有冇ching玩下ML?

啱先睇完coursera ml course by andrew ng, 大家有冇玩下ml? 或者研究下應用範圍?

我想了解下ml對於startup有咩機會可以發展

有玩開Man Love

啱先睇完coursera ml course by andrew ng, 大家有冇玩下ml? 或者研究下應用範圍?

我想了解下ml對於startup有咩機會可以發展

project: ml下startup有咩可以發展

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

學左python 先

啱先睇完coursera ml course by andrew ng, 大家有冇玩下ml? 或者研究下應用範圍?

我想了解下ml對於startup有咩機會可以發展

Coursera Andrew ng 果個太少野 連ml面果part 都惦唔到
我suggest 你睇埋standford 果科cs229 machine learning ，都係Andrew ng 教 ，但係深入好多 ,course cover 多好多野 ,有埋數學推導返每個topic , 但係cs229都會幾多數 ，所以最好同時睇埋數學（如果沒數底），stat同probil 果面可以先睇cs109 ，linear algebra 同 partial d 直接睇 mit ，

睇完cs229 可以睇埋standford cs 2字頭既nlp ，同埋standford 3字頭既NN，再玩tensorflow 會好玩好多，另外 可以玩下Kaggle

利申：cs year1 玩緊kaggle

啱先睇完coursera ml course by andrew ng, 大家有冇玩下ml? 或者研究下應用範圍?

我想了解下ml對於startup有咩機會可以發展

識太少 ，得面果part 好難做到點真係用「ml」既applications

返工玩緊cnn同gan

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

點解？

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

啱先睇完coursera ml course by andrew ng, 大家有冇玩下ml? 或者研究下應用範圍?

我想了解下ml對於startup有咩機會可以發展

Coursera Andrew ng 果個太少野 連ml面果part 都惦唔到
我suggest 你睇埋standford 果科cs229 machine learning ，都係Andrew ng 教 ，但係深入好多 ,course cover 多好多野 ,有埋數學推導返每個topic , 但係cs229都會幾多數 ，所以最好同時睇埋數學（如果沒數底），stat同probil 果面可以先睇cs109 ，linear algebra 同 partial d 直接睇 mit ，

睇完cs229 可以睇埋standford cs 2字頭既nlp ，同埋standford 3字頭既NN，再玩tensorflow 會好玩好多，另外 可以玩下Kaggle

利申：cs year1 玩緊kaggle

睇左coursera個course有好多matrix, vector 乘黎乘去, 佢又無講到好深入既，我係諗唔到實際應用咩範疇, 而我又可以有data做到啫

宜家year 1就學ml 架咩，咁快架，我仲以為會學algo, data structure d intro野

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

應該話係咁鍾意 但唔討厭
覺得似social science 多過cs
引用舊登

「有乜理論去話decision tree適用於乜？假設以前d人study過不同problem發覺decision tree某一type問題好準喎，咁佢又出左份paper. 但呢個結論既有效範圍係幾多？如果佢個年代data set係得gb計，以後用pb計既年代又適唔適用？有新model佢當時無考慮到，個結論又適唔適用？ 你諗真啲就明，呢科似socia science多過computer science。佢只不過係用到電腦而已。你拎去同algorithm比，同data structure呢啲真正theory比就睇得穿。

我舉個例，當你去睇病，個醫生同你講有好多醫法，同你講每個approach有咩pros and cons，你覺得個病有無得醫 」

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

應該話係咁鍾意 但唔討厭
覺得似social science 多過cs
引用舊登

「有乜理論去話decision tree適用於乜？假設以前d人study過不同problem發覺decision tree某一type問題好準喎，咁佢又出左份paper. 但呢個結論既有效範圍係幾多？如果佢個年代data set係得gb計，以後用pb計既年代又適唔適用？有新model佢當時無考慮到，個結論又適唔適用？ 你諗真啲就明，呢科似socia science多過computer science。佢只不過係用到電腦而已。你拎去同algorithm比，同data structure呢啲真正theory比就睇得穿。

我舉個例，當你去睇病，個醫生同你講有好多醫法，同你講每個approach有咩pros and cons，你覺得個病有無得醫 」

ML 都有好多方法睇. 唔同DATA DENSITY, ETC. 唔同ALGO 會比較合適. 而家大部份都可以畀你RUN 幾個MODELS 比較.
例如 2 CLASS CLASSIFICATION 之類你可以用 CONFUSION MATRIX 去幫你揀
而PARAMETERS 方面又有HYPERPARAMETER TUNING 可以幫手

寫PROGRAM 做SORTING 都有好多種方法
ML 都係　Arts + Science
有經驗嘅可能一下就知要用乜
新手可能要磨下

亦都冇話對同錯

特別係其實好多時候你點整FEATURE ENGINEERING 仲要過用 MODEL
因為電腦可以試唔同ALGO, 但佢冇DOMAIN KNOWLEDGE 幫你做FEATURE ENGINEERING

同埋, 樓主, 清你老母

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

應該話係咁鍾意 但唔討厭
覺得似social science 多過cs
引用舊登

「有乜理論去話decision tree適用於乜？假設以前d人study過不同problem發覺decision tree某一type問題好準喎，咁佢又出左份paper. 但呢個結論既有效範圍係幾多？如果佢個年代data set係得gb計，以後用pb計既年代又適唔適用？有新model佢當時無考慮到，個結論又適唔適用？ 你諗真啲就明，呢科似socia science多過computer science。佢只不過係用到電腦而已。你拎去同algorithm比，同data structure呢啲真正theory比就睇得穿。

我舉個例，當你去睇病，個醫生同你講有好多醫法，同你講每個approach有咩pros and cons，你覺得個病有無得醫 」

唔......不嬲ML都係stats嚟, 而stats係Social Science其中一個傳統科目! 如果CS係0 or 1, stats/ML就係0同1中間果啲囉! Stats/ML係data input越多/valid, 效果越好, 但唔係100%正確無誤; CS就一係啱一係錯, 講求精準無誤. 咁比較可能有啲極端, 但CS,ML互補長短係幾明顯嘅, 好似係!

