… extrememulti-labeltextclassification (XMC) problem: given an input text, return the most relevant labels from a large label … On the right is the clusters distribution after our semantic label …
… relevant labelclusters (instead of labels themselves). If B ≈ … framework for extrememulti-label textclassification where the … objectives defined by a hierarchical label tree. This allows the …
D Zong, S Sun - arXiv preprint arXiv:2012.05860, 2020 - arxiv.org
… Extrememulti-labeltextclassification (XMTC) aims … clusters varying with different label clustering approaches. 75.8% of the clusters have more than 100 training instances by our label …
R You, Z Zhang, S Dai, S Zhu - arXiv preprint arXiv:1904.12578, 2019 - arxiv.org
… Different from Parabel, we only keep the leaves of this tree as its clustering result. Now we … label-wise attention network, which can solve extrememulti-labeltextclassification efficiently …
… data into many clusters, and learns embeddings for each separate cluster, then kNN search is performed only within the cluster this novel document belongs to. Since clustering high di…
R You, Z Zhang, Z Wang, S Dai… - Advances in neural …, 2019 - proceedings.neurips.cc
… Extrememulti-labeltextclassification (XMTC) is an important … text with the most relevant multiple labels from an extremely … clusters until the cluster size (the number of labels in each …
… of a Deep Learning architecture devoted to textclassification, considering the extreme multi-class and multi-labeltextclassification problem, when a hierarchical label set is defined. The …
X Huang, B Chen, L Xiao, J Yu, L Jing - Neural Processing Letters, 2022 - Springer
… Extrememulti-labeltextclassification (XMTC) aims at tagging a document with most relevant labels from an extremely large-scale label set… set into several clusters, and in each cluster it …
K Bhatia, H Jain, P Kar, M Varma… - Advances in neural …, 2015 - proceedings.neurips.cc
… extrememulti-label learning is to train a classifier that can automatically tag a novel data point with the most relevant subset of labels … To scale our algorithm, we perform a clustering of …