We consider the problem of building detectors for high-level concepts using only unsupervised feature learning. For example, we would like to understand if it is possible to learn a face detector using only unlabeled images downloaded from the internet. To answer this question, we trained a simple feature learning algorithm on a large dataset of images (10 million images, each image is 200x200). The simulation is performed on a cluster of 1000 machines with fast network hardware for one week. Extensive experimental results reveal surprising evidence that such high-level concepts can indeed be learned using only unlabeled data and a simple learning algorithm.

"It takes, say, thousands of person hours spread over a handful of scholars to create and ‘debug’ a single conceptual trope. When that’s done the trope can show up in casebooks and undergraduate texts. And from there, it goes into the knowledge-hungry minds of our students. And when one of them writes reviews for The New York Times, BINGO! a conceptual trope enters the self-styled paper of record. And, from there, the world."

"That’s how culture works."

"The story of Goat Island has parallels for us engineers.  Because we cannot predict results, we know that patience, hope and courage are functions of the design process.  Every so often, we have to remind ourselves of that.  We also know that patience is a function of a good retrospective.  Just as it took a certain amount of time for the snappers to grow large enough to take on the urchins, I think there is a certain – and measurable – amount of time for participants in a retrospective to open up and start moving beyond the superficial.  That amount of time is more than two hours."

"This page provides information on the use and implementation of novelty search, an evolutionary search method that takes the radical step of ignoring the objective of search and instead rewarding only behavioral novelty. This visual demonstration (requires modern browser, IE users may need to install a plugin) contrasts a search for novelty with a search for the objective."

"You’ll be told it’s because it makes communicating your vision easier, and that’s true.  But there are two more important reasons.  First, if you know how to be a sound man, you know how to make the sound man’s job easier. This has the potential to make you very popular with sound men (or editors, or cinematographers, etc), something you’ll need when your only currency is good will.  Second, when you begin producing your own work, this renaissance approach to filmmaking will allow you to start before anyone else signs on.  Knowing you can finish in a pinch, if you have to, will lend you a confident relentlessness that makes others want to get involved."

At some point, I’ll turn to my favourite web application combo: Sinatra + MongoDB + Highcharts, to visualize these data dynamically on a web page. For now though, we can get a quick idea and create even more Friday fun by learning how to use RApache to run and view R code in the browser.

"Whether because they're Canadian or because they're engineers, or both, they are unusual among aid organizations because they focus on understanding what didn't work. For the last three years, they've published Failure Reports detailing their specific failures. The reports make an interesting read, not just for aid organizations, but for anyone trying to manage engineering teams."

"Spatially confined rigid membranes reorganize their morphology in response to the imposed constraints. A crumpled elastic sheet presents a complex pattern of random folds focusing the deformation energy while compressing a membrane resting on a soft foundation creates a regular pattern of sinusoidal wrinkles with a broad distribution of energy. … The physical model, exhibiting an analogy with parametric resonance in nonlinear oscillator, is a new theoretical toolkit to understand the morphology of various confined systems, such as coated materials or living tissues, e.g., wrinkled skin, internal structure of lungs, internal elastica of an artery, brain convolutions or formation of fingerprints. Moreover, it opens the way to new kind of microfabrication design of multiperiodic or chaotic (aperiodic) surface topography via self-organization."

"We present a method for the reconstruction of networks, based on the order of nodes visited by a stochastic branching process. Our algorithm reconstructs a network of minimal size that ensures consistency with the data. Crucially, we show that global consistency with the data can be achieved through purely local considerations, inferring the neighbourhood of each node in turn. The optimisation problem solved for each individual node can be reduced to a Set Covering Problem, which is known to be NP-hard but can be approximated well in practice. We then extend our approach to account for noisy data, based on the Minimum Description Length principle. We demonstrate our algorithms on synthetic data, generated by an SIR-like epidemiological model."

"Multi-task learning is an approach to machine learning, that learns a problem together with other related problems at the same time, using a shared representation. This often leads to a better model for the main task, because it allows the learner to use the commonality among the tasks. Therefore, multi-task learning is a kind of inductive transfer."

"Project Euler is a series of challenging mathematical/computer programming problems that will require more than just mathematical insights to solve. Although mathematics will help you arrive at elegant and efficient methods, the use of a computer and programming skills will be required to solve most problems.

The motivation for starting Project Euler, and its continuation, is to provide a platform for the inquiring mind to delve into unfamiliar areas and learn new concepts in a fun and recreational context."

