Sign in or Join FriendFeed
FriendFeed is the easiest way to share online. Learn more »

Michael Nielsen › Comments

Michael Nielsen
Eerie Video Shows Masdar City--The Sustainable City Of The Future--Has No One In It | Co.Exist | ideas impact - http://www.fastcoexist.com/3035446...
Another example of the difficulties of central planning. - Michael Nielsen
Michael Nielsen
An Oral History of Burning Man, the Biggest, Weirdest, Most Clothing-Optional Desert Carnival on the Planet | Events | OutsideOnline.com - http://www.outsideonline.com/outdoor...
Nice history of the early years. - Michael Nielsen
Michael Nielsen
So you wanna try Deep Learning? - Exchangeable random experiments - http://snippyhollow.github.io/blog...
Fun overview of some of what we know about deep learning, with plenty of useful links (and nice Theano source code) - Michael Nielsen
Michael Nielsen
Lester Dent's Master Plot Formula - http://www.multiverse.org/fora...
More in the Moorcock vein. It's easy to imagine the reaction of the critic, holding their nose at writing to formula. But you can turn that around, regarding Dent (and, more plausibly, Moorcock) as a student and theoretician of structure. And that's a pretty powerful point of view. Of course, word-by-word Dent is a poor writer, suffering from an insufficiency of revision, and an overuse of cliche. But I rather suspect that with a little more revision, a little less tolerance of cliche, he would have been pretty good. (Cory Doctorow suffers some of the same problems, and would also be excellent if he simply revised more, eliminated weaker material, and had less tolerance of cliche.) - Michael Nielsen
Michael Nielsen
How to Write a Book in Three Days: Lessons from Michael Moorcock - http://www.wetasphalt.com/content...
Fascinating both intrinsically, and for the commentary. The commentary first: part of the interest is from people who desire an easy way to write (or, more accurately, to have written). But there is also clearly a genuine interest on the part of many: what does this guy know that I don't about storytelling? You may not like Moorcock's writing, you may even regard it as shabby, but it does have its moments. And that's more than one can say for the writing of many critics. My condensation of the article: study structure. And work to find story structures that work. - Michael Nielsen
Michael Nielsen
Kevin Kelly interviews Brian Eno. Slow to get going, but fascinating. Eno proposes "process, not product", says that it's his "ease of seduction" that means he often gets things first, talks about putting more "Africa" into computers, and generally makes many interesting comments. - Michael Nielsen
Michael Nielsen
Conveys well how much force of personality and intellect Sontag brought to her writing. - Michael Nielsen
Michael Nielsen
The Truth of Fact, the Truth of Feeling by Ted Chiang — Subterranean Press - http://subterraneanpress.com/magazin...
Ted Chiang on the impact of lifelogging, and the expansion of self it will cause. - Michael Nielsen
Michael Nielsen
The Zen of Gradient Descent | Moody Rd - http://mrtz.org/blog...
Nice article explaining a principled way of deriving Nesterov's method. Also makes the interesting point that Nesterov is more brittle than standard gradient descent, and thus may be less useful in practice. - Michael Nielsen
Michael Nielsen
Have some coffee — TheLi.st @ Medium — Medium - https://medium.com/theli-s...
I've heard people talk a lot about the "need for mentors". It's a slightly odd obsession. This little essay has a lot that's smart to say about the notion: (1) It's a mistake to think in terms of "mentors", just think in terms of what you want to learn, and who you want to learn from; (2) Approach people with very specific questions; (3) Pick people who you think would be particularly well suited to answer your question; (4) Think carefully about how to structure the conversation (without going overboard); and (5) Make sure to followup, simply because people like to hear that they've done something useful. - Michael Nielsen
Michael Nielsen
The problems with object-oriented coding - http://mollyrocket.com/casey...
A thoughtful criticism of object-oriented programming, from an experienced programmer. The analogous argument for English would be something like: don't start by thinking about the nouns. "The fallacy of “object-oriented programming” is exactly that: that code is at all “object-oriented”. It isn’t. Code is procedurally oriented, and the “objects” are simply constructs that arise that allow procedures to be reused. So if you just let that happen instead of trying to force everything to work backwards, programming becomes immensely more pleasant." - Michael Nielsen
Michael Nielsen
Shadow DOM — Acko.net - http://acko.net/blog...
Nice summary of some of the problems in the design of the web. - Michael Nielsen
Michael Nielsen
The Ultimate Guide to Making Your Wacom Tablet Work on Windows 7 - http://viziblr.com/news...
