Everything you need to know about Machine Learning in 30 minutes or less

http://www.hilarymason.com/presentations-2/devs-love-bacon-everything-you-need-to-know-about-machine-learning-in-30-minutes-or-less/

Core Problem: "How can we build computer systems that automatically improve with experience, and what are the fundamental laws that govern all learning processes?"

Resources:
Data
https://bitly.com/bundles/hmason/1

Video
http://shop.oreilly.com/product/0636920017493.do

ML intro

Data vs. RDF

http://static.googleusercontent.com/external_content/untrusted_dlcp/research.google.com/en/us/pubs/archive/35179.pdf

- Related to graph
- Make me think of the link data session on ODW 2013 yesterday, that's the reason why RDF's approach won't work on real world: too dynamics structure for too complex data.

The Science of Insecurity

- I like the approach: to determined what is safe (regex and context-free)
- Halting problem: "Given a description of an arbitrary computer program, decide whether the program finishes running or continues to run forever".
- html5 and css3 (w/o js) is turing complete!

The Great Porn Experiment: Gary Wilson at TEDxGlasgow

http://www.theguardian.com/world/2013/oct/20/young-people-japan-stopped-having-sex

Hilary Mason: "In Search of the Optimal Hamburger"

- Data from various food site
- Proximity, Price, Quality from community

The first 20 hours to learn anything

- Deconstruct the skill, set specific goal.
- Learn just enough to self correct
- Practice at least 20 hours

Mark Zuckerberg at Startup School

- Not that it's difficult to do, but that you believed it very strongly.

Charlie Todd: The shared experience of absurdity

- I think we need that.

TED: Mike Rowe: Learning from dirty jobs

Peripeteia: realization and tragedy.

The Future of Programming

http://worrydream.com/dbx/

- Technology moves quickly, but people's mind moves slowly
- Furious at students wasting computing time on assembly, that could have been doing manually

coding -> direct manipulation of data "direct manipulation vs. markup language"
procedure -> goals and constrains "expressing the programm on what you want vs. how"
- Programming goal. Different program that can communicate with each other; they will figure out procedure that can achieve the high level goal vs. API (1. have to know each other 2. dependent 3. human that doing doing low level work)

text dump -> spatial representation
- Smalltalk: representing text spatially
- Interactive design


sequential (actor model) -> concurrent (threads and locks)

Actor model: every program is doing its own thing, interact with people nearby. (Lisp?)

The real tragedy, is that people forget that there can be more way of programming.

Why all these good idea happened there? Because nobody knew what programing was, they just tried everything.

The most dangerous thing for a programming person, is to "think that they know what they are doing".

 "I don't know what I am doing."

Lastly, here's some advice Alan Kay gave me (as I was going through a small personal crisis as a result of reading Jerome Bruner's "Toward a Theory of Instruction"):
I think the trick with knowledge is to "acquire it, and forget all except the perfume" -- because it is noisy and sometimes drowns out one's own "brain voices". The perfume part is important because it will help find the knowledge again to help get to the destinations the inner urges pick.

Bret Victor - Inventing on Principle

- "Follow your principle"
- injustice, responsibility not only opportunity-> social wrong. It's not just for social activity, but in technology as well. A personal mission. Larry Tesler didn't only invent "ctr+c", or "modes", but he recognize a culture wrong that was invisible in the culture. 

- Fighting for a particular idea. 1. The world try to find you by your skill. 2. Path of the problem solver: certain problem, work it, make contribution. 3. Go after his problem.

It can take time to find your principle. It took me ten years. Experience many things, and see whether it resonates with me. Confine yourself to a single skill, may limit the ability to find the principle.

A principle should be clear, to dictate right from wrong. 

- Creator needs an immediate connection
- So much of creation is discovery, if you can see the immediate connection, you can't create it, trying it immediately
- Creator needs to see what they are doing, been able to try idea as you think about.

- If we are writing program on a computer, why do we need to simulate computer in our head?

- I always think about million of pieces lock in million of heads.

Sergey Brin: Why Google Glass?

- When you need information, you don't need to search for it, it just come to you.
? It implies that we can connect with the real world more. Is it true?

Louis CK honors George Carlin

- Spent 15 years telling the same joke
- Throw it out, start all over again, go deeper

- Get it out there.

Keller Rinaudo: A mini robot -- powered by your phone

- Great high level programming add-ons to make robot more personalized!
- Muyueh:Looking forward for a romo that fly 

- USD 150
- Romo
- Follow motion, and move around

- Completely intuitive for people of all age

Katherine Kuchenbecker: The technology of touch

Muyueh: guesses that there is also a high potential in the porn industry.

