arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.15093cs.LG
收藏

跨多任务排序的表示学习

Representation Learning by Ranking across multiple tasks

Tianjin University(天津大学)

Lifeng Gu

AI总结:

本文提出将不同任务下的表示学习统一转化为排序问题,通过优化排序损失实现统一学习,实验证明其在分类、检索、多标签学习、回归及自监督任务中均优于现有方法。

摘要:

近年来,表示学习已成为机器学习社区的研究焦点。大规模神经网络是实现通用智能的关键一步,其成功很大程度上归功于其学习数据抽象表示的能力。多个学习领域正在积极讨论如何学习表示,但缺乏统一的视角。我们将不同任务下的表示学习问题转化为排序问题。通过采用排序问题作为统一视角,表示学习任务可以通过优化排序损失以统一方式解决。在多种学习任务(如分类、检索、多标签学习和回归)下的实验证明了基于排序的表示学习框架的优越性。此外,在自监督学习任务下的实验表明,该排序框架在处理无监督训练数据方面具有显著优势,数据增强技术进一步提升了其性能。

英文摘要:

In recent years, representation learning has become the research focus of the machine learning community. Large-scale neural networks are a crucial step toward achieving general intelligence, with their success largely attributed to their ability to learn abstract representations of data. Several learning fields are actively discussing how to learn representations, yet there is a lack of a unified perspective. We convert the representation learning problem under different tasks into a ranking problem. By adopting the ranking problem as a unified perspective, representation learning tasks can be solved in a unified manner by optimizing the ranking loss. Experiments under various learning tasks, such as classification, retrieval, multi-label learning, and regression, prove the superiority of the representation learning by ranking framework. Furthermore, experiments under self-supervised learning tasks demonstrate the significant advantage of the ranking framework in processing unsupervised training data, with data augmentation techniques further enhancing its performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