Google A.I. Platform For Free


FULL Nvidia GPU Cloud Presentation | Computex 2017


BAT 馬雲马化腾李彦宏谈人工智能

Google, AI, Machine Learning

Google cloud platform (note5)

Target : the enterprise cloud market

  • Position- To be a developer-friendly platform
  • Weakness

1) Not strong and no impression  on cloud service and the enterprise segment

2) No contribution to the open source community before.


  • Stength : ASIC, GPU and TPU hardware in its cloud
  • Opportunity

1) begin to work with open source projects (note3)

  • Cloud Native Computing Foundation- the open-source container management tool

run by the Linux Foundation (note4)

Other partners in the new foundation include AT&T, Box, Cisco, Cloud Foundry Foundation, CoreOS, Cycle Computing, Docker, eBay, Goldman Sachs, Huawei, IBM, Intel, Joyent, Kismatic, Mesosphere, Red Hat, Switch SUPERNAP, Twitter, Univa, VMware and Weaveworks.

Absent : Microsoft, Amazon, Pivotal, and  Taiwanese tech companies)

the popular open source container orchestration system

  • TensorFlow for machine learning

-Spanner for launching massive distributed databases

-Draco for 3D graphics compression

2) To be a  developer-friendly platform


1)letting customers run whatever open source stack they choose on Google’s infrastructure,

2)releasing and supporting open source projects and making the ecosystem

3)the partners who build tools and technologies on top of GCP, a first class citizen on the platform.

4) treating them as part of the whole and the net is bringing the tech you want and using Google technology or using any of the [partner] services

The KSF :

1)Being open to win the mind shares of developers

2) much more supportive of the open source community makes people feel better about Google and makes developers feel better about working with their tools because they can avoid lock-in

  • Threat : peers: AWS (2006. 1st public cloud, market leader, 1st mover), Microsoft, IBM
  • Strategy

UsingKubernetes, the popular open source container orchestration system offer robust open source tools, something that surprised some people in this market.

  • 4 ways Google will enable enterprises to adopt machine learning and AI (note2)

1). Machine learning computing in Google Cloud

a deep learning algorithm can have tens of millions of parameters, training these machine learning models requires enormous computational resource

the Cloud Machine Learning Engine. This capability is designed for companies with data scientists and machine learning experts who are able to build their own unique machine learning models with libraries such as Tensorflow.

Google’s infrastructure as the solution to speed training times and improve the return on investment. Google has specialized ASIC, GPU and TPUhardware in its cloud to accelerate training and improve the ROI with on-demand cloud resource utilization. After the model is trained, it is deployed in range of platforms—from on-premise to mobile devices.

2. Algorithms and pretrained machine learning models

建ML model 需用 the machine learning engine, 用 Google’s pre-trained models (full list) using APIs to add machine learning capability to their applications, such as understanding natural language, images and natural language.

An API beta for understanding videos

demo: This 3-minute video of the demonstration of the Cloud Video Intelligence beta

3. Google acquires Kaggle for data

Google acquired Kaggle for data sets and talent. Kaggle, founded in 2010, is a community of 850,000 data scientists from around the world that hosts competitions to create the most accurate predictive models and market models, as well as to acquire new public data sets in a variety of fields.

4. Expertise

the Advanced Solutions Lab for customers with ambitious goals to develop machine learning to solve complex problems.




(note4: The mission of Linux foundation :The mission of this new foundation is to “help facilitate collaboration among developers and operators on common technologies for deploying cloud native applications and services,” )


(Reference : Google Cloud Platform 入門)



Machine Learning for Marketing

The marketing big data ecosystem being impacted by machine learning in four major areas:

  1. Automated data visualization (including ML results) will become more rich, and user-friendly.
  2. Content analysis (textual, lexical, multimedia/rich) will be used to drive better marketing conversations.
  3. Incremental ML techniques will become more prevalent, leading to real-time, not just on-going and automated, changes in marketing execution.
  4. Learning from ML results will accelerate the growth and skills of marketing professionals.
  • Automated Data Visualization tools: Tableau and Qlikview

Predictive model : The objective of ML is to build predictive model for forecast.

the ability to modify a solution that is already in place by introducing new data rather than having to stop using the current solution before building a new model from scratch.

