1. 1988年，朱相竞选上海市市长完整版20分钟发言

  2. 从关系型数据库到分布式机器学习，揭秘腾讯大数据十年发展历程

  3. Uber真正的优势是数据

  4. 美团vs滴滴终局，“聚合生态”的对抗 深度

  5. 日本医疗领域中人工智能技术的应用现状及展望

  6. Smart mobility

  7. 盘后暴涨20%，特斯拉的中国速度成致胜法宝

  8. Tesla’s ‘Full Self-Driving’ feature may get early-access release by the end of 2019

  9. 龙瀛：面向智慧城市的人本尺度城市形态：理论、方法与实践

  10. 手把手教你写好论文里最关键的开头(Introduction)？

  11. 杨振宁：对于这件事情，我很高兴中国政府没有上当

    这么多年我看到了成千个研究生，很多都非常优秀，可是十年以后他们得了博士学位再看，有的人非常成功，有的人非常不成功，并不是因为这些人的本事差了这么多，得到过博士学位的人通常本事都还不坏的；也不是因为有的人努力，有的人不努力。

    主要是有人走对了方向，要是走到一个强弩之末的方向上，那就没有办法的，而且越走越不容易走出来，要换一个方向不容易，继续做那就走成了最不幸的一个人。这点我希望在座的每一个研究生都理解到这几句话的意思。

  12. Lage, Isaac, et al. “An evaluation of the human-interpretability of explanation.” arXiv preprint arXiv:1902.00006 (2019).

    • Motivation: The evaluation of interpretable machine learning systems is challenging, as explanation is almost always a means toward some downstream task.
    • Research question: how increasing different types of complexity affects the usability of explanations for three tasks.
      • Explanation Size
      • Creating New Types of Cognitive Chunks
      • Repeated Terms in an Explanation
    • Methods:
      • Tasks
        • simulating the system’s response
        • verification of a suggested response
        • counterfactual reasoning because they are general enough to be relevant across a variety of domains
      • Metrics
        • response time
        • accuracy
        • subjective satisfaction
    • Results
      • The magnitude of performance effect varies by the type of complexity.
      • Surprisingly, implicit cognitive chunks were faster for people to process than explicit cognitive chunks. Participants
      • The type of question significantly impacted response time.

  13. 滴滴获运筹学领域顶级奖项瓦格纳杰出实践奖