The Wacom Bamboo tablet has lots of problems when installed by default on Windows 7. This page is a pretty good guide to how to fix those problems. It's working much better now! - Michael Nielsen
Michael Nielsen
▶ Game Changers: Trip Hawkins - http://www.youtube.com/watch...
▶ Game Changers: Trip Hawkins
Play
Hawkins on where games (and in particular educational) games are going. - Michael Nielsen
Michael Nielsen
Build Your Own Adobe Creative Suite with Free and Cheap Software - http://lifehacker.com/5976725...
undefined - Michael Nielsen
Michael Nielsen
Bitcoin As Protocol | Union Square Ventures - http://www.usv.com/posts...
Nice abstraction of Bitcoin as a way of maintaining a distributed, public ledger - Michael Nielsen
Michael Nielsen
Leonard Kleinrock - Wikipedia, the free encyclopedia - http://en.wikipedia.org/wiki...
"The first message on the ARPANET was sent by UCLA student programmer Charley Kline, at 10:30 p.m, on October 29, 1969 from Boelter Hall 3420, the school's main building.[6] Supervised by Kleinrock, Kline transmitted from the university's SDS Sigma 7 host computer to the Stanford Research Institute's SDS 940 host computer. The message text was the word "login"; the "l" and the "o" letters were transmitted, but the system then crashed. Hence, the literal first message over the ARPANET was "lo". About an hour later, having recovered from the crash, the SDS Sigma 7 computer effected a full "login". The first permanent ARPANET link was established on November 21, 1969, between the IMP at UCLA and the IMP at the Stanford Research Institute. By December 5, 1969, the entire four-node network was established.[7]" - Michael Nielsen
Michael Nielsen
HyperCard: What Could Have Been - http://www.wired.com/gadgets...
Atkinson's attitude to the web: "Atkinson feels that if only he'd realized separate cards and stacks could be linked on different people's machines through the Net -- instead of cards and stacks on a particular machine -- he would have created the first Internet browser." - Michael Nielsen
Michael Nielsen
PLOS ONE: Big Science vs. Little Science: How Scientific Impact Scales with Funding - http://www.plosone.org/article...
Some evidence to suggest that there's little payoff from larger grants. Lots of caveats, of course. Interesting references, too. - Michael Nielsen
Michael Nielsen
Auctions and bidding: a guide for computer scientists - http://www.sci.brooklyn.cuny.edu/~parson...
A useful review. - Michael Nielsen
Michael Nielsen
Research Blog: Improving Photo Search: A Step Across the Semantic Gap - http://googleresearch.blogspot.ca/2013...
On Google's use of Hinton-Ng-Dean-LeCun's work to do automatic photo tagging. - Michael Nielsen
Michael Nielsen
Evaluation of Pooling Operations in Convolutional Architectures for Object Recognition - http://www.ais.uni-bonn.de/papers...
Reports two experiments in which max pooling handily beat out subsampling. - Michael Nielsen
Michael Nielsen
Agre on what it means to be critical: to be aware of one's own biases, and the cultural assumptions one brings to something. He talks about close reading an AI text, and trying to understand all the sociological background. - Michael Nielsen
Michael Nielsen
Tera-scale deep learning - Quoc V. Le on Vimeo - http://vimeo.com/52332329
On the 2012 Google-Stanford deep learning paper. - Michael Nielsen
Michael Nielsen
An Analysis of Single-Layer Networks in Unsupervised Feature Learning - http://www.stanford.edu/~acoate...
More learned features helps a lot; smaller stride length helps a lot. Larger local receptive fields barely helps at all. - Michael Nielsen
Michael Nielsen
Classification datasets results - http://rodrigob.github.io/are_we_...
Results for MNIST, CIFAR-10, CIFAR-100, STL-10 and SVHN - Michael Nielsen
Michael Nielsen
Best practices for convolutional neural networks applied to visual document analysis - http://citeseer.ist.psu.edu/viewdoc...
Simard, Steinkraus, and Platt on the value of image distortion and convolutional networks. - Michael Nielsen
Michael Nielsen
Good interview with an online troll: "I am lucky to have not had people who liked me for who I was when I was truly awful" - Michael Nielsen
Michael Nielsen
How My Neural Net Sees Blackboards | Christopher Olah's Blog - http://christopherolah.wordpress.com/2013...
Chris uses neural nets to clean up handwritten blackboard images. - Michael Nielsen
Michael Nielsen
Convolutional Deep Belief Networks for Scalable Unsupervised Learning of Hierarchical Representations - videolectures.net - http://videolectures.net/icml09_...
Honglak Lee's 2009 paper on hierarchical infererence in convolutional DBNs. - Michael Nielsen
Other ways to read this feed:Feed readerFacebook