Deep Learning

It’s a beautiful Friday night in Kirkland, WA. A Mandarin speaking meetup was hosted by the SeattleStartup on the subject of “Deep Learning” at 7:30pm. On my way there, I wondered who’s going to show up and why was Mandarin being used. By 7:40pm, to my surprise, there were 70+ young Chinese professionals in the room; a good number of Microsoft employees mixed with a good number of Amazon employees.

Deeplearning.net summarizes Deep Learning as follows: “Deep learning is a new area of Machine Learning research, which has been introduced with the objective of moving Marching Learning closer to one of its original goals: Artificial Intelligence.

This meetup agenda was as follows:

  1. Dr. Dong Yu, a principle researcher at Microsoft Research, Deep Learning Talk
  2. Panel Discussion about the Deep Learning Application with two AI companies:
  • Kitt AI founder Xuchen Yao and Guoguo Chen. Kitt AI is backed Paul Allen’s AI2, Amazon Alexa Fund and Madrona Venture.
  • Orbeus CEO Yi Li and her team. Orbeus was an AI startup from Silicon Valley in Image Recognition. It was acquired by Amazon and the company has recently moved to Seattle.

Dr. Yu introduced the basic concept of Deep Learning and described the key models such as Deep Neural Networks, Convolutional Neural Networks and Long Short-term Memory Recurrent Networks. He also illustrated the core design principles of Deep Learning models when introducing other new models.

He explained that Deep Learning is feasible only with the computer power and big data available today. It’s actually a rebranding and extension of neural networks. Among the three Deep Learning definitions he gave, the shortest one was “Any system that involves more than one layer of nonlinear processing”

He stated that Deep Learning’s essentials are:

  • Learn complicated feature representation through many layers of nonlinear processing
  • Learn representation automatically and jointly with classification (or whatever) tasks (end-to-end optimization)
  • Key: design the model structure and training criterion.

One of the examples given was AlphaGo; a computer program developed by Google DeepMind in London to play a board game.

AlphaGo’s deep learning is as follows

  • Supervised learning on expert games
  • Reinforcement learning; improve through self-play
  • Build a strategy network and a value network
  • Monte Carlo Tree Search (MCTS) to determine moves through real-time play

Dr. Yu continued to explain Artificial General Intelligence (AGI)

  • The intelligence of a machine that could successfully perform any intellectual task that a human being can (Wikipedia)
  • An emerging field aiming at the building of “thinking machines”; i.e.: General purpose systems with intelligence comparable to that of the human mind (agi-society.org)
  • The general-purpose mechanisms and learning principles that allow machines to explore the world, form connects and clusters, develop and validate theories, learn and generalize from a small number of examples and reason and plan with uncertainty.

The Panel Discussion was quite interesting. In order to learn these two startup’s journey and products, the audience was very actively asking questions and obtaining information. Since these are Chinese startups, it seems using native language Mandarin is a natural way to network and brain storm. That answered my question: why Mandarin.

Personally, I was very happy to have the opportunity to learn about Deep Learning and was impressed by the positive energy brought by these young professionals. I see entrepreneurship emerging among the corporate employees and that’s a good sign. My next blog will explain my perspective about the entrepreneurial culture in corporation. Stay tuned!

[This was originally posted on Linkedin on 5-3-16]

NFC, Big Data, Internet of Things and a Connected Society

Sensors are powerful enablers for Internet of Things.

GE uses data collected by sensors to help in the reduction of millions in fuel costs [1]. Transport for London uses data collected by sensors to understand how people travel and respond quickly when a disruption occurs [2]. The healthcare industry uses data collected by sensors to monitor the activity of patients and provide better patient care [3].

All of these sensor data feed into BigData. Business intelligence (BI) that emerges from sensor data analysis (SDA) helps create new products and services as well as to enhance the existing ones.

Using GE as an example, last year The New York Times notedThomas Edison would be proud. General Electric, the company he started, still knows how to make a buck off cutting-edge technology.” Predix, an IoT (Internet of Things) big data product was implemented entirely with sensor equipped GE machines. Every day Predix gathers 50 million pieces of data from 10 million sensors; including those hooked up to jet engines.