"Performance tuning of Database Management Systems(DBMS) is both complex and challenging as it involves identifying and altering several key performance tuning parameters. The quality of tuning and the extent of performance enhancement achieved greatly depends on the skill and experience of the Database Administrator (DBA). As neural networks have the ability to adapt to dynamically changing inputs and also their ability to learn makes them ideal candidates for employing them for tuning purpose. In this paper, a novel tuning algorithm based on neural network estimated tuning parameters is presented. The key performance indicators are proactively monitored….The tuner alters these tuning parameters using the estimated values using a rate change computing algorithm. The preliminary results show that the proposed method is effective in improving the query response time for a variety of workload types."

"We study the degree to which a character string, $Q$, leaks details about itself any time it engages in comparison protocols with a strings provided by a querier, Bob, even if those protocols are cryptographically guaranteed to produce no additional information other than the scores that assess the degree to which $Q$ matches strings offered by Bob. We show that such scenarios allow Bob to play variants of the game of Mastermind with $Q$ so as to learn the complete identity of $Q$."

"Imagine the scene. You are eleven years old, and in the five minutes before the end of the lesson, your Maths teacher decides he should make his class more "fun" by introducing a "game". He explains that he is going to point at each pupil in turn and ask them to say the next number in sequence, starting from one. The "fun" part is that if the number is divisible by three, you instead say "Fizz" and if it is divisible by five you say "Buzz". So now your maths teacher is pointing at all of your classmates in turn, and they happily shout "one!", "two!", "Fizz!", "four!", "Buzz!"... until he very deliberately points at you, fixing you with a steely gaze... time stands still, your mouth dries up, your palms become sweatier and sweatier until you finally manage to croak "Fizz!". Doom is avoided, and the pointing finger moves on. Until the next time."

"Another problem I've often had (as recently as last semester!) is that my goals for students--what they're expected to be able to do when the semester is over--are often not well defined. When we don't have a sense of where we're going, our 15-week courses often fall apart somewhere around week 7 or so. But this should not be such an issue in high school."

"If you spent hundreds of hours playing Dungeons & Dragons in your youth, it turns out that time wasn't wasted. Three successful authors tell Suvudu that D&D gave them the experience points to write decent novels and stories."

"There are a number of ways to “solve” mazes but there’s a wide scope for you to be as straightforward or as clever as you like with this challenge (tip: I’d love to see some clever/silly solutions!). Your “solvable?” and “steps” methods could share algorithms or you might come up with alternate ways to be more efficient in each case. Good luck!"

"Columbia University's Computer Vision Labaratory is testing out a product called the BigShot, a digital camera intended to be taken apart and assembled by children, in order to remind them that yeah, someone actually designed and built this thing."

"Since that interview Howard has written a memoir, The man who forgot how to read. While he was writing it, I met him on the street one day, and he said he was feeling a bit miffed because he had wanted to write a memoir about several aspects of his life, but his editor wanted "the stroke, the whole stroke, and nothing but the stroke." In the book he has sneaked in something of his very interesting life, as well as what happened in the aftermath of the stroke. Between them, Howard and those who read his externalized thoughts back to him have written a wonderfully insightful and engaging book."

"Today I met an extraordinary Iranian man in the Geneva airport. He's written a 1000 page book in Arabic about (at least in part) his experiences in Cyprus. He approached me, asked if my English was really really good, sat next to me, and started pulling out several pages of hand-wrtten uppercase English. He had me go over them, improve them, write some new text as he read his Arabic in halting English, told me exactly how he wanted it to sound, pressed me to find shorter ways to say things, and finally got me to write out (for his next helper, no doubt) a clean copy of all my work...."

"By the time we tell you the right answer, it'll be too late."

"And so as the acting class is going on I just realized I just knew more about cinema than the other people in the class. I cared about cinema and they cared about themselves. But two, was actually at a certain point I just realized that I loved movies too much to simply appear in them. I wanted the movies to be my movies."

"The second driver is a consistent ignorance, apathy and a serious underestimation of the impact of new technology on the businesses that HR supports (particularly social technologies). Technology moves so quickly and for HR leaders and professionals it can seem so easy (and sometimes necessary) to remain in their comfort zone of policy creation and enforcement, employee relations, or compliance reporting."

"Fluency, in the fluency game paradigm, means you don’t learn, you teach; either you teach yourself, or you teach others. In doing so, you achieve a major milestone: all your skills and knowledge “come alive”, because they can readily jump from you into others. As living skills, they can spread throughout the people in your family, community, and work life. And your fluency in one skill signifies a fluency in self-teaching. With any new skill, you know just where to start, and where to go after that."

