- 1、本文档共38页,可阅读全部内容。
- 2、原创力文档(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)。
查看更多
Some Interesting Problems with Big
Social Media Data & Machine Learning
Huan Liu
Arizona State University Some Interesting Problems for ML 7/9/19, Tencent 1
Data Mining and Machine Learning Lab
From AIK to AID
• “Knowledge is Power”: AI was then solely about K
– Expert Systems or Rule-based Systems
• “Intelligence is ten million rules.”
– Knowledge-based Systems (Cyc)
• “Data is the New Oil”: AI is now hyped up with D
– Data is ubiquitous and big
– Data can be used or misused
• Data + Computing power + ML algorithms
– All problems are solved!?
– No so for computer scientists
Arizona State University Some Interesting Problems for ML 7/9/19, Tencent 2
Data Mining and Machine Learning Lab
Social Media – a New Phenomenon
Arizona State University Some Interesting Problems for ML 7/9/19, Tencent 6
Data Mining and Machine Learning Lab
Unique Social Media Data
• Social media data is
–big and relatively new
–user-generated, noisy, partial, multi-modal
–a new lens to human behavior
• “Data is the new oil”
–It empowers AI and machine learning
–It is used to scout talents, e.g., competitions
• What else can we use SMD for?
–We turn this question to …
Arizona State University Some Interesting Problems for ML 7/9/19, Tencent 7
Data Mining and Machine Learning Lab
Some Interesting Problems for Machine Learning
• Is our social media data really big?
–How can we make data bigger?
• Is it necessary to make a trade-off between
utility and privacy?
• How can we evaluate without ground truth?
–Ground truth exists in competitions
–Without GT, how can we know if our machine
learning result is of any value?
Arizona Sta
文档评论(0)