Memo 20180126.

- Accumulative learning - We build on a foundation of “obtainable, given, working hypothesis” to reach what we want to get. - In-depth learning - First, there is “what you want to get,” and from there, “what is necessary to get it. - Don’t learn what you don’t need. - Better to learn” is True, but it is True for most things, so it is uninformative advice in terms of what to invest limited resources of time in.Investment Strategies - After digging deeper and getting to the resources that can be procured, it is possible to create “what you want to get”. - In other words, this is the process of connecting between “want to get” and “can get. - If it’s connected, we can make it. - new combination - percolation - The state of lack of digging is “Rootless Knowledge.” - Gap between the availability line and the - The gap makes it impossible to actually create and generate real benefit. - It would still be redeeming if they had realized there was a gap and gone digging into it. - Some people have a drug addiction-like condition that makes them feel better about putting rootless knowledge into their mouths. - The state of feeling good about spouting out knowledge that has no real benefit is like being drunk on the street, an embarrassment that is hard to bear to look at. - It can be divided into three levels - Rootless condition - They’re saying things like, “Strong AI…” - Can you use that knowledge to have some beneficial effect on your work?” What happens when you ask - For the most part, NO. - State that can be made using the library - The “what you get” line is bumpy, not flat. - For example, in machine learning, the “what you get” line is in a state of great excitement with the advent of scikit-learn - So, by pulling the digging direction from what you want to make in that direction, you can reach the mountain and “make it” more efficiently. - State in which the library itself can be created - Like, “I made a Chainer. - In terms of getting what you want to get, which is what we’re focusing on right now, is there any difference between essentially being able to make it using a library and being able to make it using a - Don’t be satisfied with just being able to create with a library, but delve into that library to gain deeper knowledge. - That in-depth knowledge will affect whether or not you can make it when the circumstances around you change or what you want to get out of it. - If the gap between “what you want to get” and “what you can get” is too big, you won’t find a path on how to get there - In-depth knowledge makes it easier to find a path.

• Twitter
  • Whether the knowledge is Rootless Knowledge,
    • Learning stacked type and digging down to the bottom of a matter,
    • percolation 、
  • Coupling between the space of resources available for procurement and the space of resources to be obtained

---

This page is auto-translated from /nishio/メモ20180126 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.