Using Transport for London (TfL) as an example, not only is data collected through NFC enabled Oyster cards but also from sensors attached to vehicles and traffic signals. TfL provides open APIs to software developers to access the data in order to create new services and products. That is an open and forward looking approach since community effort always generates a win-win for consumers and businesses.

In the case of health care, sensors can be embedded in the hospital beds, medical devices or wearable. There are so many use cases to show the power of connected devices. For example, a monitor system that learns normal patient physiological and activity patterns would send an alert when abnormal data (change in blood sugar, for example) appears.

When all of these connected devices are talking to each other through a wireless infrastructure or the internet, we are in a world of Internet of Things. Products and services that collect, store and analyze data are just beginning of a transformation enabled by technologies like sensors, cloud and mobility. BI and business decisions are the focus right now. At the same time, new business opportunities are unlimited. How to seize the opportunity requires visionary ideas and technology agility [4]. The end result would be a Connected Society [5]; where people are more connected to each other and to their environment.

It’s an exciting time! What is one challenge you are facing?

[1] Enterprise Big Data (Data Lake) https://www.linkedin.com/pulse/20140815010520-12418284-enterprises-big-data-data-lake?trk=mp-reader-card&trk=mp-reader-card

[2] How Big Data and the Internet of Things Improve Public Transport in London http://www.forbes.com/sites/bernardmarr/2015/05/27/how-big-data-and-the-internet-of-things-improve-public-transport-in-london/

[3] Sensor Data Analytics – Unlocking Value in ‘Big Data’ http://www.eetimes.com/author.asp?section_id=36&doc_id=1326715

[4] State Technology Welcome Idea Economy https://www.linkedin.com/pulse/state-technology-welcome-idea-economy-meg-whitman?trk=prof-post

[5] Beyond an Internet of Things: Building the Connected Society http://www.boozallen.com/insights/2015/05/beyond-an-internet-of-things/thank-you

Big Data Analytics trends and Sensors’ Role

I attended the Big Data = Big Business Meetup last Thursday and a panel of experts shared their perspectives on the topic “Big Data Solutions – A look into Emerging Tools of the Trade”. It was a good session with 40+ participants.

One of the speakers, Tony Cosentino, VP at Ventana Research, shared the Big Data Analytics trends as follows:

  • Moving from 20th century designed data to 21st century organic data; from confirmatory analytics to exploratory analytics
  • Moving from sample type of analytics to sensors type. Analytics and data are coming together into one environment instead of being separate.
  • Moving the conversation from data to outcome or business orientation

I was particularly interested in Tony’s speech so I did some research about these trends as follows:

  • Designed data vs. organic data:

This Census Bureau’s blog explains that the Census Bureau has created “designed data” based on pre-specified purpose. In contrast, data collected  through internets, sensors and other systems are organic data. The blogger believed “The combination of designed data with organic data is the ticket to the future”.

  • Sample data vs. sensor data:

Sample analytics is used widely in the conventional market research. The research population is generally too big to be covered in a survey; therefore, researchers usually choose a portion of the population (i.e. sample) to do a survey. The sample size and selection are carefully determined in order to capture the representation of the whole population.

A sensor is a device that measures a physical quantity / activity and transforms it into a digital signal. Sensors are always on, capturing data at real time and powering the “Internet of Things.” Sensors can collect enormous data and Cloud computing and storage help to make the analytics possible.

  • Conversations on data vs business:

Data itself is not the focus of the conversation anymore. Nowadays, the business value provided by the big data is the focus.

I agree that the combination of organic data and design data will create valuable data. I believe we need to have a sampling mechanism with organic data since the volume is big. For example, NFC is one of the sensors. When the technology takes off, it will provide interesting data sets. How to translate the data into value added information for businesses take specific design.

