Skip to main content

Google's Teachable Machine - What it really signifies

The Teachable Machine is an effort by Google to make Machine Learning and AI accessible to the wider public, without requiring any specialized training, knowledge in Computer Science or coding.

The site https://teachablemachine.withgoogle.com/ is a move that reflects the current trend of the personalization of AI in shifting the algorithms from the Cloud to the user's space, be it their desktop, their phone or other smart device.
That's not the biggest problem though; the real issue is that the models used for training the algorithms under the common supervised learning model, require massive datasets and excessive amounts of CPU power.
So as the situation currently stands, the bulk processing is done on the Cloud by Platform as a Services which offer Machine Learning as plug and play API's which encapsulate the necessary pre-trained  algorithms, with offerings including tone analysis, visual recognition or conversation analysis. Prevalent examples of such PaaS are Haven OnDemand and IBM's Watson/BlueMix.
But the tide seems to turning. Back in August 2016, in AI Linux, we  identified a divergence to this trend:
Things seem to be shifting though, with those elaborate algorithms looking to move on to run locally on mobile devices. That includes their training too; the pictures, notes, data and metadata that reside in the device and which are going to be worked upon, will also serve to train the network and aid its learning activities such as the recognizing, ranking and classifying of objects. The difference is that now all of that is going to happen locally.
Qualcomm's Snapdragon 820 processors and their accompanying Snapdragon Neural Processing Engine SDK are behind such a move which would allow manufactures to run their own neural network models on Snapdragon powered devices, such as smart phones, security cameras, automobiles and drones, all without a connection to the cloud. Common deep learning user experiences that could be realized with the SDK, would be scene detection, text recognition, object tracking and avoidance, gesturing, face recognition and natural language processing. 
So instead of the ML algorithms being bred on the cloud, satisfying their hunger with user-collected data, the alternative idea is to shift both the algorithms as well their training offline and onto the source generating the data in the first place.

full article on i-programmer.info

Comments

Popular posts from this blog

Insider's Guide To Udacity Android Developer Nanodegree Part 3 - Making the Baking App

Continuing to chart my experience of Udacity's Android Developer Nanodegree we step up in level, embarking on the advanced part of the super-course.
Completing project "Popular Movies" (see Part 2 of this series) signaled the end of "Android Developer". Now we are ready to tackle the second element of the program "Advanced Android Developer", a new class with a new syllabus and project. Continuing to chart my experience of Udacity's Android Developer Nanodegree we step up in level, embarking on the advanced part of the super-course.

Completing project "Popular Movies" (see Part 2 of this series) signaled the end of "Android Developer". Now we are ready to tackle the second element of the program "Advanced Android Developer", a new class with a new syllabus and project.

"Advanced Android Developer" is a mixed bag of self contained material and of coding seven different sample apps to learn about the…

JSON Feed - The New RSS?

SON Feed is a new take on the web syndication format, but unlike RSS and Atom it's in JSON, not XML. So what does it try to do better?

Mainly overcome the perils of XML; it's complex, heavyweight, difficult to parse and not in sync with the current trend wanting web data exchange happening almost exclusively in JSON document representation.

In contrast, JSON is easier to both write and parse, manipulate and consume, especially given that its data types are exact reflections of their native Javascript counterparts.
Devised by Brent Simmons, the original developer of the popular NetNewsWire and Manton Reece creator of Micro Blogs, both with a great background on publishing with RSS, it's a certainty that JSON Feed will emerge as a strong competitor to both Atom and RSS, being based upon their decade long experience on decentralized formats.

On top of that it also tries to tackle a few other issues plaguing RSS, mainly the lack of realtime client notification whe…

How Much Gameplay Can You Pack In Just 13K?

Given our expectations of Xbox games, you might consider writing a game within a 13K limit, which is the challenge for the annual js13K competition far too restrictive. Its results are now out and prove that it is possible to produce a game that is fun to play. 

Back in the tape loading days and on platforms the likes of Commodore64 games came in sizes of 4K or less. As proof of concept, here's a list of a few such 4K titles, copied over from Lemon64 's archive:
Alien SidestepBug CrusherDot GobblerClose EncountersDot Gobbler v2GridrunnerLaser CyclesMarios BrewerySpace ActionSpace RicoshayTank WarsHesmon64Retro Ball  Fast forward to now, at a time when Javascript's eating the world by making all sorts of applications or  games available to everyone through the medium of the browser, rendering the need of dedicated platforms and Operating systems obsolete, 13K is sufficient enough to pack both gameplay AND cool graphics due to the advanced browser engines and HTML5.