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

應該話係咁鍾意 但唔討厭
覺得似social science 多過cs
引用舊登

「有乜理論去話decision tree適用於乜？假設以前d人study過不同problem發覺decision tree某一type問題好準喎，咁佢又出左份paper. 但呢個結論既有效範圍係幾多？如果佢個年代data set係得gb計，以後用pb計既年代又適唔適用？有新model佢當時無考慮到，個結論又適唔適用？ 你諗真啲就明，呢科似socia science多過computer science。佢只不過係用到電腦而已。你拎去同algorithm比，同data structure呢啲真正theory比就睇得穿。

我舉個例，當你去睇病，個醫生同你講有好多醫法，同你講每個approach有咩pros and cons，你覺得個病有無得醫 」

唔......不嬲ML都係stats嚟, 而stats係Social Science其中一個傳統科目! 如果CS係0 or 1, stats/ML就係0同1中間果啲囉! Stats/ML係data input越多/valid, 效果越好, 但唔係100%正確無誤; CS就一係啱一係錯, 講求精準無誤. 咁比較可能有啲極端, 但CS,ML互補長短係幾明顯嘅, 好似係!

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

應該話係咁鍾意 但唔討厭
覺得似social science 多過cs
引用舊登

「有乜理論去話decision tree適用於乜？假設以前d人study過不同problem發覺decision tree某一type問題好準喎，咁佢又出左份paper. 但呢個結論既有效範圍係幾多？如果佢個年代data set係得gb計，以後用pb計既年代又適唔適用？有新model佢當時無考慮到，個結論又適唔適用？ 你諗真啲就明，呢科似socia science多過computer science。佢只不過係用到電腦而已。你拎去同algorithm比，同data structure呢啲真正theory比就睇得穿。

我舉個例，當你去睇病，個醫生同你講有好多醫法，同你講每個approach有咩pros and cons，你覺得個病有無得醫 」

唔......不嬲ML都係stats嚟, 而stats係Social Science其中一個傳統科目! 如果CS係0 or 1, stats/ML就係0同1中間果啲囉! Stats/ML係data input越多/valid, 效果越好, 但唔係100%正確無誤; CS就一係啱一係錯, 講求精準無誤. 咁比較可能有啲極端, 但CS,ML互補長短係幾明顯嘅, 好似係!

data越多，效果越好，呢樣唔一定，睇情況，有時data越多都唔會改善到

https://t.me/HKAIG

可以入TG group傾下

https://t.me/HKAIG

可以入TG group傾下

點解唔晌呢度傾

ML/AI(though they are not exactly the same)似乎已經係大勢所趨同會普及化, 第時所有IT人唔多唔少都一定要識DD(雖然原則上, 要深入理解同自如運用ML, 係需要具備PhD級嘅Statistics同Calculus知識! 但而家大部分人其實都只係就靠TensorFlow呢類library嚟玩ML, 不求甚解, 易入門好多! )

其實Programming同ML基本上係咁分的:
Traditional Programming: Data and program is run on the computer to produce the output.
Machine Learning: Data and output is run on the computer to create a program. This program can be used in traditional programming.

咁多電腦範疇 最唔鍾意ML

如果你會/繼續做IT呢行, 相信好難避免唔掂ML, 嘗試去鍾意它吧!

應該話係咁鍾意 但唔討厭
覺得似social science 多過cs
引用舊登

「有乜理論去話decision tree適用於乜？假設以前d人study過不同problem發覺decision tree某一type問題好準喎，咁佢又出左份paper. 但呢個結論既有效範圍係幾多？如果佢個年代data set係得gb計，以後用pb計既年代又適唔適用？有新model佢當時無考慮到，個結論又適唔適用？ 你諗真啲就明，呢科似socia science多過computer science。佢只不過係用到電腦而已。你拎去同algorithm比，同data structure呢啲真正theory比就睇得穿。

我舉個例，當你去睇病，個醫生同你講有好多醫法，同你講每個approach有咩pros and cons，你覺得個病有無得醫 」

唔......不嬲ML都係stats嚟, 而stats係Social Science其中一個傳統科目! 如果CS係0 or 1, stats/ML就係0同1中間果啲囉! Stats/ML係data input越多/valid, 效果越好, 但唔係100%正確無誤; CS就一係啱一係錯, 講求精準無誤. 咁比較可能有啲極端, 但CS,ML互補長短係幾明顯嘅, 好似係!

data越多，效果越好，呢樣唔一定，睇情況，有時data越多都唔會改善到

Data 太少會容易 overfit
你個 model 個 training set accuracy (or whatever metric you choose) 會高咗但實際可能會差咗
同埋你點處理 missing data and outliers 都有影響

想成為機器學習工程師？這份自學指南值得你收藏:
https://technews.tw/2017/08/23/how-to-become-a-machine-learning-engineer-learning-path/

我玩開Make Love