Tag Archives: Data

Man Trains Dog. Dog Trains AI Model. Cats Rule the World.

Researchers Teach AI to Think like a Dog

Source: Researchers teach AI to think like a dog and find out what they know about the world – The Verge

Animals could provide a new source of training data for AI systems.

To train AI to think like a dog, the researchers first needed data. They collected this in the form of videos and motion information captured from a single dog, a Malamute named Kelp. A total of 380 short videos were taken from a GoPro camera mounted to the dog’s head, along with movement data from sensors on its legs and body.

They captured a dog going about its daily life — walking, playing fetch, and going to the park.

Researchers analyzed Kelp’s behavior using deep learning, an AI technique that can be used to sift patterns from data, matching the motion data of Kelp’s limbs and the visual data from the GoPro with various doggy activities.

The resulting neural network trained on this information could predict what a dog would do in certain situations. If it saw someone throwing a ball, for example, it would know that the reaction of a dog would be to turn and chase it.

The predictive capacity of their AI system was very accurate, but only in short bursts. In other words, if the video shows a set of stairs, then you can guess the dog is going to climb them. But beyond that, life is simply too varied to predict. 

 

Dogs “clearly demonstrate visual intelligence, recognizing food, obstacles, other humans, and animals,” so does a neural network trained to act like a dog show the same cleverness?

It turns out yes.

Researchers applied two tests to the neural network, asking it to identify different scenes (e.g., indoors, outdoors, on stairs, on a balcony) and “walkable surfaces” (which are exactly what they sound like: places can walk). In both cases, the neural network was able to complete these tasks with decent accuracy using just the basic data it had of a dog’s movements and whereabouts.

Dog AI Model Training
Dog AI Model Training

 

G.E. Plans Big Entry into IoT, Providing Analytics and Predictive Rules

G.E. Plans App Store for Gears of Industry

The investment of $500 million annually signals the importance of the so-called Internet of Things to the future of manufacturing.

G.E. expects revenue of $6 billion from software in 2015, a 50 percent increase in one year. Much of this is from a pattern-finding system called Predix.  G.E. calls its new service the Predix Cloud, and hopes it will be used by both customers and competitors, along with independent software developers. “We can take sensor data from anybody, though it’s optimized for our own products,” Mr. Ruh said.

[Competitive solutions from IBM, Microsoft, and Google] raises the stakes for G.E. “It’s a whole new competition for them,” said Yefim Natis, a senior analyst with Gartner. “To run businesses in a modern way you have to be analytic and predictive.”

G.E. is running the Predix Cloud on a combination of G.E. computers, the vast computing resources of Amazon Web Services, and a few [local] providers, like China Telecom.

China, along with countries like Germany, [are] sensitive about moving its data offshore, or even holding information on computers in the United States.  
The practice of “Ring fencing”  data exists in dozens of jurisdictions globally.  Ring fencing of data may be a legal and/or regulatory issue, that may inhibit the global growth of cloud services moving forward.

Source: G.E. Plans App Store for Gears of Industry

Wireless Carriers: Data Only Plans

Is there any traction on ‘Data Only’ wireless plans which augment your primarily carrier?

E.g. when you reach the allocated capacity of the primary plan, one of N secondary wireless providers may be selected ‘on the fly’ or in  settings.  Each wireless provider may offer their own competitive ‘data only’ plan.  The subscriber may choose a “10 GB for 10 USD.”

Alternatively, the consumer may direct traffic of a certain type, e.g. Network packets for movies to ABC wireless.

Prime example, my family shared plan offers 10 GB to share among the 4 of us.  One of us, for the last two months, chewed up our data plan.   At 75%, 90%, and 100% AT&T sent us warning notifications, which was very good.

After we bust the quota, we are automatically charged one GB for 15 USD.  It may be just me, but that sounds pricey.  Both months I stopped the data component for one of the phones.  Also, a great feature AT&T, but it doesn’t go far enough, block only streaming movies when not in WiFi.  I’d prefer if the middle tier, the wireless solutions management implement the feature.  The kids don’t have access to it as would a device implementation.

Adjustable Data Quota Limits on Member’s of Family/Shared Data Plans

A cap on the amount of shared data each person in the plan is allowed to use should be adjustable at any point in the billing cycle. Good Use Case:  my kids are on my family plan, and I want to limit their data usage.  They always use more than my wife and I.

Another possible use case, from a small business perspective, if you add a few lines on your plan, then you may allocate to specific type of employees, such as sales reps., specific amounts of data.  There are several types of widgets can be used, such as a pie chart, and the total pie represents the total data package, and each slice represents an allocation to each member of the shared plan.

