Sparse Local Embeddings for Extreme Multi-label Classification

By Kush Bhatia, Himanshu Jain, Purushottam Kar, Manik Varma, Prateek Jain. Neural Information Processing Systems, 2015

464 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/7eb44c7bae8f364b0777c2bf3840cb0872667233

How researchers see this paper

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

1. Dacheng Tao
2. Zhen Wang
3. Percy Liang
4. Liu Liu
5. Erik Cambria
6. Trevor Darrell
7. Yoshua Bengio
8. Zhe Gan

Citations in the corpus

• AttentionXML: Label Tree-based Attention-Aware Deep Model for High-Performance Extreme Multi-Label Text Classification (2018, unknown): “They can be categorized into the following four types: 1) 1-vs-All [3,4,30,31], 2) Embedding-based [7,27], 3) Instance [10,25] or label tree-based [11, 13, 24, 28]) and 4) Deep learning-based methods [17] (see Appendix for more descriptions on these methods).”
• Long-context LLMs Struggle with Long In-context Learning (2024, unknown): “Existing methods to tackle Extreme-label Classification tasks range from embedding-based approaches to fine-tuned retrievals (Bhatia et al., 2015; Vuli´c et al., 2021), focusing on efficiently managing and leveraging the large label space.”
• Contextual Memory Trees (2018, unknown): “We compare CMT-augmented OAS to OAA on three multi-label datasets, RCV1-1K (Prabhu & Varma, 2014), AmazonCat-13K (McAuley & Leskovec, 2013), and Wiki-31K (Zubiaga, 2012; Bhatia et al., 2015).”
• Neural Message Passing for Multi-Label Classification (2019, background): “The second group learns a shared latent space representing both input features and output labels, and then upsamples from the space to reconstruct the target labels [57, 6].”
• One-vs-Each Approximation to Softmax for Scalable Estimation of Probabilities (2016, methodology): “Notice thatBIBTEX is originally a multi-label classification dataset (Bhatia et al., 2015). where each example mayhave more than one labels.”
• Learning Deep Latent Spaces for Multi-Label Classification (2017, methodology): “…is also a popular technique for multi-label classification (Balasubramanian and Lebanon 2012; Tai and Lin 2012; Chen and Lin 2012; Bi and Kwok 2013; Hsu et al. 2009; Zhang and Schneider 2012; Zhou, Tao, and Wu 2012; Lin et al. 2014; Yu et al. 2014; Li and Guo 2015; Bhatia et al. 2015).”
• Towards Scalable and Reliable Capsule Networks for Challenging NLP Applications (2019, methodology): “, SLEEC (Bhatia et al., 2015).”
• The Emerging Trends of Multi-Label Learning (2020, methodology): “Embedding: SLEEC [1] Oðn$2 þ n$CÞ Oðndþ kLÞ”
• Zero-Shot Recognition via Structured Prediction (2016, background): “The motivation for ZSR stems from the need for solutions to diverse research problems ranging from poorly annotated big data collections [1] to the problem of extreme classification [2].”
• General Multi-label Image Classification with Transformers (2020, background): “(2) Shared Embedding Space: The second group learns to project input features and output labels into a shared latent embedding space [57, 3].”
• A statistical perspective on distillation (2021, methodology): “We now assess the negativeaware distillation objective on benchmark datasets for multiclass retrieval, AMAZONCAT-13K and AMAZONCAT670K (McAuley & Leskovec, 2013; Bhatia et al., 2015).”
• Adversarial Multi-label Prediction for Spoken and Visual Signal Tagging (2019, background): “Alternative label embedding approaches include principal label space dimension reduction (PLST) [5] and sparse local embedding (SLEEC) [6].”
• Deep Streaming Label Learning (2020, methodology): “…and exploiting the correlations across labels (Zhang & Zhou, 2006; Read et al., 2011; Nam et al., 2019), and (2) reducing the label space by embedding original high-dimensional label vectors into the low-dimensional representation (Tsoumakas et al., 2011; Bhatia et al., 2015; Liu & Shen, 2019).”
• Order-Free RNN with Visual Attention for Multi-Label Classification (2017, methodology): “On the other hand, probabilistic graphical model based methods [27], [28] learn label dependencies with graphical structure, and latent space methods [29], [30] choose to project features and labels into a common latent space.”
• Learning for Tail Label Data: A Label-Specific Feature Approach (2019, methodology): “In this experiment, we compare the performance of TAIL with state-of-the-art methods: FastXML, LEML, SLEEC, DiS- MEC, CoH, PD-Sparse, and REML.”
• Augment and Reduce: Stochastic Inference for Large Categorical Distributions (2018, unknown): “, 2015), EURLex-4K (Mencia & Furnkranz, 2008; Bhatia et al., 2015), and AmazonCat-13K (McAuley & Leskovec, 2013).”
• CIFDM: Continual and Interactive Feature Distillation for Multi-Label Stream Learning (2021, methodology): “For example, C2AE [23] applies the DNN model as feature encoder to utilize features and label embedding jointly; SLEEC[4] learns a set of embeddings to preserve pairwise distances between only the nearest label vectors, and accurately predict infrequently occurring (tail) labels.”