(Source from How Machine Learning Will Be Used For Marketing In 2017)

TensorFlowonSpark system architecure



(Yahoo supercharges TensorFlow with Apache Spark)


AI will reduce management

AI用的好,會減少例行性的行政工作,  省下的時間可以專注在抉擇判斷, 反思, 創意企劃發想,實驗驗證(試一試), 在科技界,學界一直在往人工智慧方面前進,但我是沒看到台灣社會有這麼把AI 當回事,  從概念,應用,教育,沒特別感覺有在轉變, 社會上還有其他重要的議題要關注, 這也是事實,但產業,個人如果跟科技業太脫軌, 這樣的落後是個警訊.  產業大概落後15 年.  我不希望見到台灣落後矽谷太多.   社會上新技術的普及速度相見還是緩慢滿多.  學這駕馭工具, 機器, 專注在設計, 創意,企劃,  邁進AI化, 社會化的技能跟人的網絡與連結 也會更重要, 但前提是AI推進社會進化.


AI 領域包含的東西很多, 儘量學, 學的起來分享給親朋好友, 用在好的地方, 讓社會往對的方向前進

How Artificial Intelligence Will Redefine Management




這領域台大資工 Vivian教授有開課

(Artificial neural network)



Sentiment analysis

  • 收集分析意見,態度,意向
  • 用在品牌行銷管理,收集消費者, 用在選民意向,  外匯走勢…等等,  發現資料的深度意義.



  • Subjectivity/objectivity identification (主客觀認定)


classifying a given text (usually a sentence) into one of two classes: objective or subjective.



  • Feature/aspect-based sentiment analysis

It refers to determining the opinions or sentiments expressed on different features or aspects of entities, e.g., of a cell phone, a digital camera, or a bank. (在不同裝置上的各種意見)

A feature or aspect is an attribute or component of an entity, e.g., the screen of a cell phone, the service for a restaurant, or the picture quality of a camera.  (特色是實體的一個屬性)

The advantage of feature-based sentiment analysis is the possibility to capture nuances about objects of interest.  ( 可以抓一些關於偏好比較細微之處)

  • 方法

1 knowledge-based techniques
2 statistical methods
3 hybrid approaches

新媒體的分析就是要用到 sentiment analysis.  從 blog, weibo 抓出偏好.

online opinion has turned into a kind of virtual currency for businesses" ,這是商機之所在.

democratizing data mining of all the content that is getting published."

“As businesses look to automate the process of filtering out the noise, understanding the conversations, identifying the relevant content and actioning it appropriately, many are now looking to the field of sentiment analysis."

用在machine learning , 分辨贊成 和反對的意見. 找用戶的意向,  根據這偏好, 跟用戶推薦, 用Python,



在twitter 上,找 “Sentiment analysis" 這句話的Feature,


例如, Brexit這議題,就很適合做 sentiment analysis. 這樣民意可以徹底反映出來.

Twitter Reveals That the UK Will Vote to Leave the EU
Twitter sentiment analysis.


分析結果的呈現 又跟資料的視覺化有關聯, 我是有用一些資料視覺化軟件,


  • 台大 企業風險管理與商業情報分析研究中心


文字探勘(Text mining)技術不只能分析,媒體記者報導的文章內容,也可以分析散落於網路世界各個部落格的網誌、政府網站等公開資訊內容,事實上很多資訊,在正式管道還未公開前,已經可以在部落格上,看到端倪。像是在部落格的文章,多少帶有部落客的「情緒反應」,透過系統分析,可為文章內的關鍵文字,下「分數」,企業可以根據最後總分的分析結果,做為評估決策判斷的參考,尤其目前這方面之技術成熟度已相當高。

傳統負責銀行貸款的受信業務的人員,在評估放款風險時,以往只能從財務報表與基本書面資料,執行人為判斷,不免有判斷失誤的風險。但透過系統分析,可理清該公司的企業關係網絡(social network)及經營者之人格特質與社會評價,據此可做出更準確之授信風險評估。故若能運用資料與文字探勘(data and text mining)技術,同時有系統的分析財務資訊與網路上「文字類」的商業資訊,結合量化與質化分析,強化商業情報的可信度,協助企業更有效的制訂決策與管控風險。

臺大管理學院「企業風險管理(ERM)與商業情報(Business Intelligence)分析研究中心」未來將整合院內不同系所老師與博士班學生,結合產業資源長期有深度跨議題共同合作,此外本中心未來將與本院「校特聘講座教授」陳炘鈞博士位於美國的研究中心,進行雙邊資源共享交流,帶動本院跨領域之研究風氣,創造世界頂尖之研究領域,最後把研究結果從學界延伸到實務界,帶動台灣產業快速成長。




(Ref: How to Use Social Media Sentiment Analysis in Your Listening?)

(Ref:  )

%d 位部落客按了讚: