Paragraph2vec's work is under suspicion

Quoc v. Le ICML2014 's work "distributed representations of sentences and Documents" has been controversial, and many scholars have claimed that their work cannot be reproduced.

ICLR2015 article "ENSEMBLE of generative anddiscriminativetechniques forsentimentanalysis ofmoviereviews" in the article directly to the evaluation of this work:

IN Our experiments, to match the results from (Le & Mikolov), we followed the suggestion by quocle to use hi Erarchical Softmax instead of negative sampling. However, this produces the 92.6% accuracy resultonly when the training and test data is not shuffled. Thus, we consider this result to be invalid.

"ENSEMBLE of generative anddiscriminativetechniques forsentimentanalysis ofmoviereviews" article Address: http://arxiv.org/pdf/ 1412.5335v6.pdf

Paragraph2vec's work is under suspicion