Skip-gram with negative sampling, a popular variant of Word2vec originally
designed and tuned to create word embeddings for Natural Language Processing,
has been used to create item embeddings with successful applications in
recommendation. While these fields do not share the same type of data, neither
evaluate on the same tasks, recommendation applications tend to use the same
already tuned hyperparameters values, even if optimal hyperparameters values
are often known to be data and task dependent. We thus investigate the marginal
importance of each hyperparameter in a recommendation setting through large
hyperparameter grid searches on various datasets. Results reveal that
optimizing neglected hyperparameters, namely negative sampling distribution,
number of epochs, subsampling parameter and window-size, significantly improves
performance on a recommendation task, and can increase it by an order of
magnitude. Importantly, we find that optimal hyperparameters configurations for
Natural Language Processing tasks and Recommendation tasks are noticeably
different.
Description
Word2Vec applied to Recommendation: Hyperparameters Matter
%0 Generic
%1 casellesdupr2018word2vec
%A Caselles-Dupré, Hugo
%A Lesaint, Florian
%A Royo-Letelier, Jimena
%D 2018
%K recommendation toread word2vec
%T Word2Vec applied to Recommendation: Hyperparameters Matter
%U http://arxiv.org/abs/1804.04212
%X Skip-gram with negative sampling, a popular variant of Word2vec originally
designed and tuned to create word embeddings for Natural Language Processing,
has been used to create item embeddings with successful applications in
recommendation. While these fields do not share the same type of data, neither
evaluate on the same tasks, recommendation applications tend to use the same
already tuned hyperparameters values, even if optimal hyperparameters values
are often known to be data and task dependent. We thus investigate the marginal
importance of each hyperparameter in a recommendation setting through large
hyperparameter grid searches on various datasets. Results reveal that
optimizing neglected hyperparameters, namely negative sampling distribution,
number of epochs, subsampling parameter and window-size, significantly improves
performance on a recommendation task, and can increase it by an order of
magnitude. Importantly, we find that optimal hyperparameters configurations for
Natural Language Processing tasks and Recommendation tasks are noticeably
different.
@misc{casellesdupr2018word2vec,
abstract = {Skip-gram with negative sampling, a popular variant of Word2vec originally
designed and tuned to create word embeddings for Natural Language Processing,
has been used to create item embeddings with successful applications in
recommendation. While these fields do not share the same type of data, neither
evaluate on the same tasks, recommendation applications tend to use the same
already tuned hyperparameters values, even if optimal hyperparameters values
are often known to be data and task dependent. We thus investigate the marginal
importance of each hyperparameter in a recommendation setting through large
hyperparameter grid searches on various datasets. Results reveal that
optimizing neglected hyperparameters, namely negative sampling distribution,
number of epochs, subsampling parameter and window-size, significantly improves
performance on a recommendation task, and can increase it by an order of
magnitude. Importantly, we find that optimal hyperparameters configurations for
Natural Language Processing tasks and Recommendation tasks are noticeably
different.},
added-at = {2018-06-22T16:35:47.000+0200},
author = {Caselles-Dupré, Hugo and Lesaint, Florian and Royo-Letelier, Jimena},
biburl = {https://www.bibsonomy.org/bibtex/2bbe77a8d4c63d12234d72adc5ed9c3e6/thoni},
description = {Word2Vec applied to Recommendation: Hyperparameters Matter},
interhash = {b19a5e9dd9342faa5c5d384a7b49b605},
intrahash = {bbe77a8d4c63d12234d72adc5ed9c3e6},
keywords = {recommendation toread word2vec},
note = {cite arxiv:1804.04212},
timestamp = {2018-06-22T16:35:47.000+0200},
title = {Word2Vec applied to Recommendation: Hyperparameters Matter},
url = {http://arxiv.org/abs/1804.04212},
year = 2018
}