- 1、本文档共29页,可阅读全部内容。
- 2、有哪些信誉好的足球投注网站(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。
- 3、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载。
- 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
- 5、该文档为VIP文档,如果想要下载,成为VIP会员后,下载免费。
- 6、成为VIP后,下载本文档将扣除1次下载权益。下载后,不支持退款、换文档。如有疑问请联系我们。
- 7、成为VIP后,您将拥有八大权益,权益包括:VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
- 8、VIP文档为合作方或网友上传,每下载1次, 网站将根据用户上传文档的质量评分、类型等,对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档
查看更多
Dual Coordinate Descent Algorithms for EfficientLarge Margin Structured Prediction
Ming-Wei Chang and Scott Wen-tau Yih
Microsoft Research
1
Motivation
Many NLP tasks are structured
Parsing, Coreference, Chunking, SRL, Summarization, Machine translation, Entity Linking,…
Inference is required
Find the structure with the best score according to the model
Goal: a better/faster linear structured learning algorithm
Using Structural SVM
What can be done for perceptron?
2
Two key parts of Structured Prediction
Common training procedure (algorithm perspective)
Perceptron:
Inference and Update procedures are coupled
Inference is expensive
But we only use the result once in a fixed step
3
Observations
4
Observations
Inference and Update procedures can be decoupled
If we cache inference results/structures
Advantage
Better balance (e.g. more updating; less inference)
Need to do this carefully…
We still need inference at test time
Need to control the algorithm such that it converges
5
Questions
Can we guarantee the convergence of the algorithm?
Can we control the cache such that it is not too large?
Is the balanced approach better than the “coupled” one?
6
Contributions
We propose a Dual Coordinate Descent (DCD) Algorithm
For L2-Loss Structural SVM; Most people solve L1-Loss SSVM
DCD decouples Inference and Update procedures
Easy to implement; Enables “inference-less” learning
Results
Competitive to online learning algorithms; Guarantee to converge
[Optimization] DCD algorithms are faster than cutting plane/ SGD
Balance control makes the algorithm converges faster (in practice)
Myth
Structural SVM is slower than Perceptron
7
Outline
Structured SVM Background
Dual Formulations
Dual Coordinate Descent Algorithm
Hybrid-Style Algorithm
Experiments
Other possibilities
8
Structured Learning
9
The Perceptron Algorithm
10
Structural SVM
Objective function
Distance-Augmented Argmax
11
Dual formulation
12
Outline
Structured SVM Background
Dual Formu
您可能关注的文档
- Document Analysis TeacherWeb文献分析teacherweb.ppt
- Document Design Weber State University文件设计韦伯州立大学.ppt
- Documenting a Dying Language University of 记录一个垂死的语言大学.ppt
- DOCUMENT TITLE SECOND LINE AND THIRD LINE 文件标题二号线和第三号线.pptx
- Dodge City, Kansas USD 281美国堪萨斯的道奇城281.pptx
- Documents officiels ORBi Home文件或officiels 首页.ppt
- Documenting the Family History An Overview of 记录家族病史的概述.ppt
- Documenting Sources s文件来源.bellevuecollege.pptx
- documents文件.routledgeinteractive.s3.amazonaws.pptx
- DOE Lessons Learned from SF6 Reductions从SF6减少学习DOE的教训.pptx
最近下载
- 专题10 自我成长-2025年中考语文现代文阅读热门主题预测与答题技巧指导(原卷版).docx VIP
- 基于知识图谱的高中生数学科个性化学习路径设计与实施策略教学研究课题报告.docx
- 医院消防安全管理制度汇编(最全精编无错版).doc VIP
- 中小学肺结核病防控知识培训课件.pptx VIP
- 中长期激励办法.pdf VIP
- 《跨区域调水系统生态产品价值核算规范》(DB3210T1177-2024).pdf VIP
- 2025年中小学教师公开招聘中学语文学科笔试大纲.docx
- 《通信发展史》ppt课件.pptx VIP
- 2025年秋季新人教版8年级上册物理全册教学课件(新版教材).pptx
- 医院感染管理制度全套汇编(2022年修订版).pdf VIP
文档评论(0)