Archive for the ‘Learning’ Category


Friday, February 17th, 2017

Mindstorms by Seymour Papert.

From the webpage:

Seymour Papert’s Mindstorms was published by Basic Books in 1980, and outlines his vision of children using computers as instruments for learning. A second edition, with new Forewords by John Sculley and Carol Sperry, was published in 1993. The book remains as relevant now as when first published almost forty years ago.

The Media Lab is grateful to Seymour Papert’s family for allowing us to post the text here. We invite you to add your comments and reflections.

From the introduction:

…I believe that certain uses of very powerful computational technology and computational ideas can provide children with new possibilities for learning, thinking, and growing emotionally as well as cognitively….

You should read Mindstorms along with Geek Heresy by Kentaro Toyama.

Toyama gives numerous examples that dispel any naive faith in technology as a cure for social issues.

Given the near ubiquitous presence of computers in first world countries, how do you account for the lack of children with

…new possibilities for learning, thinking, and growing emotionally as well as cognitively….

If new learning or thinking has developed, it’s being very well hidden in national and international news reports.

Reinforcement Learning: An Introduction

Tuesday, September 27th, 2016

Reinforcement Learning: An Introduction, Second edition by Richard S. Sutton and Andrew G. Barto.

From Chapter 1:

The idea that we learn by interacting with our environment is probably the first to occur to us when we think about the nature of learning. When an infant plays, waves its arms, or looks about, it has no explicit teacher, but it does have a direct sensorimotor connection to its environment. Exercising this connection produces a wealth of information about cause and effect, about the consequences of actions, and about what to do in order to achieve goals. Throughout our lives, such interactions are undoubtedly a major source of knowledge about our environment and ourselves. Whether we are learning to drive a car or to hold a conversation, we are acutely aware of how our environment responds to what we do, and we seek to influence what happens through our behavior. Learning from interaction is a foundational idea underlying nearly all theories of learning and intelligence.

In this book we explore a computational approach to learning from interaction. Rather than directly theorizing about how people or animals learn, we explore idealized learning situations and evaluate the effectiveness of various learning methods. That is, we adopt the perspective of an artificial intelligence researcher or engineer. We explore designs for machines that are effective in solving learning problems of scientific or economic interest, evaluating the designs through mathematical analysis or computational experiments. The approach we explore, called reinforcement learning, is much more focused on goal-directed learning from interaction than are other approaches to machine learning.

When this draft was first posted, it was so popular a download that the account was briefly suspended.

Consider that as an indication of importance.



International Conference on Learning Representations – Accepted Papers

Monday, February 8th, 2016

International Conference on Learning Representations – Accepted Papers

From the conference overview:

It is well understood that the performance of machine learning methods is heavily dependent on the choice of data representation (or features) on which they are applied. The rapidly developing field of representation learning is concerned with questions surrounding how we can best learn meaningful and useful representations of data. We take a broad view of the field, and include in it topics such as deep learning and feature learning, metric learning, kernel learning, compositional models, non-linear structured prediction, and issues regarding non-convex optimization.

Despite the importance of representation learning to machine learning and to application areas such as vision, speech, audio and NLP, there was no venue for researchers who share a common interest in this topic. The goal of ICLR has been to help fill this void.

That should give you an idea of the range of data representations/features that you will encounter in the eighty (80) papers accepted for the conference.

ICLR 2016 will be held May 2-4, 2016 in the Caribe Hilton, San Juan, Puerto Rico.

Time to review How To Read A Paper!


I first saw this in a tweet by Hugo Larochelle.

Making Learning Easy by Design

Tuesday, October 20th, 2015

Making Learning Easy by Design – How Google’s Primer team approached UX by Sandra Nam.

From the post:

How can design make learning feel like less of a chore?

