What's the Machine learn?
01.What's the learning?
what's the kill?
improve some performance measure
提高一些績(jī)效
從Data出發(fā)柠逞，經(jīng)過機(jī)器的學(xué)習(xí)焙蹭，得到技能的加強(qiáng)
Why use machine learning?
of example
這張圖里面是什么冰寻？
如何定義樹？如何讓程序識(shí)別樹剃毒？
你是這么認(rèn)識(shí)樹的？
不是你父母告訴你特征而是你的觀察
而是你看來很多樹(●'?'●)
你眼睛的觀察
1.exists some 'underlying pattern' to be learned
--so 'performance measure' can be improved
2.but no programmable(easy)definition
--so 'ML' is needed
3.somehow there is data about the pattern
--so ML has some 'inputs' to learn from
看看冰山一角ML的應(yīng)用
01.Food(Sadilek et al.2013)
data:Twitter data(words+location)
skill:tell food poisoning likeliness of restaurant properly
2.Clothing(Abu-Mostafa,2012)
data:sales figures + client surveys
skill:give good fashion recommendations to clients
3.Housing(Tsansa and Xifara,2012)
data:characteristics of buildings and their energy load
skill:predict energy load of other buildings clousely
4.Transportation(Stallkamp et al,2012)
data:some traffic sign images and meanings
skill:recognize traffic signs accuratelyML is everywhere!
A Possible ML Solution
answer correctly ≈ [recent strngth of student > difficulty of question]
1.give ML 9 million records form 3000 students
2. ML determines (reverse-engineers)strength and difficulty automatically
電影推薦系統(tǒng)構(gòu)想
特征
機(jī)器學(xué)習(xí)深入：
example:
用戶信用評(píng)估&行用卡發(fā)行：
data:
data
Basic Notations
ML輸入X
ML輸出Y
目標(biāo)函數(shù)F
data
Data <=> training examples: D={(x1,x1),(x2,y2).....(Xn,Yn)}
(historical records in bank)
hypothesis
G:x->y
