Today we will talk about using Monte Carlo tree search for risk assessment.
First we need to understand his basic principles:
From the above figure, we can know that the Monte Carlo tree search is divided into five steps:
Selection: From the root node R down, use recursion to select the optimal node to a certain node L.
Expansion: If a new action A needs to be performed, a child node needs to be created under node L, and one of the child nodes C is selected for expansion.
Simulation: Start at node C and proceed randomly down until the final result is found.
Backpropagation: Record the result of this iteration and score the new child nodes.
Repeat steps 1 to 4, and finally select the node with the highest score for each item, which is the optimal solution we need.
The advantage of Monte Carlo Tree Search is that...
Stop, stop, we're not here to learn algorithms!
We engineers often make a mistake: when he wants to introduce how useful a technology is, what benefits it can bring, and what benefits, he starts to introduce the technology itself.
In the end, the engineer couldn't understand: "Why did I explain it so clearly, but he still doesn't understand what I'm talking about?"
Because you may not have noticed: when you understand a technical principle, you automatically eliminate the memory when you don't understand it.
Why is it tiring to teach children?
Because what you think is common sense, he doesn't understand. You look at this issue from the perspective of "understanding", while he is from the perspective of "doesn't understand". And there is a deep gulf between them.
Although the adults who communicate with you are adults, in the field of technology, they are only newbies, and they are still looking at the problem from the perspective of "don't understand". Dry technical explanations will not make your communication effective.
So, how can we make others "understand"?
We need to do the following four things:
State the conclusion, state the utility
Use examples to allow for deviations
Less jargon, more graphics
Communicate with each other and enhance feedback
1. State the conclusion and explain the utility
When eating at a restaurant, you don't care how skilled the chef is at cutting vegetables, but how good he is at cooking.
When buying a perfume, you don't care what the ingredients are in it, but whether it smells good or not.
When picking a mobile phone, you don't care about the craftsmanship inside, but whether it works well or not.
Therefore, when you introduce a technology, it is far more important to clarify what is useful than to talk about the technology itself.
If you've read other articles I've written, you'll find Latest Mailing Database that my articles have clearly bolded keywords .
Just like the table of contents of a book, the role of these keywords is that even if you don't read the article itself, you can roughly estimate the content of the article from it.
If you find a section that interests you, you can read selectively. This greatly reduces reading time and effort.
Similar to reading, if you want to make your ideas interesting to others in a limited time, you first need to state the conclusion and utility .
Technology is used to solve problems. Most people are not interested in the technology itself, they come to know the technology out of a "technological realization" attitude. Your technical introduction is used to prove the feasibility of the technology, not to carry out knowledge popularization.
We need to capture the core needs of our audience:
If he's the boss, tell him about the increase in efficiency.
If he's the manager, tell him about the time savings.
If he is the customer, tell him the advantages of the product.
2. Use examples to allow deviations
In many cases, a seemingly profound problem can be easily explained using a common example.
For example, in junior high school textbooks, when we explain electric current, it is difficult for students to imagine how electric current flows in wires.
But if current is analogous to water flow, voltage is analogous to water pressure, and resistance is analogous to flow resistance, then the basic principle of current can be clearly and intuitively explained.
Think this is too easy?
Let's explain the blockchain.
As normal, you need to explain about public ledgers, distributed ledger, proof-of-stake, etc.
If you want to go deeper, you will explain SHA-256, block storage units, Merkle trees, etc.
And I'll use a very simple example:
Let's say we have a small village where everyone is worried that everyone else is the bad guy. So how can we sell Zhang Sanjia's cabbage to Li Si under such circumstances? We stipulate that there is a mailbox at the door of each house. The mail of this mailbox can be read, but the content cannot be changed. When Zhang San wants to sell cabbage to Li Si, they will send the matter of "selling cabbage" to everyone's mailbox, and then everyone will reply "I know". Then, Zhang San can sell this cabbage to Li Si. Because everyone knows about "Zhang San sold cabbage to Li Si", then Zhang San can't sell the same cabbage to other people. That mailbox is the blockchain.