Extreme Multi-label Loss Functions for Recommendation, Tagging, Ranking & Other Missing Label Applications

By Himanshu Jain, Yashoteja Prabhu, Manik Varma. Knowledge Discovery and Data Mining, 2016

357 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/a9bd477ba37472178e99664f5c5fb3da2380f14e

How researchers see this paper

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

In Artificial Intelligence

Citations in the corpus

AttentionXML: Label Tree-based Attention-Aware Deep Model for High-Performance Extreme Multi-Label Text Classification (2018, methodology): “We chose P@k (Precision at k) [10] as our evaluation metrics for performance comparison, since P@k is widely used for evaluating the methods for XMTC.”
Sampled Softmax with Random Fourier Features (2019, methodology): “Following the convention of extreme classiﬁcation literature [34, 35, 4], we report precision at k ( PREC @ K ) on the test set.”
Learning for Tail Label Data: A Label-Specific Feature Approach (2019, background): “The question on accurately build models for TLD has not been thoroughly studied, though it is widely stated that tail labels are informative and rewarding if predicted correctly [Bhatia et al., 2015; Jain et al., 2016; Xu et al., 2016].”
SVP-CF: Selection via Proxy for Collaborative Filtering Data (2021, methodology): “…missing to decompose over the user and item as follows: pu,i = P (r u i = 1 | r∗ui = 1) = P (ru = 1 | r∗u = 1) · P (ri = 1 | r∗i = 1) = pu · pi Following (Jain et al., 2016), we assume the user and item propensities to lie on the following sigmoid curves: pu := 1 1 + Cu · e−A·log(Nu+B) pi := 1 1…”
Deep Streaming Label Learning (2020, methodology): “…our proposed method was evaluated in terms of both bipartition-based metrics, i.e., Hamming loss, macro-F1, micros-F1, and instance-F1, and ranking-based metrics, i.e., Precision@k, coverage, ranking loss, average precision (AP), macro-AUC, and micro-AUC (Wu & Zhou, 2017; Jain et al., 2016).”