This Hadoop blog suggests that “sensors can be used to collect data from many sources, such as:

  • To monitor machines or infrastructure such as ventilation equipment, bridges, energy meters or airplane engines. This data can be  used for predictive analytics, to repair or replace these items before  they fail.
  • To monitor natural phenomena such as meteorological patterns, underground pressure during oil extraction or patient vital statistics during recovery from a medical procedure.”

I think sensors go beyond these domains. For example: an NFC embedded wearable device can monitor body movements and vitals, such as heart rate and blood sugar. Digital health and fitness mentioned in a blog of Aaron Rose  is possible because of the sensors. The Fujitsu NFC smart glove shows a use case beyond digital health and there is unlimited space for monitoring these types of innovations.

These thoughts were triggered by a two hour Meetup. Can you imagine what thoughts will be triggered in two days? I am looking forward to attending the Big Data Innovation Summit held in Santa Clara on April 9 and 10th. With 80+ sessions, it will definitely broaden my vision and expand my imagination.

What are your thoughts on these trends?

Big Data Innovation Summit 2014

Big Data Innovation Summit 2014

Take Control of Your Data

“Big Data” and “analytics” have been in a lot of conversations lately. Many businesses want to jump on the wagon to use big data for marketing and product development. At the same time, most of the companies don’t really know what to do with a flood of data. How to capture, analyze and utilize this data for business insights is critical.

O’Reilly published an ebook titled Business Models for Data Economy, and it is a good reference book for this topic. It says, “Whether you call it Big Data, data science, or simply analytics, modern businesses see data as a gold mine.” Indeed, we live in a fast-evolving world of data economy. Forrester predicted that 2014 would be the year that marketing leaders will put insights to use.

My interest in Big Data started while I was writing my bookEveryday NFC”. I realized that sensors play a crucial role in helping big data move forward, and I wanted to learn more about Big Data. Going to a conference on the topic was a quick way for me to pick up the terminologies, use cases and visions. Networking with early adopters and watching their progress were good learning points. After this initial overwhelming learning experience, I was able to dive deeper into a specific area that applied to my work.

Here is an opportunity for you to learn about Big Data or to share your experience.

The Big Data Innovation Summit will be held in Santa Clara on April 9-10 in 2014, to guide you in taking control of your data. The Summit brings leading thinkers together for presentations, workshops and panels. To network with these people would be a big win in the Summit. There are four tracks: Data Analytics, Hadoop & No SQL, Data Science and the Machine Learning & Algorithms. You can view the schedule here. Follow #datawest14 and @IE_BigData for further updates.

Are you ready to jump on the wagon?

Image

Reflections on Big Data Confereneces

Last week, I attended two technical events. One was Big Data Techcon in San Francisco and another was Seattle Biz-Tech Summit 2013. My focus was on Big Data and proximity sensors.

For Big Data Techcon, there were many sessions about tools; for example, how to collect data, analyze the data and make a correct interpretation of the analytics. The emphasis is on engineering data. Two things that stood out for me was the graph data base and the keynote speech by Doug Cutting.

The graph data base has an advantage to visualize the connections between Big Data. The book, “Graph Databases”, was given away at the session led by Max De Marzi. He was passionate about Neo4j and showed us the connections between Facebook accounts using code. The connections were visualized regardless of the privacy setting in Facebook.

In the keynote speech, Doug Cutting, the founder of Hadoop claimed that “Hadoop 2 is the Big Data OS” and “Open source’s time has come”. After the keynote, Doug was available to talk to people who wanted to obtain his insights or wanted to have a photo with him. Regarding my inquiry about his view on proximity sensors and Big Data, he saw the significance of the sensor impact to Big Data and made an example with retail stores “What would be the value to the retail stores when they can figure out the shopper’s favorable route.

Seattle Biz-Tech Summit 2013 also focused on Cloud and Big Data. I particularly enjoyed the panel “Innovation and Impacts of Cloud Computing and Big Data”. Dave Segleau, Director, Oracle described the phases of the customer adoption of Big Data as:

  1. What is Big Data?
  2. What can Big Data do?
  3. I have a Big Data (or NoSQL) problem. How can I use your product help me build and deploy a Big Data (or NoSQL) based solution?
  4. I’m starting to understand the issues (limitations, requirements, administration) around managing a Big Data (or NoSQL) solution.
  5. Here’s how I can leverage Big Data to benefit the Enterprise and our customers.

