Harvard EE | Machine Learning with Strategic Consideration

分享项目信息给更多好友,平台将为您当日的首次分享捐出0.1元公益善款。清帆公益行动期待你的加入!
项目经历 学习机会
2016.10.10-2016.10.30
难度
86人已参加

简介

Machine Learning enables us to discover patterns from data and make predictions/decisions directly from the data we have. It has been a workhorse which is widely used in biomedical, industry engineering, economics and other area.However, in some real world application, especially in economics, the data collection is not an ideal and free process.For one thing, respondents may lie about their answer for the questions because the question includes their sensitive information or they want to interfere the learning model on purpose; for another thing, collecting data will cost money according to their collection approach,contents of data and other facts. Thus recent years, there are many works with this kind of strategic considerations like how to elicit true answers from respondents, how to make learning algorithm more robust when confronted with strategic answers, and how to make the learning algorithm accurate with lowest cost. In this project, you are expected to do a literature survey on relevant papers, taste a flavor on how scientists deal with real world problem with mathematical models and try to propose (and realize) your own model.

工作时间

  • 项目周期:3周
  • 预估每周工作量:8-10小时(均为弹性工作时间)

工作内容

任务1

Task 1:

Get Started. This task need some prior knowledge on Probably Approximately Correct Learning Model[1] and Game theory[2]. So in this week you are asked to survey the literature on PACL and game theory, and answer the questions below so that you can smoothly move to the main topic of this project.
任务2

Task 2:

In this week you are expected to do literature research on learning with strategic considerations. Please answer the following questions so that you can have a first step understanding on how we deal with different kinds of strategic situation in machine learning.[3, 4,5, 6, 7, 8, 9, 10]
任务3

Task 3:

This week you are expected to propose your own model dealing with strategic agentsin learning process. It is definitely a wide topic and here are some reference directions that you can think of

工作方式

Duration: 3 weeks (From 2016-10-10 to 2016-10-30)
Estimated workload: 8-12 hours/week

工作成果评估方法

The final score is peer-review score (30%) plus project host score (70%)

收获

免费学习机会
在实际项目中全面了解机器学习技术
团队素质培养
在真实项目环境建立团队合作意识和能力,培养领导力和技术队伍协作能力
获得工作经历
按要求完成实习任务者,获得项目方认可的清帆在线项目经历

常见问题

清帆(qingfan.com)是一个创新2.0平台,为企业、大学和社会机构提供外部人才管理技术与服务。清帆通过高质量的公开在线项目(Massive Open Online Projects),连接机构与来自全球的优秀人才。
清帆通过在线项目的形式,帮助项目提供方(一流企业、大学和社会机构)连接来自全球的优秀人才。
清帆平台上的项目完全来自一流企业、高校与社会机构,项目相关信息绝对真实可靠。
参与机构发布的项目不需要付费,也不提供现金报酬。
参与清帆项目可以获得以下好处:
(1)不受时间空间限制,可以更多机会参与一流企业、大学和社会机构的高水平项目。打破专业限制,可以尝试不同行业岗位的项目。
(2)提升个人职业能力。通过项目积累,可以更好地了解自己专业知识和通用能力的水平。
(3)了解行业信息与岗位能力需求。针对性地逐步提升职业能力。
(4)获得清帆平台的项目经历。该经历记录了项目中的详细表现和输出成果,并可以追溯。
(5)表现优秀者有机会获得机构活动邀请和工作职位。
可以通过清帆平台进行组队,或是线下组好队一起参加。若无线下小组,清帆推荐系统也会为你找到匹配的队友。可以通过清帆平台认识更多有趣的小伙伴。
组队后通过小组内部推荐选择组长。组长将负责任务分配、时间节点把控、组内沟通协调,并承担部分具体工作内容。
更新
统计