63. 脳研究のプロジェクト
Blue Brain Project
EU Human Brain Project
NIH Human Connectome Project
US BRAIN Initiative
64. Blue Brain Project
Henry Markram
“I wanted to model the brain because
we didn’t understand it.”
“The best way to figure out how
something works is to try to build it
from scratch.”
87. NN & Deep Learningの
Lecture Note
Geoffrey Hinton
“Neural Networks for Machine Learning”
http://www.cs.toronto.edu/~tijmen/csc321/
http://www.cs.toronto.edu/~hinton/csc2535/lectu
res.html
Andrew Ng
“Machine Learning”
https://class.coursera.org/ml-007/lecture
http://cs229.stanford.edu/materials.html
100. 機械による細胞のがん化の判定
自然な細胞死と悪性ガン腫瘍と細胞の壊死を、
人間より正確に判定するという。
“LEARNING INVARIANT FEATURES OF
TUMOR SIGNATURE”
http://cs.stanford.edu/~quocle/InvariantLearningISBI12.
pdf
101.
102. Andrew Ngらの新しい知覚研究と
Googleの協力
“Building High-level Features Using Large
Scale Unsupervised Learning”
Jeff Dean & Andrew Y. Ng et al
2012年7月
http://arxiv.org/pdf/1112.6209.pdf&embed
ded=true
108. Now You Can Build Google’s $1M
Artificial Brain on the Cheap
On Monday, he’s publishing a paper that
shows how to build the same type of
system for just $20,000 using cheap,
but powerful, graphics microprocessors,
or GPUs. It’s a sort of DIY cookbook on
how to build a low-cost neural network.
http://www.wired.com/2013/06/andrew_ng/
GPUを利用した、NNのコモディティ化
126. 無限性Recursion
I ate a banana
I know I ate a banana
I think I know I ate a banana
I declare I think I know I ate a banana
What do you declare I think I know I ate?
This is the banana I declare I think I know I
ate
There are more bananas I declare I think I
know I ate than bananas that still grow on the
tree
This is the cat that caught the rat that ate the
cheese that... http://bit.ly/1uc4Taj
127. The Faculty of Language:
What Is It, Who Has it,
and How Did It Evolve?
http://www.chomsky.info/
articles/20021122.pdf
言語のRecursionの能力は、
数の概念の形成能力とも結
びついている。ハトは、3以上
の数を区別出来ない。サルは、
個別の数を個別に学ぶが、
人間の子のように一般化出来
ない。
133. Power of Data
Googleの言語へのアプローチ
Ben Jai “What’s Google Doing” から
http://life.math.ntu.edu.tw/sites/
life.math.ntu.edu.tw/files/
Fall-2006-Campus-Talk-TW.pdf
151. "An example of what recurrent neural nets
can now do (to whet your interest!) "
It is less favorable to the good boy for when to
remove her bigger. In the show’s agreement
unanimously resurfaced. The wild pasteured
with consistent street forests were
incorporated by the 15th century BE. In 1996
the primary rapford undergoes an effort that
the reserve conditioning, written into Jewish
cities, sleepers to incorporate the .St Eurasia
that activates the population. Mar??a Nationale,
Kelli, Zedlat-Dukastoe, Florendon, Ptu’s
thought is. To adapt in most parts of North
America, the dynamic fairy Dan please believes,
the free speech are much related to the
http://bit.ly/1qN5Pes
152. "An example of what recurrent neural nets
can now do (to whet your interest!) "
while he was giving attention to the second
advantage of school building a 2-for-2 stool
killed by the Cultures saddled with a halfsuit
defending the Bharatiya Fernall ’s office . Ms .
Claire Parters will also have a history temple
for him to raise jobs until naked Prodiena to
paint baseball partners , provided people to
ride both of Manhattan in 1978 , but what was
largely directed to China in 1946 , focusing on
the trademark period is the sailboat yesterday
and comments on whom they obtain overheard
within the 120th anniversary , where ......
