re: Supervised Learning: Instance-based Learning and K-Nearest Neighbors VIEW POST


Depending on the number of dimensions you can try an embedding (word2vec, gloVe, fastText, ...) or something like a dimensionality reduction (PCA, ISOMAP, ...). I have made mostly positive experience with PCA + kNN. When working on NLP problems word2vec and kNN did well too.

BTW: Thrilled for you VC dimension post. This topic can be confusing.


fyi i wrote up VC Dimensions. it is confusing indeed but i learned a lot from examples.


thanks - i'm very keen on NLP type problems and unfortunately it doesnt seem to be covered in any detail in this course. so after i end this series i may do a further deep dive on this, or maybe if you wanna write up some things i can explore (a pseudo syllabus?) i will happily do that

code of conduct - report abuse