Ying Li, Director, ACM SIGKDD suggested that we would move from an engineering data phase to a data knowledge sharing phase in the future. She was an advocate for open data. Jay Mozek, Chief Architect & Director, iSoftStone thought that we need to be clear about the business goal before engineering data. Chris Garvery, Senior Director, Expedia encouraged us to think what we can’t do today and use data to discover the possibility. Panelists had their own perspectives and their unique views made the session informative and interesting.

Yesterday, I found this article that shared how graphics chips can help process big data sets in milliseconds and “opening up new ways to visually explore everything from Twitter posts to political donations.” This trend of facilitating big data visualization is certainly in full swing.

IMG_0163[1]

Sensors and Big Data Analytics

After learning about Google’s Sensing Lab, I did some reading on Big Data and sensors.

In the book of “Taming the Big Data Tidal Wave” by Bill Franks, the value of sensor data was demonstrated with the case of industrial engines and equipment. It discussed how the embedded sensors were utilized from aircraft engines to tanks in order to monitor the second-by-second or millisecond-by-millisecond status of the equipment. All data was fed into “Big Data” analytics.

IBM and The Beacon Institute also collaborated on an effort to use a sensor-enabled monitoring network In order to track temperature, salinity and pollution of the Hudson River. Actually IBM Big Data Technology is used to develop several environmental protection projects like this one.

What about proximity sensors and Big Data? Coca Cola is using NFC tags and QR codes in 100 selected retail stores to collect the data about user behavior and handsets. The backend platform collects analytics such as time, location, frequency of interaction, tap vs. scan, phone model, operating system, service provider and browser type. SocialTagg, a startup in LA, offers an event management platform to enrich attendees’ networking experience by using Big Data analytics on QR codes/NFC tags that were assigned to the event participants.

I will be leading a panel on “Building a Link Between NFC/Proximity Technologies & Big Data” in WIMA USA – NFC and Proximity Solution conference on October 29th in San Francisco. I am looking forward to having a rich discussion with the participants. If you are a “Big Data” expert and would like to join the panel, please contact me at info@everydaynfc.com.

Big Data and Proximity Sensors

I attended “Social Data Week Seattle” hosted by the Tableau Software today. An overview of the workshop stated that “Social Data Week Seattle will explore the business opportunities and practicalities of creating a socially intelligent business by leveraging big data, social data and analytics.”

I arrived 30 minutes late, and there were no seats left. In front of the packed room, Rahul Khandkar from Google was giving a presentation on “Google and Big Data”. He mentioned that smart sensors, machines and social data sources will generate large volume data which will grow with time. He also demonstrated how data collected by Google Sensing Lab is analyzed.

  • Sensors (RFID or NFC?) pick up data.
  • Endpoints/App Engine ingests and processes data.
  • Datastore stores data.
  • Compute Engine BigQuery computes data.
  • Data is visualized by a Dashboard after big data analytics work.

In a previous blog, I’ve mentioned how Google is collecting purchaser’s data. Now, I’ve learned that Google is planning to collect more data from proximity sensors. What a vision!

Daniel Hom from Tableau spoke on “Using Social Analytics for Insights”. He demonstrated analyzing social data with Tableau:

  • Datasift, a Cloud platform for extracting value from social data, collects data.
  • Googlebigquery, a web service, performs interactive analysis.
  • Tableau displays the analytics in a dashboard instantaneously.

Tableau seems to be a very robust tool at making sense of and visualizing big data.  How fortunate it is that we have these innovative companies in Seattle.

I like the idea of “creating a socially intelligent business by leveraging big data, social data and analytics.” Web2.0 provides a self-expression platform and we have a large self-expressed population. What could be easier to understand our customers than understanding their needs and wants through social data?

