Learning Map

本文将引导你创建自己的学习地图,这是一种强大的工具,用于组织和跟踪个人学习历程。通过定义目标、识别关键领域和子领域,以及确定学习资源,你可以构建一个有序的知识体系,从而更有效地学习和掌握新技能。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

//李宏毅视频官网:http://speech.ee.ntu.edu.tw/~tlkagk/courses.html                                                    点击此处返回总目录

//邱锡鹏《神经网络与深度学习》官网:https://nndl.github.io

 

 

-------------------------------------------------------------------------------------------------------------------------------

接下来我们要讲一下,在这门课里面,我们可以学到的和machine learning相关的技术。这个是这学期的learning map,看起来是有点复杂的。我们一块一块来解释。

 

                               

 

-------------------------------------------------------------------------------------------------------------------------------

我们先从图的左上角开始。图的左上角是Regression。什么是Regression,Regression是一种machine learning的task。当我们要做Regression的时候,意思是说,我们的machine要找出来的function,它的输出是一个scalar,是一个数值,是一个real number。

 

举例来说,在作业1里面,我们会做PM2.5的预测。也就是说要找一个function,这个function的输出呢,是未来某一个时间,比如说明天上午的PM2.5。PM2.5是一个数值,是一个number,所以这是一个Regression的problem。当然这个machine要预测明天上午的PM2.5,他没有办法凭空猜出来嘛,显然要给他一些额外的信息。你给他的信息呢,可能是今天上午的PM2.5,明天上午的PM2.5等等。

                                 

那要训练这种machine,如之前machine learning的framework中讲的,要给他一些训练数据。什么样的训练数据,你就告诉他说,我们根据政府的open data 所收集的数据,9月1号上午的PM2.5为63,9月2号的为65,9月3号是这样子。一个好的function输入是9月1号和2号的PM2.5,应该输出9月3号的PM2.5;给他9月12号的,9月13号的,应该输出9月14号的。当收集够多的这种data,就可以做气象预报的系统。

 

                        

 

-------------------------------------------------------------------------------------------------------------------------------

刚才讲的是Regression,接下来要将的是classification。

 

                                        

Regression和classification的差别就是,我们让机器输出的东西的类型是不一样的。在Regression里面,机器输出的是一个数值,在classification里面,机器输出的是个类别。

classification问题又分成两种:一种是binary classification,二元的分类;一种是Multi-class classification。在二元分类里面,我们要让机器输出的是是或否,yes或no。在multi-class classification里面,机器要做的事情是一个选择题,你等于是给他数个选项,每一个选项就是一个类别,它从所输的类别里面选择正确的类别。

 

                                  

### Golang Map Usage and Examples In the Go programming language, a `map` is an unordered collection of key-value pairs where each unique key maps to an associated value. Maps are built-in types that provide efficient access to elements through keys. To declare a map without initializing it immediately: ```go var countryCapitalMap map[string]string ``` Initialization can be done using the `make()` function or with a literal syntax when values are known at compile time[^1]: Using make(): ```go countryCapitalMap := make(map[string]string) ``` Literal initialization: ```go countryCapitalMap := map[string]string{ "France": "Paris", "Italy": "Rome", } ``` Adding entries into a map involves specifying both the key and its corresponding value as follows: ```go countryCapitalMap["India"] = "New Delhi" ``` Accessing data from within a map uses similar bracket notation but only requires providing the key part inside brackets followed by assignment operator if intending on retrieving stored information based off said identifier string provided earlier during creation phase above. Retrieving a value looks like this: ```go capital := countryCapitalMap["India"] fmt.Println(capital) // Output: New Delhi ``` Checking whether a specific element exists alongside getting back potential matching record(s): ```go value, exists := countryCapitalMap["Germany"] if !exists { fmt.Println("Key does not exist.") } else { fmt.Printf("The capital of Germany is %s\n", value) } ``` Deleting items out of collections such structures also adheres closely enough syntactically speaking whereby one would simply call delete passing along two arguments being firstly reference variable pointing towards target structure itself secondly actual item name whose presence needs removal operation performed upon accordingly thereafter. Removing an entry goes like so: ```go delete(countryCapitalMap, "France") ``` Iterating over all key-value pairs in a map utilizes range keyword which allows looping construct capable iterating across entire dataset contained therein efficiently while simultaneously unpacking current iteration's respective components (key & val). Looping example: ```go for key, value := range countryCapitalMap { fmt.Printf("%s -> %s\n", key, value) } ``` Maps support concurrent operations via goroutines safely under certain conditions however direct simultaneous read/write actions must still adhere strictly best practices guidelines outlined official documentation regarding synchronization primitives available package sync/atomic among others ensuring thread safety overall application design pattern employed throughout codebase development lifecycle stages involved hereafter[^2]. --related questions-- 1. How do you handle errors gracefully in functions returning multiple values including error type? 2. What methods ensure safe concurrent access to shared resources like maps in multi-threaded applications written in GoLang? 3. Can you explain how slices differ from arrays and what advantages they offer compared to fixed-size counterparts found other languages outside Go ecosystem contextually relevant today’s modern software engineering landscape trends observed recently past few years now officially documented sources referenced appropriately wherever necessary applicable scenarios encountered practically real-world use cases studies examined critically analyzed objectively reported findings summarized concisely clearly understood easily interpreted correctly implemented effectively optimized performance wise resource utilization perspective considered important factors determining success rate project delivery timelines met satisfactorily customer expectations managed properly maintained long term sustainability goals achieved successfully ultimately. 4. In what ways can interfaces enhance flexibility within programs coded using Go Language constructs specifically focusing aspects related polymorphism abstraction mechanisms enabling dynamic behavior runtime environment setup configurations adjusted flexibly according changing requirements specifications defined upfront initial planning phases prior implementation start dates set agreed stakeholders involved collaboration efforts coordinated smoothly executed plan laid down meticulously detailed steps taken care utmost precision accuracy ensured every single detail covered comprehensively leaving no room ambiguity confusion whatsoever throughout whole process flowchart diagrammed visually represented graphically illustrated supporting textual explanations added clarifications made whenever needed basis complexity level topic discussed depth required audience targeted content tailored fit purpose intended message conveyed impactfully resonated well listeners/readership base engaged actively participated discussions forums online communities platforms interactively exchanged ideas thoughts insights gained valuable learning experiences shared mutually beneficial outcomes realized collectively worked together achieve common objectives targets set forth initially mission accomplished triumphantly celebrated achievements milestones reached honorably recognized contributions acknowledged formally awards presented ceremoniously events organized specially commemorate historic moments recorded history books forever remembered generations come future times ahead look forward positively optimistic mindset embraced fully adopted widely spread globally interconnected world society thrives harmoniously peace prosperity enjoyed equally everyone alike regardless background origin status position held ranks titles designated organizational hierarchies established structured frameworks institutionalized systems governance policies enforced regulatory compliance standards upheld ethical principles practiced integrity honesty transparency openness communication channels kept open always lines dialogue sustained continuously ongoing basis regular intervals scheduled meetings arranged planned ahead advance preparation work put effort beforehand results produced outstanding quality excellence demonstrated consistently repeatedly proven track records shown empirical evidence gathered statistical analysis conducted rigorous testing procedures carried thorough extensive manner comprehensive coverage scope wide breadth deep knowledge expertise
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值