プレゼン?ポスターで自分の研究を「伝える」 (How to do technical oral/poster presentation)Toshihiko Yamasaki
?
MIRU2020若手プログラム招待講演のスライドを一般公開用にアレンジしたものです。日本語で書かれています。下記の点にご注意ください
?セリフが伴ってないので内容は限定的です
?著作権等に配慮しているので中身は結構無味乾燥です。
This is an arranged version of my invited talk at MIRU 2020 young researchers' forum. This is written in Japanese.
This document introduces deep reinforcement learning and provides some examples of its applications. It begins with backgrounds on the history of deep learning and reinforcement learning. It then explains the concepts of reinforcement learning, deep learning, and deep reinforcement learning. Some example applications are controlling building sway, optimizing smart grids, and autonomous vehicles. The document also discusses using deep reinforcement learning for robot control and how understanding the principles can help in problem setting.
プレゼン?ポスターで自分の研究を「伝える」 (How to do technical oral/poster presentation)Toshihiko Yamasaki
?
MIRU2020若手プログラム招待講演のスライドを一般公開用にアレンジしたものです。日本語で書かれています。下記の点にご注意ください
?セリフが伴ってないので内容は限定的です
?著作権等に配慮しているので中身は結構無味乾燥です。
This is an arranged version of my invited talk at MIRU 2020 young researchers' forum. This is written in Japanese.
This document introduces deep reinforcement learning and provides some examples of its applications. It begins with backgrounds on the history of deep learning and reinforcement learning. It then explains the concepts of reinforcement learning, deep learning, and deep reinforcement learning. Some example applications are controlling building sway, optimizing smart grids, and autonomous vehicles. The document also discusses using deep reinforcement learning for robot control and how understanding the principles can help in problem setting.
- A job can be modeled as observing circumstances, judging if conditions are met, and taking action. Certain jobs have a "clear frame" that makes them replaceable by AI.
- Framing is selecting key attributes to define a job's frame. Only jobs with clear frames defined by key attributes can currently be done by AI.
- As AI technologies advance, they can analyze more data types and orchestrate "micro jobs". However, finding proper frames and designing new processes still requires human-level heuristic abilities not found in current AI.
The document discusses the concepts of plan, recognition, and execution as they relate to performing an action like cooking. It explains that an action involves first planning, then recognizing when it's time to execute each step of the plan through recognition cues. Mistakes occur when the plan is wrong, lapses happen when the recognition is faulty, and slips are errors in the execution itself. An example is provided of making an omelette to illustrate the different phases of plan, recognition, and execution.
17. Copyright アイデアクラフト 2017
OODAは間違って理解されている?(2)
16
BoydのOODAコンセプトと、その通俗解釈との違い
OODAはループではない! The OODA Loop is not a loop (1). That came as quite a shock. Maybe I
have been reading with my eyes closed. Time for one of those reality
checks. Maybe we are all too familiar with the concept of the Shewhart
cycle that Deming taught (2). Or too familiar with one of the many
derivatives – such as the plan-do-check-act or PDCA cycle of TQM.
Theory of Constraints Thinking Process Cloud OODA Loop
http://www.dbrmfg.co.nz/Thinking%20Process%20Cloud%20OODA.htm
Implicit Guidance &
Controlが重要!
Thus we come back to the two “implicit guidance and control” arrows that
we mentioned earlier. Let’s return to Robert Coram once again (1). “Note
that Boyd includes the ‘Implicit Guidance & Control’ from ‘Orientation’ with
both ‘Observations’ and ‘Action.’ This is his way of pointing out that when
one has developed the proper Fingerspitzengefuhl [ fingertip feel ] for a
changing situation, the tempo picks up and it seems one is then able to
bypass the explicit ‘Orientation’ and ‘Decision’ part of the Loop, to ‘Observe’
and ‘Act’ almost simultaneously. The speed must come from a deep intuitive
understanding of one's relationship to the rapidly changing environment.
This is what enables a commander seemingly to bypass parts of the Loop. It
is this adaptability that gives the OODA Loop its awesome power.
Theory of Constraints Thinking Process Cloud OODA Loop
http://www.dbrmfg.co.nz/Thinking%20Process%20Cloud%20OODA.htm
つまり、OODAがObserveから始まることには大した意味は無い???ってこと(^_^;)