- Haptics /Simultate touching (Haptic phtography)

force sensor /motion tracking /accelerator / vibration sensor (10 sec)

surface gives vibration feedback

(online shopping)

Dentist use teeth's feeling to detect cavities. Able to learn this before people touch it. Potential in medical training.

Game /Sport to be more realistic.

Ken Jennings: Watson, Jeopardy and me, the obsolete know-it-all

Hard Question
- When information is so easy accessible, is memory/processing that much important?


- "I came to believe that there is two advantage for knowing:"
1. Massive different of facts (information are doubling every 18 months); need to have more basics to know where to search.

Muyueh: isn't this the reason why one should not be memorizing that much information ? The fundamental basic that ones need to manipulate the system (ie: searching on Google; or like the process of problem solving in consulting) is different than the memorize it all approach.

One interesting place where this is still disconnected is education: especially the test system, that so far still consist of testing the processing/memorizing ability of humain. 

There will come a day people should be allowed to use Google in their exam, lookup for problem that has been already been solved. As computer is very bad at framing the problem, making decision and pattern recognition.

2. Advantage of time, little girl that remember the right fact that save people from tsunami - > share connection between people.

Muyueh: this is a moving story, but what about all the people that doesn't recall this fact? Isn't it the case where it shows that people among one thousands people, only one little girl recalled, hence, having people to memorize is in fact a bad strategy that we should not rely on?



It is always better to know a thing, then not to know. Last cultural heritage that people shares.

-> Question where who leads the future. Machine taken over where machine takes over.

Muyueh: false dichotomy,

-> Spirit of Jeopardy: people that are curious about everything, constantly learning.





- He is a programmer beaten by IBM computer, understand the difficulty for AI to underst and Jeopardy (a little line that comes closer and closer)

- "I felt an Detroit worker that is been replaced by robot"

- Middle class they can do taken away vs. People will have more time for leisure

- "Is all our of brain going to shrink?"

Kees Moeliker: How a dead duck changed my life

- Ig Nobel Prize: first make people laugh and then makes them think: ultimate goal to make more people interested in science. (high price fake medicine is more popular than low price price fake medicine)

- Why does it mean? Only dead is instant.

- Bird fight its own image in glass.

Sanjay Dastoor: A skateboard, with a boost

Great HW startup idea.

#OpenDataDay

Open Data 開放資料可以做到什麼

“這個 Chrome 瀏覽器的套件，求職者在 104、yes123、1111 找工作時。公司假如曾違反勞基法，會自動跳出警告視窗警告應徵者，這公司過去的違規事項，或近期內是否放過無薪假”。”把模擬城市的遊戲界面，做在真的地圖上：這樣就可以看到每一點最近的消防局、警察局、廟會。還可以計算每一個地址的幸福指數“。“評律網上，可以找到每一個律師的訴訟勝負比例，進而挑選最適合案情的律師。因為根據我們的調查，台灣超過 7 成的民眾遇到法律糾紛的時候，是依賴朋友推薦，尚無一個可以客觀選擇律師的平台。”

星期天 (2/24) 我去參加了 Open Data Day （國際開放資料日），我在哪邊遇到好些有趣的人，上述的問題，描述了一些他們在試著解決的問題。

近幾個月，從開放政府預算、公車時刻查詢 APP、以及實價登錄網，台灣也開始有些開放資料的有趣應用。 我的朋友問我說，為什麼開放資料現在才崛起？從某方面來說，資料是越來越多了：公車裝上追蹤器，也是這幾年內的事情而已。

但另一方面，有些資訊一直以來都存在著：勞委會一直以來都有記錄違反勞基法的公司，而法院的判決也一直都存在。是什麼原因造成這些應用最近出現？我們可不可以加速這些應用的生成？

目前政府的做法，似乎就是拿這些資料，推出一堆自家的應用程式。（令人懷疑，或許這已經成為公家單位考績的一部份了）。如此產出的應用程式，通常是功能強大，但不大好用。追根究底，這些應用程式多半是以招標的方式承包：在預算限制內，完成特定的功能。功能的數量可以清楚界定，但使用起來是否方便，就難以驗收了。

這背後基本的假設是：公家單位完全清楚這些資料可以開發做什麼用途，就像一位種哈密瓜的農夫知道自己的作物應該怎麼樣來吃。許多年來，我們被告知哈密瓜是一種飯後水果，在吃完後，用湯匙挖著哈密瓜往嘴裡送。

