Unsupervised Context Retrieval for Long-tail Entities

Garigliotti, Darío; Albakour, Dyaa; Martínez, Miguel; Balog, Krisztian

doi:10.1145/3341981.3344244

Cited by 1 publication

(1 citation statement)

References 17 publications

(25 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The phenomenon of long-tail distributions is ubiquitous in IR [8,14,17,18,47,58,82,85]. Specifically, for learning from datasets with a skewed, long-tail distribution of class labels, several strategies have been proposed in previous studies.…”

Section: Learning From Long-tail Datamentioning

confidence: 99%

Long-Tail Hashing

Chen

Hou

Leng

et al. 2021

Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

Hashing, which represents data items as compact binary codes, has been becoming a more and more popular technique, e.g., for large-scale image retrieval, owing to its super fast search speed as well as its extremely economical memory consumption. However, existing hashing methods all try to learn binary codes from artificially balanced datasets which are not commonly available in real-world scenarios. In this paper, we propose Long-Tail Hashing Network (LTHNet), a novel two-stage deep hashing approach that addresses the problem of learning to hash for more realistic datasets where the data labels roughly exhibit a long-tail distribution. Specifically, the first stage is to learn relaxed embeddings of the given dataset with its long-tail characteristic taken into account via an end-to-end deep neural network; the second stage is to binarize those obtained embeddings. A critical part of LTHNet is its dynamic meta-embedding module extended with a determinantal point process which can adaptively realize visual knowledge transfer between head and tail classes, and thus enrich image representations for hashing. Our experiments have shown that LTHNet achieves dramatic performance improvements over all state-of-theart competitors on long-tail datasets, with no or little sacrifice on balanced datasets. Further analyses reveal that while to our surprise directly manipulating class weights in the loss function has little effect, the extended dynamic meta-embedding module, the usage of cross-entropy loss instead of square loss, and the relatively small batch-size for training all contribute to LTHNet's success.

show abstract