Subscribe via RSS Feed Connect on Google Plus Connect on Flickr

Tag: machine learning

“Hey Siri”

Monday, 23 October, 2017 0 Comments

Apple has published a a paper on how its devices listen for people using the “Hey Siri” command/query. There’s a lot of machine learning going on here:

“The ‘Hey Siri’ detector uses a Deep Neural Network (DNN) to convert the acoustic pattern of your voice at each instant into a probability distribution over speech sounds. It then uses a temporal integration process to compute a confidence score that the phrase you uttered was ‘Hey Siri’. If the score is high enough, Siri wakes up. This article takes a look at the underlying technology. It is aimed primarily at readers who know something of machine learning but less about speech recognition.”

Note: Apple says: “Hey Siri” works in all languages that Siri supports, but “Hey Siri” isn’t necessarily the phrase that starts Siri listening. For instance, French-speaking users need to say “Dis Siri” while Korean-speaking users say “Siri 야” (Sounds like “Siri Ya.”) In Russian it is “привет Siri” (Sounds like “Privet Siri”), and in Thai “หวัดดี Siri”. (Sounds like “Wadi Siri”.)


@HumanVsMachine

Monday, 13 February, 2017 0 Comments

Hardly a week goes by without some “expert” or other predicting that by, say, 2020, millions and millions of jobs will be lost in developed economies due to robotics, AI, cloud computing, 3D printing, machine learning and related technologies. Hardest hit will be people doing office and factory work, but other sectors, from trucking to healthcare, will be affected “going forward,” as lovers of business cliché love to say.

The Twitter feed @HumanVSMachine features images showing the increasing automation of work. The footage of people doing a job side-by-side with videos of robots doing the same thing suggests a sombre future of post-human work.

Philippe Chabot from Montreal is the human behind @HumanVSMachine. He was a graphic artist in the video industry and he had plenty of work, once upon a time. But companies began outsourcing their artwork and Chabot found himself competing a globalized market where rivals can create a logo for $5 and software automatically designs avatars. Today, Philippe Chabot works in a restaurant kitchen and he feeds @HumanVSMachine in his free time.

Note: This image of “Robot Baby Feeder; Robot, baby bottle, crib, toy” by Philipp Schmitt is included in the “Hello, Robot. Design between Human and Machine” exhibition at the Vitra Design Museum in Weil am Rhein in Germany.

Raising robot


The gathering storm that is Industry 4.0

Thursday, 9 February, 2017 0 Comments

All our posts about machine learning this week have been prompted by a dramatic shift going on right now called “Industry 4.0.” In essence, this is the end-to-end digitization of all physical assets and their integration into digital ecosystems. Along with machine learning, Industry 4.0 buzzwords include connectivity, supercomputing, artificial intelligence, robots, self-driving cars gene editing and globalization.

The preceding industrial revolutions freed us from animal power, made mass production possible and opened digital doors for billions of people. This Fourth Industrial Revolution, Industry 4.0, is profoundly different in that its technologies are melding the physical, digital and biological worlds and forcing us to confront uncomfortable questions about work, identity and life itself.

The World Economic Forum, which is “committed to improving the state of the world,” produced this clip about Industry 4.0. It’s a positive view, overall, but it does not address the issue that’s roiling politics today: the conflict between the elites, who stand to gain from early access to the upsides of this transformation, and the precariat, which stands to lose the jobs that glue their communities together. More on this here tomorrow.


Machine Learning for Dummies

Wednesday, 8 February, 2017 0 Comments

That’s the title of an wry and informative take by Grzegorz Ziemoński in DZone. His definition of Machine Learning is worth memorizing: “Computer doing statistics on Big Data.” If you want to learn some Machine Learning, but don’t know where to start, his “text is for dummies just like us” is recommended.

Along with offering a concise definition of Machine Learning, Ziemoński takes readers through the difference between supervised and unsupervised Machine Learning and he shows us how to use Amazon Machine Learning to make a simple prediction. The key questions for those wishing to do more are: What do you want to predict? What data do you have? What can you do to make it work?

For those still foggy about the relevance of this stuff, substitute Machine Learning for Big Data in this quote by Dan Ariely, founder of The Center for Advanced Hindsight: “Big Data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it.”

