This page:

Manufacturing an Artificial Intelligence Revolution, Yarden Katz, SSRN, Feb 09, 2018

This is an interesting paper from last November that reminds me a lot of the work being done by Audrey Watters. The value of this paper is an informed history of the history of AI and its critics (though I would have given credit for 'the view from nowhere' to Thomas Nagel rather than Alison Adam). It has a number of funny examples of deep learning systems misinterpreting images, though in all fairness a lot of humans would misinterpret them as well. That said, there's no doubt ... [Direct Link]

How to Automatically Generate Textual Descriptions for Photographs with Deep Learning, Jason Brownlee, Machine Learning Mastery, Nov 16, 2017

OK, you're not actually going to learn how to do this simply by reading the article, but you will learn how it's done, and more importantly, that it can be done. The task breaks down into three parts: classifying images (do you see a cat, a rabbit?), describing images (providing a natural language summary of the content), and annotating images (generating text descriptions for specific parts of the image). So basically we're associating object recognition with ... [Direct Link]

Building Competencies for Careers, Maria Ferguson, Diane Stark Rentner, Nancy Kober, Matthew Frizzell, Matthew Brau, Center on Educational Policy, May 31, 2017

This report (16 page PDF) explores how deeper learning competencies apply to the workplace through an analysis of the Occupational Information Network (O*NET) database. The deeper learning competencies include such things as learning to learn, critical thinking, academic mindset, collaboration and communication. The deep learning competencies were required in all 301 occupations evaluated, especially necessary in those occupations with a "bright" outlook. According to the authors, this " ... [Direct Link]

Deep Learning in 7 lines of code, gk_, Medium, Apr 13, 2017

This is a technical article with some good less-technical points. First, we have the idea of how straightforward machine learning has become, as noted in the title. Second, though, those seven lines embody considerable depth of function. The data is run through several layers of neural networks (five of the seven lines in question). Finally, this: "The essence of machine learning is recognizing patterns within data." But it's not just the essence of machine learning, it's the essence ... [Direct Link]

Neural Networks and Deep Learning, Michael Nielsen, Mar 13, 2017

Between meetings with notaries today I was wondering to myself whether work had been done on using one neural network to train another neural network. I didn't find the answer (if you know, send me a note!) but I did find this nice guide to neural networks and deep learning. Michael Nielsen explains these a bit differently than I've seen before, but in such a way as to make some things clearer to me, so I felt it was certainly worth passing along. There are also examples you can work ... [Direct Link]

Bots go bust, Baptiste Parravicini, ReadWrite, Mar 07, 2017

I like predictions that go against the grain, especially when I am fundamentally in agreement with them. Here are the predictions: Bots go bust Deep learning goes commodity AI is cleantech 2.0 for VCs MLaaS dies a second death Full stack vertical AI startups actually work One explanatin summarizes a lot of this: "The bottom line on why it doesn’t work: the people that know what they’re doing just use open source, and the people that don’t will not get anything to ... [Direct Link]

Deep Learning is Revolutionary, Oliver Cameron, Medium, Nov 02, 2016

Yes, this article is pretty superficial (and a "ten reasons" listicle) but if you haven't been looking at some of the things neural networks are doing you may want to take a look. Also, it makes me feel good, because I always knew they'd perform like this. [Direct Link]

Has AI (Finally) Reached a Tipping Point?, Irving Wladawsky-Berger, Oct 24, 2016

Irving Wladawsky-Berger offers a useful overview of contemporary artificial intelligence (AI) from a non-technical perspective referencing Stanford University's One Hundred Year Study on Artificial Intelligence  (AI100, 52 page PDF) including the list of 'hot' areas of current study (quoted, p.9): Large-scale machine learning - algorithms to work with extremely large data sets. Deep learning - has facilitated object recognition in images, video labeling, and ... [Direct Link]

The Commoditization of Deep Learning, Geoffrey Bradway, Medium, Oct 06, 2016

'Deep Learning' is the use of neural networks to do smart things, like grade papers or make recommendations. This article addresses the "commoditization" of deep learning, that is, the trend toward making the data and algorithms available for free. That's why you could use an open source library like Tensor Flow to do neat things with open data. It still takes some smarts, but it's getting easier. The point of this article, though, is that it still takes computing power - ... [Direct Link]

Why does deep and cheap learning work so well?, Henry W. Lin, Max Tegmark, arXiv, Sept 10, 2016

There's a good Technology Review summary of this article. In a nutshell: why do deep learning algorithms, which simulate neural networks, work so well? Mathematically, they should be much less effective, because they are attempting to select the best answer from an enormous number of possible outcomes. According to this paper, the reason is that the laws of physics are biased toward certain outcomes, and neural networks - which emulate physical processes - are biased in a similar ... [Direct Link]