It’s not as easy as it sounds. Flat out, people usually won’t go out of their way to learn something new. Research shows that only 3% of adults in the U.S. spend time learning during their day.¹

Think about that for a second: Despite all the information available at our fingertips, and all the new technologies that emerge seemingly overnight, 97% of people won’t spend any time actively seeking out new knowledge for their own development.

That was the challenge at hand when our team at Google set out to create Primer, a new mobile app that helps people learn digital marketing concepts in 5 minutes or less.

UX was at the heart of this mission. Learning has several barriers to entry: you need to figure out what, where, how you want to learn, and then you need the time, money, and energy to follow through.

A short read that makes it clear that designing a learning experience is not easy or quick.

Take fair warning from:

only 3% of adults in the U.S. spend time learning during their day

when you plan on users “learning” a better way from your app or software.

Targeting 3% of a potential audience isn’t a sound marketing strategy.

Google is targeting the other 97%. Shouldn’t you too?

Comparing supervised learning algorithms

Friday, February 27th, 2015

Comparing supervised learning algorithms by Kevin Markham.

From the post:

In the data science course that I instruct, we cover most of the data science pipeline but focus especially on machine learning. Besides teaching model evaluation procedures and metrics, we obviously teach the algorithms themselves, primarily for supervised learning.

Near the end of this 11-week course, we spend a few hours reviewing the material that has been covered throughout the course, with the hope that students will start to construct mental connections between all of the different things they have learned. One of the skills that I want students to be able to take away from this course is the ability to intelligently choose between supervised learning algorithms when working a machine learning problem. Although there is some value in the “brute force” approach (try everything and see what works best), there is a lot more value in being able to understand the trade-offs you’re making when choosing one algorithm over another.

I decided to create a game for the students, in which I gave them a blank table listing the supervised learning algorithms we covered and asked them to compare the algorithms across a dozen different dimensions. I couldn’t find a table like this on the Internet, so I decided to construct one myself! Here’s what I came up with:

Eight (8) algorithms compared across a dozen (12) dimensions.

What algorithms would you add? Comments to add or take away?

Looks like the start of a very useful community resource.

Announcing Digital Pedagogy in the Humanities: Concepts, Models, and Experiments

Tuesday, December 23rd, 2014

Announcing Digital Pedagogy in the Humanities: Concepts, Models, and Experiments by Rebecca Frost Davis.

From the post:

I’m elated today to announce, along with my fellow editors, Matt Gold, Katherine D. Harris, and Jentery Sayers, and in conjunction with the Modern Language Association Digital Pedagogy in the Humanities: Concepts, Models, and Experiments, an open-access, curated collection of downloadable, reusable, and remixable pedagogical resources for humanities scholars interested in the intersections of digital technologies with teaching and learning. This is a book in a new form. Taken as a whole, this collection will document the richly-textured culture of teaching and learning that responds to new digital learning environments, research tools, and socio-cultural contexts, ultimately defining the heterogeneous nature of digital pedagogy. You can see the full announcement here:

Many of you may have heard of this born-digital project under some other names (Digital Pedagogy Keywords) and hashtags (#digipedkit). Since it was born at the MLA convention in 2012 it has been continually evolving. You can trace that evolution, in part, through my earlier presentations:

For the future, please follow Digital Pedagogy in the Humanities on Twitter through the hashtag #curateteaching and visit our news page for updates. And if you know of a great pedagogical artifact to share, please help us curate teaching by tweeting it to the hashtag #curateteaching. We’ll be building an archive of those tweets, as well.

After looking at the list of keywords: Draft List of Keywords for Digital Pedagogy in the Humanities: Concepts, Models, and Experiments, I am hopeful those of you with a humanities background can suggest additional terms.

I didn’t see “topic maps” listed. 😉 Maybe that should be under Annotation? In any event, this looks like an exciting project.


Getting Good Tip

Monday, August 11th, 2014

I first saw:

“if you want to get good at R (or anything really) the trick is to find a reason to use it every day”

in a tweet by Neil Saunders, quoting Tony Ojeda in How to Transition from Excel to R.