IMG_0123

IMG_0124

IMG_0129

Big Data and NFC Mobile Payment

Today, while I was finishing an online purchase, a window popped up on my screen to ask if I wanted a free purchase protection service from Google. (See screen shot at the end of the blog)

If I were regular online shopper, I might think, “How nice, why not to take advantage of the free protection?” However, being into Big Data, I see this as a smart attempt on behalf of Google to collect data. Currently Google collects a lot of information about their customers using Google search, Gmail, Google docs, maps and calendar. Offering free purchase protection gives them the opportunity to further develop their big data analysis by collecting information on online purchase.

In its recent blog, Google said if any information is stored in Google+, Gmail,  or Google calendar, it can instantly accessible through voice search. This information, which will be secured through encryption, includes information on flights, reservations, purchases, plans, and photos. This is how transparent our life. The more data we provide to a service provider, the more they can make our life easier. The cost, however, is that we have less privacy. For example, A Wall Street Journal article titled “NSA Reaches Deep Into U.S. To Spy on Net” reported yesterday that NSA systems have the capacity to reach roughly 75% of all US internet traffic.

When NFC mobile payment takes off, telecoms will own mobile purchase information. That data will enable them to understand their customers better. If I were a leader in the telecom industry, I would want to obtain clarity about what consumer mobile purchase data means to the bottom line of my business, to the lives of my consumers and to the whole ecosystem in order to generate the business intelligence that benefits all.

In the article “One Man’s Trash is Another Man’s Big Data”, the author pointed out ” We have a long way to go to get rid of the mindset that data is all about marketing and advertising.” I trust innovation comes from this kind of awareness, sharing and discussion. What do you think?

croppedpic

NFC and Big Data

Near Field Communication (NFC) is a wireless connectivity technology. The key word here is “connectivity”. It is not rocket science; it is similar to WIFI or Bluetooth, the technologies that we are familiar with and widely use. NFC enables devices within short proximity (4 cm) to connect with each other in order to exchange information. It also enables devices to read and write from/to NFC tags in order to retrieve or distribute data.

The term WIFI was used commercially in 2000. I remember that in 2002, I used three long Ethernet cables in order to connect  my home computers to the internet. The next year, I bought a WIFI router and started taking advantage of wireless connectivity. Nowadays, when we go to Starbucks, we connect our laptops to the WIFI network and start browsing the web within seconds. The hardware (wireless adapter) and user interface make the connectivity so easy that we don’t even think of the underlying mechanism.

Bluetooth specifications were developed in 1994 and many products are using the technology now. In the tele-healthcare industry, Bluetooth enabled medical devices provide significant value to the consumer. For example, a patient can take a blood pressure reading with a Bluetooth device at home and a nurse can remotely monitor the readings through a server.

NFC is based on RFID technology. The first RFID patent was granted in 1983. The NFC Forum was established in 2004. In 2006, the first NFC-enabled phone was released by Nokia. Today, most newly released smart phones are NFC enabled; they have an NFC chip inside the phone that communicates with NFC tags or other NFC enabled phones, consumers know very little about this capability because of lacking of education.

With your NFC enabled phone,  you can tap your phone with a new friend’s NFC phone and exchange contact information upon agreement. That’s the simplicity of connectivity. This kind of data exchange will add more data into the Big Data realm.

Big data is a hot topic nowadays. Enterprises want to leverage it in order to serve their customers or to develop customer desired products. Whatever we say, write, connect to or exchange with; structured or unstructured data all are part of the Big Data. With the broad usage of mobile devices, big data can be collected easily with or without our permission. When combined with Cloud computing, Big Data analysis can be performed easily and create tremendous impacts to the businesses.

AWS Summit NYC 2012, Werner Vogel’s keynote shared how a drug designer used computational chemistry algorithms with 21 Million compounds to develop a medication that would treat cancer. A Cloud platform shortened the process time and operational expense while dealing with big data analysis.

NFC, like any other connectivity enabler data traffic, will provide useful information for businesses and provide value for consumers. Any business that is thinking of capturing NFC data will be ahead of the game. I believe when consumers start to adopt NFC, it will become part of our life.