在 Open Data Day，我看到有人把哈密瓜當成前菜，配著煙薰火腿一起享用。看到這些令人驚艷的應用，反過來看當初我們是怎麼吃哈密瓜，反而顯得相當無趣，或許農夫知道之後也感到相當的羞愧，當初怎麼信誓旦旦的介紹這是哈密瓜的所有吃法了。或許根本性的問題，就是農夫根本不應該插手決定哈密瓜有什麼用途，他只管生產哈密瓜就是了。

資訊就是前述的哈密瓜，我們過去仰賴著資擁有者提供我們應用的選擇。但現今，所謂 Open Data 提倡：或許民間可以更有效率的把這些資訊用來解決他們的需求。有時候，是因為解決這些問題需要有多方單位的資料（比方：CityDashBoard 的計劃）；另一些時候，只是因為這些問題需要一個不同的突破方式（比方：前述求職小幫手）。

過去，要處理地理資訊等問題，需要昂貴的專業軟體。公家單位要解決一個問題的時候，就會開一個標案，由幾位台大教授帶領跟他們的學生還有幾家小型的顧問公司競標。

近幾年來，在資料收集、整理、分析、以及視覺呈現上，都多了好些免費工具，包含許多開放、半開放（ Google Map）的基礎建設。這是世界趨勢，台灣從去年底開始逐漸出現了幾場以 Open Data 為主軸的駭客松 (i.e. g0v 資料放領)。許多會寫程式的”非專業人士“進入這一個領域，百花齊放，雖偶爾幾個例子惹人爭議，但總體上還是為了社會帶來許多正向的影響。

台北、高雄市府也已經建立開放資料的平台，呼應這項趨勢。短期內還有些問題必須解決，像是：是否應該要跟民間競爭釋放應用程式？創造自己的開放資料規範或跟國際接軌？還有到底應該用什麼格式釋放這些資料。

窺視 Open Data Day 當天的活動進行，或可得到一些啓發。

當天最我最驚艷的工作，是前面提到， Chrome 瀏覽器上找工作的求職小幫手：把勞委會的資料庫運用在一個找工作的過程中，把求職程序，與資料在一起。該原型，單由 Ronny Wang 所完成（目前求職小幫手雖然還無法網路上找到，只有另外一個相關專案，是把關係企業視覺化）。

由 Tim Hong 所帶領的城市儀表板，則希望可以把更多的資料來源整合到 Google Map，希望可以做成像模擬城市裡面的控制界面。

評律網，已經把上百萬筆的法院的訴訟案件進行整理，打算未來開放以 API 的方式，提供第三方可以製作其他應用。在美國，有一個工作叫做法律助理，專門協助律師處理文件工作。想像原本需要五、六個小時整理某公司的訴訟案件，可以有電腦協助。不過評律網的開發人員也有提到，訴訟文件目前遇到的問題之一，就是錯字很多。

當天跟傳統駭客松不同的。 是除了開發者，難得還有需求者帶著具體的問題到現場：地球公民基金想要解決的是水質監測的問題：特別針對新竹地區的霄裡溪，現階段政府已經有針對該地區的水質進行監測，但目前農委會網站上所公佈的數值過於籠統，僅有“優良-優良”，無判斷細項。

媒合有開發者以及需求者的確是相當好的想法。

林業試驗所的王豫煌，近一年來，他們開始搜集森林裡面的聲音、影像檔案，結合他們原本在森林裡所架設的氣候觀測站，他們希望可以建立一個物候的模型：什麼樣的生物容易在什麼樣的環境下出現。他們遇到幾個問題，一是辨識技術：一個十秒的錄影，會捕捉一個生物在鏡頭出現的蹤跡。但以現今的電腦技術，這類型的模式辨識其實相當困難。這些工作其實都需要人工進行。

當天討論到一個可能的模式，是效法 NASA ：把星空的資料釋放給大眾，用人眼協助標記照片上所出現的東西。 NASA 的專家就標記的內容進一部分析。 或許台灣會有民眾想要在無聊通勤的時候，用手機分析這一個影片中出現的是穿山甲還是猴子？

但在有限經費內，如何假設一個夠大的平台？該單位每個月約可以產生  1 TB 的音訊資料量、100 GB 的影像資料（1 TB 等於 1,000 GB，Gmail 信箱約有 10 GB 的大小）。在有限的經費底下，怎麼樣釋出這些資料是一個一大難題。現場所討論到的一個解決方案，或許是利用第三方影音平台(Youtube)上加值。

