machine learning - What is the advantage of the paperboat format in performance optimization of ML? -


the paperboat format claims provide better dataset representation machine learning routines. i'd understand nature of optimization. understand using integer representation model attributes means faster processing of data set, other improvements.

also, how tune ml algorithm work file format.

i don't know if format provides better representation, can speculate why can more efficient.

first, state @ format description, "having data of same precision consecutive enables hardware vectorization."; consider wikipedia: "vector processing techniques have since been added modern cpu designs".

second, format allows mix sparse , non-sparse features, since sparse features placed consequently, possible take them sparse matrix , optimize methods learning conjugate gradient.

how tune ml algorithm work file format?

what mean ml algorithm tuning? learning algorithm doesn't know , doesn't need know file format of dataset; , can't increase or decrease accuracy if know file format. in theory, can speed concrete optimization algorithm (like gradient descent) if can rely on properties of data (and, guess, ismion paperboat it), don't think can tune yourself.


Comments

Popular posts from this blog

Java 3D LWJGL collision -

spring - SubProtocolWebSocketHandler - No handlers -

methods - python can't use function in submodule -