Multi-label Learning with Random Circular Vectors
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: Extreme Multi-label Classification, Random Projection, Vector Symbolic Architectures
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
TL;DR: This paper provides a novel method that uses random circular vectors for efficient multi-label learning.
Abstract: Extreme multi-label classfication (XMC) task involves learning a classifier which can predict from a large label set the most relevant subset of labels for a data instance. While deep neural networks (DNNs) have demonstrated remarkable success in XMC problems, the task is still challenging because it needs to deal with a large number of output labels, which make the DNN training computationally expensive. This paper addresses solving the issue by exploring the use of random circular vectors where each vector component is represented as a complex amplitude. In our proposed framework, we can develop an output layer and loss function of DNNs for XMC by representing the final output layer as a fully connected layer that directly predicts a low-dimensional circular vector encoding a set of labels for a data instance. In this paper, we firstly conducted experiments on synthetic datasets to verify that circular vectors have better label encoding capacity and retrieval ability than normal real-valued vectors. Then, our experimental results on XMC datasets also show that these appealing properties of circular vectors contribute to significant improvements in XMC task performances, compared with a previous model using random real-valued vectors, while reducing the size of the output layers by up to 97%.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 2445
Loading