關於資料來源問題，現場也有不同的團隊試著從別的切入點下手：將資料的格式統一。政府開始推動開放資料平台，把很多公開資料都釋放出來了，但卻面臨這些資料無法被使用的窘境。有部分原因，是因為他們使用的軟體系統格式，可能是該軟體特有，沒有軟體的人無法打開（想像要用小畫家打開 Photoshop 的 .ai 檔）。這一個問題，在地理資料上特別常見。為公家單位平反：在當初規劃這些軟體的時候，沒有辦法預見十幾年後，會需要把這些付費軟體內的資料轉換為某種特定的開放格式。

這一個團隊在做的，就是拿市政府上的資料進行轉換，然後再公開。這相當層面，為了所有人謀福利。做到底層很繁瑣的格式轉換。

主辦者 CK 自己也帶了兩個小組，一個是整理台灣目前的發展狀況，未接下來要研究核能議題的人留下基礎。另外是呼應全球 Open Data Day 的專案，將 World Bank 釋出來的資料進行視覺化。

林旭宏發起了一項跟保育蛾有關的計劃，不過當天沒有機會聊到。

特別感謝精誠資訊提供場地。

Miguel Nicolelis: A monkey that controls a robot with its thoughts. No, really.

- The sound of brainstorming
- Billion of cell to capture

- 10 years ago, benchmark (monkey) artificial arm and real arm in real time ( < 0.5 sec)
- Monkey uses artificial device to play game
- Also sending back artificial feedback

- NOW: Train animal to control avatar to explore the virtual world (send back virtual world feedback, that brain learn how to control artificial hand

Global Trip takes 20 millisecond less than the monkey brain to physical.

Esther Perel: The secret to desire in a long-term relationship

- Crisis of desire: free choice and identity
- This is the first time over the history, we are trying to experience sexuality in the long term: pleasure and connection

- Two human needs that we wish to reconcile

1. Safety: Permanence 
2. Adventure: Journey and Travel

This is often a crisis of imagination.

Love -> to have
Desire (sex) -> to desire ( to explore, an bridge to cross)

When do you find yourself most drawn to your partner?
Group 1
- When the partner is away./imagine to be together
- When partner is self-sustaining, from a comfortable distance, to a somewhat distance. See my partner in a mystery.

Group 2
No needs (care-taking) in desire, different to wanting them.

Group 3
When there is novelty: what part of thing you show. Sex is not a thing to do, but  a place to go.

Imagination is the center piece of erocticism.

Those who didn't die: worry, insecure, anxious can't experience erotic.
vs.
Those who came back to life: erotic as an antidote for life.

"I turn myself on/off when" vs. "what turn me on":

When I feel old, when I felt low self-esteem
vs.
Desire is a certain part of  selfishness in the presence of other.

Exploratory needs. Need to explore while at the same to be able to come back.


Few things to do:

- A lot of sexual privacy, erotic place
- Foreplay
- An erotic place, is a safe place for imagination
- Moon, cycle vs. fall right from the sky

- Committed love is presence

Mitch Resnick: Let's teach kids to code

- Scratch, programming for kids.
- Kids create their interactive motherday card; ppt w/ animation

- Kids has became fluent w/ the technology.
- Chatting, Gaming different than create new technology ( read but not write)

- The ability to write their own code.
- 100 of websites that teach coding

- Should children learn to code?

- drag and drop programming

 - already connect to the real world/hardware

- use NB camera (kinect)
- Voice control


- The kids are learning code very young!

- Learn to code -> open up many other thing (knows why should use variable)
- exactly to know where to put the right code

- The process of design; how to fix bug, try ideas

-> The future kids are scary. 

They will have the design skills, and an born knowledge of how computer writing.

I only learn how to write and read in a very stage

How simple ideas lead to scientific discoveries - Adam Savage

- Inertia (leads to Nobel Prize)
-Simple question->Scientific discovery

-Earth is round was already know by Aristotle. see sun shadow ( 1% difference of diameter of the world )


- refining other people result.
- great technology; with technology that were available a decade ago.

- Light. use wheel to determine 2% of light speed in 1849.

- Different scientist are starting in the same place as us. More curious more hardworking.

Why the shape of your screen matters - Brian Gervase

4:5 -> 1:3 (movie)

pan scan techniques to eliminate part ( 55% stay)

16:9 (75%)

Questions no one knows the answers to (Full Version) - Chris Anderson

- Question no one knows the answers to. Great list of difficult question.

- No one knows the answer to: How many universe exist?

