Monday, May 13, 2013

MoAR: a variety, this week.

this week's roundup is a little more substantial than in previous weeks .. but not by any extra effort on my part. i looked through the same channels and there was that much more to share, than usual ..


1. teaching computers like humans.

the human brain apparently has 86 billion neurons, which is a fairly complex computing system. so if it takes us years to learn how to walk, speak, and understand the baηach-alaοglu theorem, then maybe we should allow computers the same educational license.
"With Deep Learning, Ng says, you just give the system a lot of data “so it can discover by itself what some of the concepts in the world are.” Last year, one of his algorithms taught itself to recognize cats after scanning millions of images on the internet. The algorithm didn’t know the word “cat” — Ng had to supply that — but over time, it learned to identify the furry creatures we know as cats, all on its own."

~ from "The Man Behind the Google Brain: Andrew Ng and the Quest for the New AI" @wired

2. contrary evidence, of a few kinds.

in an earlier post i alluded to some statistics about the relative success of asian-american students, as measured by test scores. one may inquire as to what factors lead to these differences .. but apparently, it might not be tiger parenting after all.
"An associate professor of human development and family sciences at the University of Texas, Kim had been following more than 300 Asian-American families for a decade when the book came out. In March, she published her results; they will no doubt surprise Chua and her admirers. Children of parents whom Kim classified as “tiger” had lower academic achievement and attainment—and greater psychological maladjustment—and family alienation, than the kids of parents characterized as “supportive” or "easygoing.”"

~ from "Poor Little Tiger Cub" @slate
that said, this looks like a comparative study within a statistical population of asian-american families. what i'd like to know is whether kids in the u.s. with tiger parents always underperform against their peers with easy-going parents regardless of ethnicity.

in other news, apparently we live on a strange planet:
"We are now beginning to understand that nature seems to overwhelmingly prefer [planetary] systems that have multiple planets with orbits of less than 100 days," says Steve Vogt, astronomer at the University of California, Santa Cruz. "This is quite unlike our own solar system, where there is nothing with an orbit inside that of Mercury. So our solar system is, in some sense, a bit of a freak and not the most typical kind of system that Nature cooks up."

~ from "Our Very Normal Solar System Isn't Normal Anymore" @krulwich-wonders (npr)

3. messages of progress and hope.

for some people (like me) the difficulty with listening to the advice of others is that it seems painfully abstract until it becomes relevant to you .. by which time it's too late to follow the given advice.

for example, even if i were to travel back in time, meet my younger self in his first year of ph.d., somehow convince him that i'm his future self, and give him the best advice that i could, then it still wouldn't be clear to he-that-was-once-me would actually listen to me (that is, the me that i am now).

then again, i could just be a nutcase. so if you're still in your grad school years, then this webpage might be worth a read:
"While you're writing or building, you might feel like your idea sucks and want to stop it early. Unless you're totally convinced after some more thought that your idea is unworkable or totally impractical, keep going. That feeling of discomfort you have might turn out to be your core research problem, and you'll feel better once you finish a draft and identify the unsolved work ahead of you. Even if you're not convinced your idea is great, you should continue. When you present the introduction to your advisor, they might think of something completely different and more exciting. Your crappy first draft will end up being a wonderful first seed for the idea you'll actually start working on!"

~ from "The N=1 guide to grad school"
as for the next article, it's a more serious take on popular science writing than, say, the up goer 5 editor from a while back. who knows? it might even survive the usual lifespan of a single news cycle.
"Your challenge is to write as though explaining your work to an intelligent, capable relative who is totally naïve to your field and its importance. That means more than avoiding jargon; it also means providing some context upfront so readers can appreciate why particular findings might matter."

~ from "Science, Meet The People" @13.7 (npr)

4. trolling, mathematically.

my guess is that the following quote was taken out of context, or that the quoted party was writing an email that wasn't re-read for clarity.
"O’Neil, however, holds Mοlchizuki accountable, saying that his refusal to cooperate places an unfair burden on his colleagues. “You don’t get to say you’ve proved something if you haven’t explained it,” she says. “A proof is a social construct. If the community doesn’t understand it, you haven’t done your job.”"

~ from "The Paradox of the Proof" @projectwordsworth
as far as i've read, Mοlchizuki did not actually submit these articles for publication in a peer-reviewed journal. they were posted on his website, a web-crawler happened to index them into a search engine, and the right search happened to bring the work to the attention of experts.

the practice is non-standard, yes, but non-binding. yes, he claims he has a proof, but he's not forcing others to acknowledge that his proof is correct or even to read it .. so the comment of this being an "unfair burden on his colleagues" sounds ridiculous to me. by that reckoning, i am being an unfair burden because my papers on analysis are not written in an expository way for algebraists to understand immediately.

if the quoted party feels so strongly about having a readable, easily verifiable proof, then (s)he should either work on the same problem to find a better proof .. or quit complaining about papers that (s)he probably wouldn't have read anyway.

that aside, project wordsworth strikes me as an interesting social experiment; they're worth a few mouse clicks.

5. about online education, mercifully put into sound (text?) bites.

in case you're sick of reading about these thing, i'll keep it short.
"Offering MOOCs through edX is hardly free. There are options available to institutions that want to build their own courses on the edX platform at no charge, but for partners who want help developing their courses, edX charges a base rate of $\$$250,000 per course, then $\$$50,000 for each additional time that course is offered; edX also takes a cut of any revenue the course generates."

~ from "Why Some Colleges Are Saying No to MOOC Deals, at Least for Now" @chronicle.

"In spite of our admiration for your ability to lecture in such an engaging way to such a large audience," the letter's authors write, "we believe that having a scholar teach and engage with his or her own students is far superior to having those students watch a video of another scholar engaging his or her students."

~ from "Why Professors at San Jose State Won't Use a Harvard Professor's MOOC" @chronicle

"Online learning startup Coursera on Wednesday announced a partnership with Chegg, a student hub for various educational tools and materials, as well as five publishers to offer students free textbooks during their courses."

~ from "Coursera leaps another online learning hurdle" @tnw

"This sounds good and I'm sure it works with some of the students, but I don't believe it would work on a large scale. One reason is that students discount the future so heavily that I don't believe that students will choose to read a textbook instead of going to work or an extracurricular activity. I think students would choose to enjoy the immediate benefits when the costs come later. "

~ from "Sir Ken Robinson on Education" @mikeroe (via a.isakov)

No comments: