision to learn the mapping from one domain to another?
- janalsncm 4y agoYes the canonical examples are called parallel corpuses. Sentence pairs of (english_sentence, french_sentence). Because we don’t know what dolphins are saying in the first place, there’s no way to generate a parallel corpus.