Label Partitioning For Sublinear Ranking

By Jason Aaron Edward Weston, Ameesh Makadia, Hector Yee. 2013

123 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Erik Cambria
2. Wei Zhao
3. Percy Liang
4. Dacheng Tao
5. Prateek Jain
6. Yoshua Bengio
7. Haiyun Peng
8. Trevor Darrell

Citations in the corpus

• Deep Neural Networks for YouTube Recommendations (2016, methodology): “Previous systems at YouTube relied on hashing [24] and the classifier described here uses a similar approach.”
• Learning Tree-based Deep Model for Recommender Systems (2018, background): “Unlikemulticlass andmulti-label classification works [26, 32] where tree is used to partition samples or labels, our recommendation tree indexes items for retrieval.”
• Binary Stochastic Representations for Large Multi-class Classification (2017, methodology): “For example, especially when an existing hierarchy is available, one can use hierarchical models [5,6,7] classifying in logarithmic time.”
• Towards Scalable and Reliable Capsule Networks for Challenging NLP Applications (2019, background): “Tree-based models (Agrawal et al., 2013; Weston et al., 2013) induce a tree structure that recursively partitions the feature space with nonleaf nodes.”
• Extreme Multi Class Classification (2013, methodology): “Also another work by Weston et al. (2013) make use of the k-means hierarchical clustering (other distortion-minimizing method could also be used) to recover the label sets for a given partition though this work primarily adresses the problem of ranking a very large set of labels rather than the…”
• Deep Extreme Multi-label Learning (2017, methodology): “oneagainst-all classifier1, as mentioned in Section 1, various tree based and embedding based models are devised. Tree based methods for XML: The label partitioning by sublinear ranking (LPSR) method [30] focuses on reducing the prediction time by establishing a hierarchy structure over a benchmark ”
• FastXML: a fast, accurate and stable tree-classifier for extreme multi-label learning (2014, methodology): “Similarly, [35] treated YouTube videos as distinct labels in an extreme classifier so as to recommend a ranked list of videos to users.”
• Deep Learning for Extreme Multi-label Text Classification (2017, background): “Another category of e‚orts that has improved the state of the art of XMTC in recent years are the new variants of tree-based ensemble methods [1, 36, 41].”
• Sparse Local Embeddings for Extreme Multi-label Classification (2015, methodology): “using a label hierarchy, such hierarchies are unavailable in many applications [1, 2].”
• The Emerging Trends of Multi-Label Learning (2020, methodology): “MLRF [70] and LPSR [71] optimize the Gini index and clustering error as their local measure of performance.”
• Extreme Multi-label Loss Functions for Recommendation, Tagging, Ranking & Other Missing Label Applications (2016, methodology): “As such, the Hamming loss [4, 6, 9, 10, 13, 18, 21, 39, 42, 47] continues to be one of the most popular losses for extreme multi-label learning along with precision [3, 5, 17, 20, 21, 33, 44, 45] and the F-measure [4, 10, 14, 18, 19, 21, 39, 49].”