arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2110.04599cs.LGcs.MM
收藏

Embed Everything:一种有效的多模态空间共嵌入方法

Embed Everything: A Method for Efficiently Co-Embedding Multi-Modal Spaces

Carnegie-Mellon University(卡内基梅隆大学);Harvey Mudd College(哈维穆德学院);Soot

Sarah Di, Robin Yu, Amol Kapoor

摘要:

任何通用的人工智能系统都必须能够在多模态潜在空间中解释、操作和生成数据,这些潜在空间可以表示音频、图像、文本等。在过去的十年中,深度神经网络在单峰数据分布方面取得了显着的成功,而迁移学习技术在相关领域的模型重用方面取得了巨大的发展。然而,从头开始训练多模态网络仍然昂贵且虚幻,而异构迁移学习(HTL)技术仍然相对不发达。在本文中,我们提出了一种新的和成本有效的HTL策略的共同嵌入多模态空间。我们的方法通过使用所有组件的预训练模型来预处理嵌入,而不通过这些模型传递梯度,从而避免了成本效率低下。我们证明了使用该系统在联合图像音频嵌入任务。我们的方法具有广泛的应用,因为成功地弥合不同潜在空间之间的差距可以为承诺的“通用”嵌入提供一个框架。

英文摘要:

Any general artificial intelligence system must be able to interpret, operate on, and produce data in a multi-modal latent space that can represent audio, imagery, text, and more. In the last decade, deep neural networks have seen remarkable success in unimodal data distributions, while transfer learning techniques have seen a massive expansion of model reuse across related domains. However, training multi-modal networks from scratch remains expensive and illusive, while heterogeneous transfer learning (HTL) techniques remain relatively underdeveloped. In this paper, we propose a novel and cost-effective HTL strategy for co-embedding multi-modal spaces. Our method avoids cost inefficiencies by preprocessing embeddings using pretrained models for all components, without passing gradients through these models. We prove the use of this system in a joint image-audio embedding task. Our method has wide-reaching applications, as successfully bridging the gap between different latent spaces could provide a framework for the promised "universal" embedding.

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。

↑