http://bit.ly/1qN5Pes
161. Google Knowledge Graph
新しい検索の三つの特徴
正しい「もの」を見つける。(Find the right
thing)
最良の要約を得る。(Get the best
summary)
さらに深く、さらに広く。(Go deeper and
broader)
GoogleのKnowledge Graphについては、2012年7月12日の
クラウド研究会での丸山の資料「Googleの新しい検索技術
Knowledge Graphについて」を参照されたい。
https://drive.google.com/file/d/
0B04ol8GVySUuUFUtbWcxNFlHY3c/edit?usp=sharing
165. Facebook
next 10-year plans
「Graph Searchは、ほとんど動かない」
2013年1月30日
Interview with Business Week
http://www.businessweek.com/articles/201
4-01-30/facebook-turns-10-the-mark-zuckerberg-
interview#p4
178. Types = “is-a”
Person
Location
Business
Movie
TV show
Fictional
character
Winery
Automobile model
Airport
Travel destination
Ingredient
Mountain
Cause of death
Political party
179. Properties = “has-a”
PersonのProperty
gender
date of birth
place of birth
profession
nationality
ethnicity
parents
children
religion
education
employment
history
... and more
180. Freebaseの基本的なコンセプト
Topics
Graph: データはグラフとして格納される。
グラフのnode は、/type/object で定義され、
グラフのedgesは、/type/link で定義される。
Topics: Freebaseは、people, places, things のよ
うな現実世界の実体について3,900万のtopicを持って
いる。topicのtypeの例。
Physical entities, e.g., Bob Dylan, the Louvre
Museum, the Saturn planet
Artistic/media creations, e.g., The Dark Knight
(film), Hotel California (song)
Classifications, e.g., noble gas, Chordate
Abstract concepts, e.g., love
181. Freebaseの基本的なコンセプト
TypeとProperty
topicは、いろいろな観点からみることが出来る。
例えば
Bob Dylan was a song writer, singer, performer,
book author, and film actor;
Leonardo da Vinci was a painter, a sculptor, an
anatomist, an architect, an engineer, ...;
Love is a book subject, film subject, play subject,
poetry subject, ...;
Freebaseのtopicには、任意の数のtypeを割り当てるこ
とが出来る。Bob Dylanのtopicには、song writer,
singer, performer, book author、film actor といっ
たtypeが割り当てられる。
182. Freebaseの基本的なコンセプト
TypeとProperty
それぞれのtypeは、そのtypeに固有のpropertyを持つ。
例えば、
The music artist type contains a property that lists
all the albums that Bob Dylan has produced as well
as all the music instruments he was known to play;
The book author type contains a property that lists
all the books Bob Dylan has written or edited, as
well as his writing school of thoughts or
movement;
Typeは、Propertyのコンテナーと考えていい。
183. Freebaseの基本的なコンセプト
DomainとID
propertyがtypeでまとめられるように、typeはdomain
にまとめられる。domainにはIDが与えられる。例えば、
/business is the ID of the Business domain
/music - the Music domain
/film - the Film domain
/medicine - the Medicine domain
それぞれのtypeもIDを持つ。例えば、
/music/album is the ID of the (Music) Album type,
belonging in the Music domain
/film/actor - the Actor type in the Film domain
/medicine/disease - the Disease type in the Medicine
domain
184. Freebaseの基本的なコンセプト
Compound Value Types
複数のフィールドを持つType。
例えば、populationというCompound Value Typeの
Valueは、number of peopleとdateという、二つの
フィールドを持つ。
188. Search Output
Find entities named "Blade Runner" and return
their contributors
filter=(all name{phrase}:"Blade Runner")
&output=(contributor)
Find US Presidents and return their date of
birth as well as their spouses' date_of_birth:
filter=(all type:/government/us_president)
&output=(/people/person/date_of_birth
(/people/marriage/spouse
/people/person/date_of_birth))
211. Gooogle
Actions in the Inbox
Schema.orgのマークアップをe-mailに埋め
込む技術。Google Nowで使われている。
https://developers.google.com/gmail/action
s/
212. Adding structured data
to the email
<html>
<head>
<title>Did you enjoy Google Cafe?</title>
</head>
<body>
<p>
Dear John, please rate Google Cafe between 1 and 5 stars and,
optionally, add a text comment.
</p>
</body>
</html>
225. Using the Advertising ID
The advertising ID is a unique but
user-resettable string identifier that lets
ad networks and other apps
anonymously identify a user. The user's
advertising ID is made available to apps
through APIs provided in Google Play
services.
226. Using the Advertising ID
Users can reset their advertising ID at
any time, right from the Ads section of
the Google Settings app on their devices.
From the same app, users can also opt-out
of targeted advertising based on the
advertising ID by setting the appropriate
ad tracking preference. When the
user opts-out of targeted ads, this ad
tracking preference is made available to
apps through a Google Play services API.
227. Android 広告ID の使用
Google Play 開発者サービスバージョン4.0
では、広告と分析のプロバイダが使用する新しい
API とID が導入されました。このAPI とID を
使用するための規約は以下のとおりです。
個人を特定できる情報またはその他のID との
関連付け: 広告ID をユーザーの明示的な同意
なしに、個人を特定できる情報にリンクしたり、永
続的なデバイスID(例: SSAID、MAC アドレス、
IMEI など)に関連付けたりしてはいけません。