Update: I stand corrected, and do see an AT&T Smart Limits for 4.99 USD per month:

Under Code, Apps Would Disclose Collection of Data – NYTimes.com

Under Code, Apps Would Disclose Collection of Data – NYTimes.com.

I do not believe the disclosure of consumer information collected by mobile applications to provide mobile app transparency should be the only disclosure channel of consumer data collected.  I think that all of the web browsers should have an option to see, for the current web site, or application, the chart, or similar as mentioned in the NYT article, which is similar to nutrition details of food from the FDA.  That would allow for a majority of the disclosure, and would satisfy the consumer lobby groups.

Consumer collected data from all electronic sources should be disclosed in a simple manner similar to the ingredients label from the FDA.

 

NFC Replaces Secure Digital Memory for Data Transfer in Art Galleries?

NFC (Near Field Communications) has significant potential in the transfer of information, and has already proven to be a lightweight technology to transfer and store data.  We have already seen at this year’s CES conference business cards enable the transfer of songs from an NFC enabled business card to a car radio.  Samsung has enabled this technology in their smartphones to transfer data such as videos and pictures.

There will come a day soon where we will have built in storage in a device, such as a picture frame, or television, and the NFC card will allow the transfer of information to this temporary buffer in the device for playing music, watching videos, or looking at pictures. This day is not far off.  Yes, those LCD picture frames in your home that take SD memory are outdated.

Apple made an acquisition of a company that has the ability to enable an LCD touch screen to raise a keyboard through the touch screen, so the user has the tactile contact of the keyboard.  We may go back to typing on the keyboard without looking, like we do with smartphones with keyboards.  I envision an art gallery that has huge LCD screens all around the room, and switching an artist on display would be as easy as walking over to each LCD picture frame and taping the frame enabled with this raised, tactile LCD technology.  In the artist’s creation, the paint of the brushstrokes may appear raised from the LCD canvas, with a three dimensional effect on the picture frame.  An artist making an art creation would make brush strokes using a digital brush, pressing like you would on a canvas, choosing the appropriate paint may record the additional information required to display a three dimensional painting.

Picture that.

Addendum:

After additional research, the one inhibitor, which may pose a significant barrier, and provides optimal data transfer of smaller data packets.

The maximum data transfer rate of NFC (424 kbit/s) is slower than that of Bluetooth V2.1 (2.1 Mbit/s), as noted in Wikipedia.

The speed of MicroSD Speed Class 10 is 10 MB/sec, significantly greater, as well as the advanced UHS, or Ultra High Speed Class, UHS-I has a 50 MB/s, and UHS-II has a theoretical maximum transfer rate of 312 MB/s.

Although, the idea of NFC, or Bluetooth for the matter, has a conceptual idea of tap and transfer high rates for large data to internal memory buffers in devices, the reality is that the  WiFi connectivity speeds outweigh both NFC and Bluetooth, and MicroSD, physical medium outweighs NFC / Bluetooth.  If this idea had merit today, you would need to apply a WiFi connected device to get the maximum throughput without physical media, such as secure digital, or continue to leverage physical media for transfer and still use the memory buffer as a temporary storage in devices, as noted in the article.

Big Data Creates Opportunities for Small to Midsize Retail Vendors

Big Data Creates Opportunities for Small to Midsize Retail Vendors through Collective Affinity Marketing outside Financial Institutions.

In the Harvard Business Review, there is an article, Will Big Data Kill All but the Biggest Retailers?  One idea to mitigate that risk is to create a collective of independent retailers under affinity programs, such as charities, and offer customers every N part of their purchase applies to the charity to reach specific goals as defined by the consumer.   Merchants, as part of this program, decide their own caps, or monetary participation levels.  Consumers belong to an affinity group, but it’s not limited to a particular credit card.  The key is this transaction data is available to all participating merchants for the affinity.  Transaction data spans all merchants within the affinity and not just the transactions executed with the merchant.

Using trusted, independent marketing data warehouses independent retail vendors share ‘big data’ to enable them to compete and utilize the same pool of consumer [habitual] spending data.

Affinity, marketing data companies can empower their retail clients/vendors with the tools for Business Intelligence and pull from the collection of consumer data.  Trusted, independent marketing data warehouses sprout up to collect consumer data and enable it’s retail vendor clients to mine the data.

These trusted loyalty affinity data warehouses, not affiliated with a single financial institution, as previously implemented with credit cards, but more in line with, or analogous to, supermarket style loyalty programs, however, all independent retail vendors may participate OR may cap these affinity program memberships for retail vendor from small to mid-size companies.

Note: Data obfuscation could be applied so customer identification on fields like social security number will not be transparent, limiting any liabilities for fraud.