That sounds more doable than saying: “I will practice R for an hour every day this week.” Some days you will and some days you won’t. But finding a reason to use R (or anything else) once a day, I suspect it will creep into your regular routine.


How to learn Chinese and Japanese [and computing?]

Saturday, March 1st, 2014

How to learn Chinese and Japanese by Victor Mair.

From the post:

Victor concludes after a discussion of various authorities and sources:

If you delay introducing the characters, students’ mastery of pronunciation, grammar, vocabulary, syntax, and so forth, are all faster and more secure. Surprisingly, when later on they do start to study the characters (ideally in combination with large amounts of reading interesting texts with phonetic annotation), students acquire mastery of written Chinese much more quickly and painlessly than if writing is introduced at the same time as the spoken language.

An interesting debate follows in the comments.

I am wondering if the current emphasis on “coding” would be better shift to an emphasis on computing?

That is teaching the fundamental concepts of computing, separate and apart from any particular coding language or practice.

Much as I have taught the principles of subject identification separate and apart from a particular model or syntax.

The nooks and crannies of particular models or syntaxes can weight until later.

Enabling action: Digging deeper into strategies for learning

Sunday, April 21st, 2013

Enabling action: Digging deeper into strategies for learning by Thom Haller. (Haller, T. (2013), Enabling action: Digging deeper into strategies for learning. Bul. Am. Soc. Info. Sci. Tech., 39: 42–43. doi: 10.1002/bult.2013.1720390413)


A central goal for information architects is to understand how people use information, make choices as they navigate a website and accomplish their objectives. If the goal is learning, we often assume it relates to an end point, a question to answer, a problem to which one applies new understanding. Benjamin Bloom’s 1956 taxonomy of learning breaks down the cognitive process, starting from understanding needs and progressing to action and final evaluation. Carol Kuhlthau’s 1991 outline of the information search process similarly starts with awareness of a need, progresses through exploring options, refining requirements and collecting solutions, and ends with decision making and action. Recognizing the stages of information browsing, learning and action can help information architects build sites that better meet searchers’ needs.

Thom starts with Bloom, cruises by Kahlthau and ends up with Jared Pomranky restating Kuhlthau in: Seeking Knowledge: Denver, Web Design, And The Stages of Learning:

According to Kuhlthau, the six stages of learning are:

  • Initiation — the person becomes aware that they need information. Generally, it’s assumed that visitors to your website have this awareness already, but there are circumstances in which you can generate this kind of awareness as well.
  • Exploration — the person sees the options that are available to choose between. Quite often, especially online, ‘analysis paralysis’ can set in and make a learner quit at this stage because they can’t decide which of the options are worth further pursuit.
  • Formulation — the person sees that they’re going to have to create further requirements before they’re able to make a final selection, and they make decisions to narrow the field. Confidence returns.
  • Collection — the person has clearly articulated their precise needs and is able to evaluate potential solutions. They gather all available solutions and begin to weigh them based on relevant criteria.
  • Action — the person makes their final decision and acts on it based on their understanding.

Many web designers assume that their surfers are at the Collection stage, and craft their entire webpage toward moving their reader from Collection to Action — but statistically, most people are going to be at Exploration or Formulation when they arrive at your site.

Does that mean that you should build a website that encourages people to go read other options and learn more, hoping they’ll return to your site for their Action? Not at all — but it does mean that by understanding what people are looking for at each stage of their learning process, we can design websites that guide them through the whole thing. This, by no coincidence whatsoever, also results in websites and web content that is useful, user-friendly, and entirely Google-appropriate.

We all use models of online behavior, learning if you like, but I would caution against using models disconnected from your users.

Particularly models disconnected from your users and re-interpreted by you as reflecting your users.

A better course would be to study the behavior of your users and to model your content on their behavior.

Otherwise you will be the seekers who: “… came looking for [your users], only to find Zarathustra.” Thus Spake Zarathustra