Thanks to Grzegorz Ziemoński, we can now get to first base, as it were.


Machine learning on smartphones

Tuesday, 7 February, 2017 0 Comments

The language used by the acolytes of the high priests of the Information Age is richly encoded. Example: “TensorFlow is now available in a Docker image that’s compatible with Python 3, and for all Python users, TensorFlow can now be installed by pip, Python’s native package manager.”

That’s from an InfoWorld story by Serdar Yegulalp in which he says machine learning will one day run on a smartphone, without cloud support. At the heart of this development is TensorFlow the open-source, deep-learning framework developed by Google. Here’s how the engineers, using human language, decode it:


Learning Machine Learning

Sunday, 5 February, 2017 0 Comments

True story: A player named Libratus sat down at a poker table in a high-stakes game of no-limit Texas Hold’em. The gruelling 20-day tournament ended a week ago in a dramatic victory for Libratus over four of the world’s top players. Libratus is no cigar-smoking dandy cowboy, however. It’s an artificial intelligence (AI).

Machines are getting smarter, and AI is entering society in all kinds of intriguing and disturbing way. But who creates these machine-learning programs and who writes the algorithms that produce everything from stock market predictions to data journalism to poker-winning strategies? It’s time we found out and it’s time to learn how to do it ourselves. But how and where and when?

The ScienceAlert Academy is offering a 73.5-hour course titled “The Complete Machine Learning Bundle” for $39. This is the kind of immersion in the stuff you’ll need to plan a career or take your hobby to the next level. The package contains 10 different courses, including “Hadoop & MapReduce for Big Data Problems” and “From 0 to 1: Learn Python Programming – Easy as Pie”.

AI


Is machine learning magic?

Saturday, 30 July, 2016 0 Comments

That’s the question posed by Adam Geitgey, a “guy who does @programminglanguage at @company.” What is machine learning? Adam Geitgey defines it thus:

Machine learning is the idea that there are generic algorithms that can tell you something interesting about a set of data without you having to write any custom code specific to the problem. Instead of writing code, you feed data to the generic algorithm and it builds its own logic based on the data.

“Machine Learning is Fun!” says Adam Geitgey in what he describes as “The world’s easiest introduction to Machine Learning.”

How is this stuff related to the real world? Alphabet CEO Sundar Pichai devoted much of Thursday’s second-quarter earnings call to the two “Ms” — the mobility that is Alphabet’s present and the machine learning that is its future. Quote:

“Machine learning is the engine that will drive our future, but it is already making our products better, and we use it every day. In fact, more than 100 teams are currently using machine learning, from StreetView to Gmail to voice search and more.”

Make that three Ms: magic, mobility, machine learning.


Learning about machine learning

Tuesday, 15 March, 2016 0 Comments

On Friday, here, we watched Stephen Wolfram speak about the next language. What it’s going to be is undefined, but if we want computers to do increasingly complex things, a shared language will be required. This “code” will express our needs, our wishes, in a way machines can understand. Wolfram’s profound belief is that coding for this future has a philosophical, humanistic, perhaps, divine, purpose. Most people, however, see it in a more prosaic light: learning about the “soul of the machine” is about getting a job.

Enrollment in machine learning classes is soaring in the US, and universities are scrambling to add classes to meet an unprecedented demand writes Jamie Beckett in an NVIDIA blog post. At Carnegie Mellon University’s Machine Learning Department, enrollment in ‘Introduction to Machine Learning’ has jumped nearly 600 percent in the past five years. Applicants to its machine learning Ph.D. program have doubled in six years and the university has added its first undergraduate course on the topic. At the University of California, Berkeley, enrollment in ‘Introduction to Machine Learning,’ has nearly tripled in less than two years says Beckett.

Quote: “In the old days, you had to take an introductory computer class so you’d know how to use a computer at work,” said Lynne E. Parker, division director for the Information and Intelligent Systems Division at the National Science Foundation. “Today, students are recognizing that whatever their chosen field, there’s going to be some automation of the knowledge work — and that’s machine learning.”

Note: Coursera is offering Machine Learning Foundations: A Case Study Approach.

machine learning