List your company
ResearchFormerly WizWand

Document Classification on Yahoo Answers

76.1AccuracyDPCNN
23.5837.21550.8564.485Jan 5, 2019Aug 30, 2019Apr 24, 2020Dec 18, 2020Aug 13, 2021Apr 8, 2022Dec 2, 2022
Updated 8mo ago

Evaluation results

MethodLinks
DPCNN2019.0976.1
EXAM2019.0974.8
DNC-CUWWriting Strategy=Cached Uniform Writing2019.0174.3
ULMFiT (Small data)Training data percentage=20%2019.0974.3
DNC-UWWriting Strategy=Uniform Writing2019.0174.2
D-LSTM2019.0173.7
Region Embedding2019.0173.7
WC-Reg2019.0973.7
VDCNN2019.0173.4
VDCNN2019.0973.4
ULMFIT (Tiny data)Training data percentage=8%2019.0973.3
ColD-Fusion2022.1272.69
Finetune2022.1272.49
MUPPET2022.1271.9
Multitask2022.1271.71
DRNN2019.0970.3
FlowGMMn_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=semi-supervised2019.1257.4
Π-modeln_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=semi-supervised2019.1256.6
3-Layer NN + Dropoutn_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=labeled data only2019.1255.6
Logistic Regressionn_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=labeled data only2019.1254.9
RBF Label Spreadingn_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=semi-supervised2019.1230.4
kNNn_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=labeled data only2019.1228.4
kNN Label Spreadingn_l=800, n_u=50k, classes=10, Backbone=BERT embeddings, training=semi-supervised2019.1225.6