Machine Learning for Designers, Patrick Hebron, O'Reilly, Jun 22, 2016

Long post that introduces machine learning for designers. It requires a (free) O'Reilly login (sorry). People already expert in machine learning won't find anything new but I think it's worth the effort if you don't have background in the field. "Conventional programming languages can be thought of as systems that are always correct about mundane things like concrete mathematical operations. Machine learning algorithms, on the other hand, can be thought of as systems that ... [Direct Link]

Is Big Data Taking Us Closer to the Deeper Questions in Artificial Intelligence?, Gary Marcus, Edge, May 04, 2016

It's easy to get excited about thee potential for big data and deep learning in artificial intelligence, but as Gary Marcus argues in this item, we are still a long way from the goal. Even a one-year old child is further ahead than a robot with it comes to doing things like climbing couches. Machine reading and comprehension is a long way from what humans can do. Siri is not really much of an advance over ELIZA. We should look again at psychology, argues Marcus. "I felt like the field had ... [Direct Link]

Deep Learning with the Analytical Engine, Adam P. Goucher, GitHub, Mar 30, 2016

Cool but challenging. "This repository contains an implementation of a convolutional neural network as a program for Charles Babbage's Analytical Engine, capable of recognising handwritten digits to a high degree of accuracy (98.5% if provided with a sufficient amount of training data and left running sufficiently long)." See also Neural Networks and Deep Learning, " a free online book by Michael Nielsen, which is almost certainly the best hands-on introduction to the subject of ... [Direct Link]

Whisper's Master Of Content Moderation Is A Machine, Harry McCracken, Fast Company, Mar 29, 2016

What if one of the most onerous online learning tasks - content moderation in online forums - could be farmed out to a machine? According to this article, that's exactly what Whisper has done. Whisper - an app that allows people to share secrets anonymously - is particularly vulnerable to abuse. "But the company has a secret weapon: The Arbiter, a piece of software that uses the artificial intelligence techniques known as deep learning to moderate content in the same way a human would,... [Direct Link]

N Cultures, Mark Liberman, Language Log, Jan 31, 2016

Here's another look at the competing 'camps' in machine learning, this time depicted by Jason Eisner as three in number, and based on real work in the field: classical, Baysean, and deep learning. Also interesting in this article is the brief account of the history of dividing fields of study into distinct 'cultures', as well as the division of progress into 'stages' or 'steps'. The two are often related: "Jason presents his intellectual ... [Direct Link]

The race for the master algorithm has begun, Pedro Domingos, Wired, Jan 28, 2016

This is all you need to read from this story: "Backpropagation, a brain-inspired learning algorithm that he co-invented, is taking the world by storm. Rebranded as 'deep learning', it's used by Google, Facebook, Microsoft and Baidu for, among other things, understanding images and speech as well as choosing search results and ads to show you." What's interesting is the method dates to the 1980s. See this paper and this chapter for example. This work got me very ... [Direct Link]

Preparing for a Renaissance in Assessment, Peter Hill, Michael Barber, Pearson Learning, Jan 20, 2016

This report cited in Contact North's Top 10 wish list for 2016 is worth a look. Not that I agree with what's in it, but it's useful to see where one of the largest educational publishers is heading and what it thinks about the state of learning and assessment. The authors look at new technologies such as adaptive testing, multiple versions of tests, data analytics, automated marking, testing for deep learning, and the like. They apply these to assessment challenges, such as ... [Direct Link]

Microsoft Neural Net Shows Deep Learning Can Get Way Deeper, Cade Metz, Wired, Jan 14, 2016

"neural nets use hardware and software to approximate the web of neurons in the human brain. This idea dates to the 1980s, but in 2012, Krizhevsky and Hinton advanced the technology... Deep neural networks are arranged in layers. Each layer is a different set of mathematical operations—aka algorithms. The output of one layer becomes the input of the next." [Direct Link]

Your Algorithmic Self Meets Super-Intelligent AI, Jarno M. Koponen, TechCrunch, Dec 23, 2015

This is the next frontier in e-learning. Yes, we have learning management and learning analytics, but as this article points out, "you can’t directly affect how your personal data is used in these systems." But "when you’re in control, you can let your personal learning system access previously hidden data and surface intimate insights about your own behavior, thus increasing your self-awareness in an actionable way." These machines will use deep learning algorithms ... [Direct Link]

Learners’ Goal Profiles and their Learning Patterns over an Academic Year, Clarence Ng, The International Review of Research in Open and Distance Learning (IRRODL), Jun 26, 2015

I have long argued that the solution to the problem of motivation lies in providing students with subjects they want to study and become proficient in. To my mind, this article to some degree validates that argument. The authors conclude, "Distance learners learn with different goal profiles that are associated with different learning patterns... distance learners who endorsed both mastery and performance-approach goals engaged in deep learning using adaptive strategies consistently ... [Direct Link]

[Next 20 results]

Creative Commons License. gRSShopper

Copyright 2015 Stephen Downes ~ Contact:
This page generated by gRSShopper.
Last Updated: Mar 23, 2018 10:58 a.m.