A.I 是怎樣煉成的？

Geoffrey似我 2018-11-27 11:30:04

我UST CS grad，都幾多嘢學完無用
data sci多d maths
如果俾我揀我會揀data sci

Data_狗 2018-11-27 13:17:55

UBER AI LAB 出咗新AI Algorithm(Go Explore), 透過重返有趣嘅State再explore 去解決sparse reward problem
http://eng.uber.com/go-explore/?fbclid=IwAR3411XVuU_r7Jrjs2kSLGYLfUnRa3qTceuagxobkzl3SZi0Li1HFpWp9qE

我沒有放棄 2018-11-27 13:23:42

讀Math + Stat再自學Python/ML

雷神雷利 2018-11-27 23:14:20

btw樓主知唔知ust個msc in data driven modeling係咩programme黎

係唔係新開架？

Shitlatan 2018-11-27 23:48:12

stat undergrad 同ai 其實冇乜關係

我沒有放棄 2018-11-28 01:58:06

Deep Neural Networks只係多層嘅logistic regression (GLMs)咋，點會冇關係

Shitlatan 2018-11-28 09:09:12

Undergrad 喎

我沒有放棄 2018-11-28 09:16:49

Undergrad唔識logistic regression?

Geoffrey似我 2018-11-28 09:30:41

我都唔知..
係今年先開

忒修斯之船 2018-11-28 09:35:27

先唔講而家啲學校通常都有開ml related course
你唔係覺得
pca
lda qda
smoothing methods
cross validation
model selection
...
全部都冇關係？

雷神雷利 2018-11-28 11:08:39

個programme由department of physics host係咩玩法

ASA 2018-11-29 18:34:17

推薦一本書，呢本幾好睇
http://www-bcf.usc.edu/~gareth/ISL/

雷神雷利 2018-11-29 22:14:29

ching識貨

ASA 2018-11-29 23:35:26

其實我識好少，個陣grad前上過堂玩過下credit scoring 咁。呢本我自己都幾鍾意，啲數唔太abstract 幾好睇

kTln2 2018-11-30 00:27:45

咁搞笑？不過現實係好多物理人轉行做DS，連中大物理都有Data science stream

忒修斯之船 2018-11-30 00:31:30

有咩科係冇乜人想轉過ds

kTln2 2018-11-30 00:39:57

Medic (好似係）

不過物理人真係好多，宜家見到d resume 基本上有一半都係

Geoffrey似我 2018-11-30 18:31:16

我身邊兩個phy major frd都做DS

RX-78-2 2018-12-01 08:21:29

因為modeling skills同實際application 好似physics

利申: physics 撚

我沒有放棄 2018-12-01 09:34:05

想學machine learning可以睇我之前開個post

我沒有放棄 2018-12-01 09:34:23

https://lihkg.com/thread/624052/page/1

ASA 2018-12-01 10:56:23

多謝哂先巴打，啱啱睇左你個post

想問下有邊本你覺得啱我睇？我係stat &optimization 加少量applied maths background

我沒有放棄 2018-12-01 10:58:45

All
我係math + stat + optimization background

ASA 2018-12-01 11:15:46

其實我想問有無書係for classification(fuzzy string match), 因為我最近寫左個python code 用cosine similarity 做，但就只係將啲string vectorize 再計similarity.
詳情係呢度https://bergvca.github.io/2017/10/14/super-fast-string-matching.html