"At the moment, any of us who use web applications tend to spend a lot of time and effort populating application databases to make them useful to us. But when we do so, we tend to lose control of our data. They go into a private database schema, and what access we have to that depends entirely on what the application allows us to do. Sometimes there are reasonable ways to get the data back out (some kind of an XML dump perhaps), sometimes not. But always the application is in control. And linking data across applications is, in general, somewhere between hard and impossible.

FluidDB can change all that by leaving the user in control of his or her data, granting the application only such permissions as necessary or desired, and ensuring that the user retains flexability and control."

"There’s only one rule at hack day: build something you can demonstrate at the end of the event (Powerpoint slides don’t count). Importantly though, our hack days are not restricted to just our development team: anyone from the technology department can get involved, and we extend the invitation to other parts of the organisation as well. At the Guardian, this includes journalists.

For our first hack day, I put together a list of “tools for non-developers”—sites, services and software that could be used for hacking without programming knowledge as a pre-requisite. I’m now updating that list with recommendations from elsewhere. Here’s the list so far:"

Definitely some targets for genetic programming

"This series of games has a consistent size and format, and any four will fit neatly into The Games Collection Stand (Pin's part number 02705)."

[more raw material for Nudge project]

"In a previous post I laid out some guidelines for helping you create stitched panoramic photographs. To help further inspire experimentation of this technique, I scoured Flickr for some prime examples of what can be accomplished. Each image links back to the Flickr page and most contain large or even original size images if you want to take a closer look."

"And I worry that the idea that learning in relation to history can easily be kept within some type of bounds implies, to a degree, that the importance of history is its factual content. Generations of captive history students, face-down and drooling on their desks, indicate that approaches of this nature are not only unfortunately limited, but also a fatal blow to any intrinsic interest in examining historical/cultural change."

"I was amazed that the Journal of Cell Science has the wherewithal to publish an essay like this. Kudos to the author of the essay and the editors of the journal. "

"... The notanemployee example helps these people to know that some people are declaring that the types of relationships in business are no longer restricted to being hierarchical, and that we can make a choice. And, that independents can work together to help the people who hire them understand that they can get a competitive edge by not trying to control those people who choose to work with them. This makes for better relationships, more adaptability and flexibilty, a higher chance of success. This is a realistic pathway for people to begin to have the freedom to start building systems that are commons and peer-based. 20 years ago, it was unheard of for independent business people to work together closely on creating an ecolgy of trust, mutual respect, and learning/participation commons."

[indirect but key]
"Practice, practice, practice!
Don't confuse experience with expertise.
Don't trust folklore -- but learn it anyway.
Take nothing on faith. Own your methodology.
Drive your own education -- no one else will.
Reputation = Money. Build and protect your reputation.
Relentlessly gather resources, materials, and tools.
Establish your standards and ethics.
Avoid certifications that trivialize the craft.
Associate with demanding colleagues.
Write, speak, and always tell the truth as you see it."

"The Big Dream is about pushing yourself. Identifying places where your reach exceeds your grasp and then committing to tackle those things."

True for nearly every discipline beside "painting" as well. Including the ones where one may be "more successful". I know a lot of useless computer scientists, for example.

When it comes to graceful expertise, I am reminded of the intentional stance Ron Jeffries and Chet Hendrickson take in their work.

wow. Just... wow.

READ THIS. All of you.

"But I’m actually thinking that there are some lessons here that the mathematics community could learn from..."

via: boingboing

"the distance between any two degrees of uncertainty equals the amount of evidence you would need to go from one to the other"

"when the only tool you have is birth control, all your problems start to look like this thing you can maybe have safe sex with"

Discovery is the addiction that drives research -- it's the crackpipe hit, the rush, the thrill, that keeps us going through the down times and the plodding; but one of the best ways to alleviate the boredom and despondency that sets in between fixes is t

Conversation from Agile 2005 conference on Beginner's Mind and Promiscuous Pairing in software development teams. Worth thinking about in a research lab context.

"The CAPM is totally useless for predicting expected stock returns via the betas it generates, but a staple of MBA finance because it's so damn pretty."

Recommended for anybody who thinks they "know" "how" to "program" and yet still expect me to hire them, ever, clueless as they clearly are.

Impressive hands-on demo of Mootools Javascript libraries.

Formalizing good habits in collaborative workflows

Scary. Scarier than you think.

Hold that thought....