- Galaxy->grain sand; milky way fill 30 by 30 vy three sands;

- String theory-> countless (?) other universes, build on different properties. 11 dimensions. 10^500 universe

infinite universe.

No agreement on this question 0->infinity


- Why can't we see evidence of aliens life?

- Half of trillion planet in the galaxy

- Other civilization in million of years of advance

- Possible answer:
- A superior species made a radio silence; and ready to eliminate other species.
- Maybe it's fare harder than we think: four billions year

- Just haven't look it up yet ( looking in too short time)
 - Maybe not looking the right way (dark matter?)

- Maybe looking at wrong scale ( smaller scale turn itself microscopic )

- Maybe biological life is just a passing phase. ( interesting)

I like the message. Haven't done fact check yet, but it is a good thing that someone is asking/discussing the difficult question.

Pamela Meyer: How to spot a liar

- Core proposition: Lying is a cooperative act. Someone else has agreed to be lied.
- Deception cost 7% of revenue.

- Henry's rule: "Everyone is ready to give you something, for the thing they are hungry for."
- Lies connect the gap.

- 10 - 200 lies/day

- stranger lies 3 times within 10 mins of meet each other

- Men lies more about themselves than other people; women lies more about other people

- Extravert > Introvert
- Married couple: one lie in ten interactions; not married: one in three interactions

- It will be good if the source are checked when talked (popcorn project)
- Conform to which study; Original Quotes;

- We are against lying, but we cover for it

- Larger the cortex-> More Lies

- One year old baby: conceal truth
- Two year old : bluff
- Nine year old: cover up

- Lies spotters : spot 90% of the time.
- Rest of us : spot 54% of the time.

- Technique to spot lie:
Formal > Informal language
Distance Language

Chatter with fingertips

Freeze upper body
Fake smile (real smile in the eyes)
Over compensate eyes contact

An honest person will cooperative; brainstorming; imply strict punishment; furious about wrong accusation

Deceptive; redrawn; way to much detail; very chronicle details; uncoordinated speech physical gesture

Leak an expression; contempt (you are been dismissed, asymmetrical lips)

Lower vocal tone, feet to the exit (need to appear in group)

Technology to deceive lies.

Challenge: over sharing is not honesty. Signal to outsider to not cooperate on lies.

Wingham Rowan: A new kind of job market

- Problem: make use of flexible labor force. Defined by a last minute of available force.
- Marketplace for rush hours. The terms that they will accept booking.

- Labour, market research people.

? only for interchangeable task; what about more specific task?

-> Modern Market for all
- using machine/algorithm to analyze the opportunity
- as compared to the current listing methodology

- what needs to be done ( the following info. needs to be open)
- ids
- credential
- license

national lottery.

start by outsourcing gov. work. to small business.
-> have more data, that can add value

Tim Berners-Lee: The year open data went worldwide

- Water and White people
- Open data, to rebuild map in Haiti very fast

Arunachalam Muruganantham: 我如何開始一場衛生棉革命

- I have to say above all, that I really thank the translator of putting the subtitle.
- Kind of long at the end.

- Inspiring to build and give away.

David McCandless: The beauty of data visualization

- Gov. Budget
- Fear
- Facebook status: Breakup 10,000

- Relative Data
- 1,000 medical study; 250 visualization; one month to fill a medical graph
http://www.informationisbeautiful.net/play/snake-oil-supplements/

Store in the Google Docs!

地球圖輯隊

A great website on the world news for Taiwan.
http://world.yam.com/

Jeff Bezos: What matters more than your talents

Recommended by Roger.
- My Grandmother bursted into teers.
- One day you will understand that it will be harder to be kind (choice) than smart (gift)

- Will you take proud of your gift, or proud of your choices?

- Choice of starting up
- How will you use your gift?

- We are our choices

The Beauty of Mathematics

http://www.cise.ufl.edu/research/sparse/Harn/Beauty_of_Mathematics.m4v

- equation -> matrix -> relationship map

Ric Elias: 3 things I learned when my plan crashed

- Bad wine collector, because if the opportunity was ready, why not open it?
- Ego, things that you don't care with people that you care
- Death is not feat, but sadness, why can I spend more time as a parent

Inge Druckrey

Cinematographer of Helvetica. 

- See first, otherwise cannot create. 

- Use drawing to see. Metal/Glass.

- Power, strength of basic element.

- Two aspects of writing: 

1. the eye that controls, 

2. the radical hand write expressively, faster and faster.

Learning: 

- Copy until internalized

Symbolize, essence of something.