Sparse and Expandable Network for Google's Pathways.

IF 2.4 Q3 COMPUTER SCIENCE, INFORMATION SYSTEMS Frontiers in Big Data Pub Date : 2024-08-29 eCollection Date: 2024-01-01 DOI:10.3389/fdata.2024.1348030
Charles X Ling, Ganyu Wang, Boyu Wang
{"title":"Sparse and Expandable Network for Google's Pathways.","authors":"Charles X Ling, Ganyu Wang, Boyu Wang","doi":"10.3389/fdata.2024.1348030","DOIUrl":null,"url":null,"abstract":"<p><strong>Introduction: </strong>Recently, Google introduced Pathways as its next-generation AI architecture. Pathways must address three critical challenges: learning one general model for several continuous tasks, ensuring tasks can leverage each other without forgetting old tasks, and learning from multi-modal data such as images and audio. Additionally, Pathways must maintain sparsity in both learning and deployment. Current lifelong multi-task learning approaches are inadequate in addressing these challenges.</p><p><strong>Methods: </strong>To address these challenges, we propose SEN, a Sparse and Expandable Network. SEN is designed to handle multiple tasks concurrently by maintaining sparsity and enabling expansion when new tasks are introduced. The network leverages multi-modal data, integrating information from different sources while preventing interference between tasks.</p><p><strong>Results: </strong>The proposed SEN model demonstrates significant improvements in multi-task learning, successfully managing task interference and forgetting. It effectively integrates data from various modalities and maintains efficiency through sparsity during both the learning and deployment phases.</p><p><strong>Discussion: </strong>SEN offers a straightforward yet effective solution to the limitations of current lifelong multi-task learning methods. By addressing the challenges identified in the Pathways architecture, SEN provides a promising approach for developing AI systems capable of learning and adapting over time without sacrificing performance or efficiency.</p>","PeriodicalId":52859,"journal":{"name":"Frontiers in Big Data","volume":"7 ","pages":"1348030"},"PeriodicalIF":2.4000,"publicationDate":"2024-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11390433/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Frontiers in Big Data","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3389/fdata.2024.1348030","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/1/1 0:00:00","PubModel":"eCollection","JCR":"Q3","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
引用次数: 0

Abstract

Introduction: Recently, Google introduced Pathways as its next-generation AI architecture. Pathways must address three critical challenges: learning one general model for several continuous tasks, ensuring tasks can leverage each other without forgetting old tasks, and learning from multi-modal data such as images and audio. Additionally, Pathways must maintain sparsity in both learning and deployment. Current lifelong multi-task learning approaches are inadequate in addressing these challenges.

Methods: To address these challenges, we propose SEN, a Sparse and Expandable Network. SEN is designed to handle multiple tasks concurrently by maintaining sparsity and enabling expansion when new tasks are introduced. The network leverages multi-modal data, integrating information from different sources while preventing interference between tasks.

Results: The proposed SEN model demonstrates significant improvements in multi-task learning, successfully managing task interference and forgetting. It effectively integrates data from various modalities and maintains efficiency through sparsity during both the learning and deployment phases.

Discussion: SEN offers a straightforward yet effective solution to the limitations of current lifelong multi-task learning methods. By addressing the challenges identified in the Pathways architecture, SEN provides a promising approach for developing AI systems capable of learning and adapting over time without sacrificing performance or efficiency.

查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
谷歌 Pathways 的稀疏可扩展网络。
简介最近,谷歌推出了下一代人工智能架构 Pathways。Pathways 必须解决三个关键挑战:为多个连续任务学习一个通用模型;确保任务之间可以相互利用,同时不遗忘旧任务;从图像和音频等多模态数据中学习。此外,Pathways 还必须在学习和部署过程中保持稀疏性。目前的终身多任务学习方法不足以应对这些挑战:为了应对这些挑战,我们提出了稀疏可扩展网络 SEN。SEN 的设计目的是通过保持稀疏性来同时处理多个任务,并在引入新任务时实现扩展。该网络利用多模态数据,整合来自不同来源的信息,同时防止任务之间的干扰:结果:所提出的 SEN 模型在多任务学习方面有显著改进,成功地管理了任务干扰和遗忘。它有效整合了各种模式的数据,并在学习和部署阶段通过稀疏性保持了效率:SEN 为解决当前终身多任务学习方法的局限性提供了一个简单而有效的解决方案。通过解决 Pathways 架构中发现的挑战,SEN 为开发能够在不牺牲性能或效率的情况下进行长期学习和适应的人工智能系统提供了一种前景广阔的方法。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
CiteScore
5.20
自引率
3.20%
发文量
122
审稿时长
13 weeks
期刊最新文献
Editorial: Visualizing big culture and history data. On explaining recommendations with Large Language Models: a review. Enhancing smart home environments: a novel pattern recognition approach to ambient acoustic event detection and localization. Balancing act: Europeans' privacy calculus and security concerns in online CSAM detection. A scalable tool for analyzing genomic variants of humans using knowledge graphs and graph machine learning.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1