什么是动态规划 in 一般?

Dynamic programming is a special approach to problem-solving. Unfortunately, there is no single definition of dynamic programming, but let's try to form one by ourselves. The idea is that the optimal solution can often be found by considering all possible ways of solving an issue and choosing among them the best one. The operation of dynamic programming is very similar to recursion with memoization of intermediate solutions, which is also called memoization. Recursive algorithms tend to divide a large concern into smaller subtasks and solve them. Dynamic algorithms divide the difficulty into chunks and compute them one at a time, building up solutions step by step. Therefore, dynamic algorithms can be thought of as recursion, working from the bottom up. It's about Richard Bellman, who invented and established the concept of “dynamic programming” in the scientific community. In 1940, he used the term for issues where the solution to one part of the problem depended on another. Then in 1953, Bellman modified and expanded the definition of dynamic programming to its current form. He describes the naming situation of this concept in an amusing way. Bellman had to come up with a name, but the official representative to whom he reported at the time strongly hated (the mathematician emphasizes that he “hated”) the terms and did not want to deal with them. Every time, all the employees had to literally tremble with fear because their boss couldn't understand something and stirred it with mud in his unfunny puns. And especially the math terms! According to the book “Dynamic Programming”, Richard Bellman spent a lot of time and effort coming up with a name. The word “programming” was chosen as an analog to the word “planning”, which did not fit for a number of different reasons (the Soviets had planned something all the time). And the word “dynamic” was decided based on the fact that, in addition to conveying the essence of the approach, it was hard to come up with something derogatory, or preparatory with it. Bellman didn't want an executive to somehow obfuscate his term. Not even an official could have done it so easily (readers, of course, will find a couple of variations). This is how the term “dynamic programming” was formed. Summing up, dynamic programming is an approach to solving problems, where the original problem is broken down into smaller subtasks that are easier to solve. And then smaller solutions of these subtasks can be used to fix the original error. Nevertheless, the approach has some essential features: It makes sense if the solutions of subtasks can be used effectively (to memorize, for example); subtasks have a common structure, so some homogeneous methods can be applied to their solution, instead of solving each separately with different algorithms.

How to Apply Dynamic Programming Algorithms to Solve Problems

The demonstration subject will be a classical dynamic programming concern—Levenshtein Distance. Despite the seemingly complex name, in fact, its difficulty is in the transformation of one word into another by adding, deleting and replacing letters with a minimum number of operations. This concern of dynamic programming can be formulated as follows: find the minimum “distance” between two words. The distance in this case will be the minimum number of operations to be applied to the first word to get the second (or vice versa). We have three available operations: insert – add one letter anywhere in the word, including its beginning and ending delete – remove one letter from any place in the word replace – replace one letter in a certain place with another 所有 these operations have an equal cost: +1 to the distance between words. Take, for example, two simple words, MONEY and MONKEY. What is the minimum number of operations needed to turn MONEY into MONKEY? The shrewd human eye quickly guesses one: add a K between the third and the fourth letter. Let's take a more complicated case. Try turning SUNDAY into SATURDAY, and we see that the number of combinations of dynamic programming to search is potentially gigantic. If we solve the problem by brute force, we can consider all possible combinations, like in the password-cracking example. Instead of a possible 94 candidate characters, we have three operations – insert, delete, and replace. Three combinations for the first letter, 3×3 for two letters, 3×3×3 for three letters, and 3^N for N letters. Memoization and tabulation Dynamic programming memoization is an optimization technique that allows you to memorize the results and all algorithms of calculations and reuse them when you need to repeat this calculation. A favorite example: when calculating the seventh Fibonacci number, you can memorize the fourth Fibonacci number and not calculate it every time, but only once. This technique is also described as top-down. Tabulation is an optimization technique that starts by solving subtasks with the simplest one and then, as you progress, deals with more and more complex subtasks until the main problem is solved. In doing so, solutions to simpler subtasks are used to solve more complex subtasks. In contrast to memoization, this approach is called “bottom-up” because you first take on the simplest tasks.

LITSLINK博客

AI白皮书

数据科学趋势

立即下载

加载更多帖子

饼干	持续时间	描述
cookielawinfo-checkbox-analytics	11个月	此cookie由GDPR cookie同意插件设置. cookie用于存储用户对“分析”类别中的cookie的同意。.
cookielawinfo-checkbox-functional	11个月	cookie是由GDPR cookie consent设置的，用于记录用户对“功能”类别中的cookie的同意。.
cookielawinfo-checkbox-necessary	11个月	此cookie由GDPR cookie同意插件设置. cookie用于存储用户对“必要”类别中的cookie的同意。.
cookielawinfo-checkbox-others	11个月	此cookie由GDPR cookie同意插件设置. cookie用于存储用户对“其他”类别中的cookie的同意.
cookielawinfo-checkbox-performance	11个月	此cookie由GDPR cookie同意插件设置. cookie用于存储用户对“性能”类别中的cookie的同意。.
viewed_cookie_policy	11个月	cookie由GDPR cookie同意插件设置，用于存储用户是否同意使用cookie. 它不储存任何个人资料.

LITSLINK博客

AI白皮书

什么是动态规划? 特点、技术 & 例子

人工智能挑战- 6个步骤，无痛开发和实施

AI和ML有什么新进展? 10项正在发展的技术

深入研究人工智能虚拟代理:为什么它们今天成为焦点?

智能自动化:中小企业和企业应该期待的真正价值

计划、控制 & 专题报告:如何建立有效的仓库管理系统

使用AI进行Web开发的利弊:你必须知道的绊脚石

什么是POC? 通过概念验证解决技术问题，构建理想的软件

构建健壮应用程序的10种敏捷软件开发方法和框架

智能和强大- 3最先进的人工智能系统概述

订阅2020十大正规网赌网址的博客更新

LITSLINK博客

AI白皮书

类别

订阅2020十大正规网赌网址的博客更新