David II
[Most Recent Entries]
[Calendar View]
[Friends]
Below are the 20 most recent journal entries recorded in
util's LiveJournal:
[ << Previous 20 ]
| Saturday, November 21st, 2009 | | 12:05 pm |
InspiringFingernail clippings wait to be swept, coins to be spent. The jar of jelly's half empty, the peanut butter half full. The bread bag is empty. Sunlight warms the arm and leg. The foot warms the floor. | | Tuesday, November 17th, 2009 | | 5:47 am |
Reid considers raising Medicare tax for high earners "Centrist Democrats prefer to tax costly 'Cadillac' insurance as a way to lower health costs." What is the rationale for saying progressively taxing health insurance will lower costs? I guess they are not arguing that it will lead to richer people spending more out of pocket and so becoming smarter consumers. What then? ( *) PBS NewsHour -- no answer here although they do seem right to note that not taxing health insurance is regressive -- the more employer provided health insurance you get, the more of a tax benefit you're getting. It seems both interviewees want to see the proposed tax adjusted so that those who get more care through their "Cadillac insurance" pay less. Would that have the desired effect (or the opposite)? Say it is mostly the wealthy who get these plans. This increases the share of GDP going to healthcare spending. Is that a bad thing for society though? Somehow it seems like less obvious of a bad thing if that extra spending is coming through the paying of what amount to bonuses. Say that you're worried about the burden of clothes spending on the nation because clothing expenditures make up a larger fraction of GDP than you think they should. Now you find a noticeable chunk of that spending is coming from the superwealthy buying haute couture. Is a rationale response then taxing high end clothing? Is this a fair analogy? // Just read this: Tyler Cowen: What should we do instead of the Obama health reform bill? (The commenters seem pretty negative on their merits overall for what that's worth.) | | Sunday, November 15th, 2009 | | 11:16 pm |
My Squeak Smalltalk response to Eric's "Trying to Start a Programming Meme". name := FillInTheBlankMorph request: 'What''s your name?'.
age := FillInTheBlankMorph request: 'How many years old are you?'.
Transcript open.
1 to: age asInteger do: [:i |
Transcript show: (i asString , ') Hello, ' , name); cr].
Transcript endEntry I'm guessing there's probably a more idiomatic way to do parts of this. Hm. I think it's been a few years at least since I wrote anything in Squeak. R: do.call(function() {
cat("What's your name: ")
name <- readLines(n=1)
cat("How many years old are you?: ")
age <- as.numeric(readLines(n=1))
cat(paste(1:age, ") Hello, ", name, "\n", sep="", collapse=""))}, list()) | | 12:24 pm |
| | Sunday, November 8th, 2009 | | 4:42 pm |
alternatives to life expectancy
I wonder how the published estimates of life expectancy are calculated. It seems they must incorporate estimates of the advance of medical technology, the performance of the economy, probability of natural and human-started pandemics, probability of world war, etc, all of which we can know only very imperfectly making the estimate perhaps very imperfect. And if they don't incorporate such, then it seems the estimates are worse. How accurate do the estimates tend to be? Picking someone born 50 years ago, how much has the official estimate of their expected total number of years to live varied per year? (Well, I guess we'll have to look at the people who died. Problem: Assuming the distribution of your choice, taking the life expectancy estimates momentarily as the true means of that distribution conditioned on attained age, now step back and estimate the divergence between the given estimate and the true one using the corresponding estimates of the probability of surviving one more year given years attained.) Would it make more sense to instead compare mortality rates per age group? No forward looking estimation is required then. One example is infant mortality, but I've heard the definition is not consistent. How would you weight age groups? How well would it work generally to account for demographics to take 25 year olds as baseline? | | Sunday, September 20th, 2009 | | 1:26 am |
Tired, sired, no longer wired, bought the farm, kicked Nantucket, sucked a goat, walked a bucket with a boat, a rote, a vote, a mote, emote, demote, promote... ( Read more... ) | | Tuesday, September 15th, 2009 | | 9:33 am |
I really do not like the way Reality Czech tastes. I drank it all any way though. The Japanese blue grass singer was fun, and he really was blue due to the lighting. I was sorry he didn't seem to do any of his own stuff though. Hopefully his own life is just too happy to serve as grounds for the blues. | | Friday, September 11th, 2009 | | 9:58 pm |
I finished The curious incident of the dog in the night-time just now. I liked it but am not completely firm in my liking of it. Also, is it necessary to find which side of the triangle is longest? Just find one side whose square is the sum other's squares and you're done, aren't you? (Or conversely, if you exhaust the three possibilities, you've proved the statement to be false.) The remains of the dog. Yesterday morning I changed a light bulb in the bathroom. In the evening, I bought some toothpaste. | | Monday, September 7th, 2009 | | 9:33 pm |
Night pizza to fill a hole and a sack a snack a meal not so healthy but something to get me thinking just in time for sleep. Eat eat. | | Saturday, August 1st, 2009 | | 8:35 am |
| | Friday, July 24th, 2009 | | 6:12 pm |
| | Tuesday, July 14th, 2009 | | 7:55 pm |
| | Monday, July 13th, 2009 | | 8:56 pm |
Mother Jones: San Francisco Rejects American Apparel: Has the Fight Against Chain Stores Gone Too Far?American Apparel Storefront TalkI guess there have been real instances where for racist reasons, neighbors used community pressure to keep new businesses or residents out. I guess the people against American Apparel would not recognize the legitimacy of these uses of community pressure. So, I guess they recognize that majority opinion does not always make right. Say that it is the state enforcing a racist law to keep people segregated, but that it is not popular. Is this worse than if it were popular? Why? I wonder when the first of the stores to give the area its unique character worth protecting came in. Did they displace something else? Would they have had a chance to get a foothold if the same laws were in place then as are here now? How much tax revenue would have been collected if American Apparel had gone in? What useful (and unuseful) things would've been done with it? Is someone unemployed now who would've been employed? How does their well-being balance against the aesthetic value of the place? How likely do you think it is that you are wrong? I would say there's exactly a 35.76203234% chance that I am wrong, but I may be off about that too. | | Sunday, July 12th, 2009 | | 9:36 am |
ToneMatrix. Beautiful! How hard is it to generate sound data in memory in Flash? Is it possible to do this in modern Java installations without scary security dialogs?... AudioTool. Holy moly. Minnesota Population Center: "The MPC is one of the world's leading developers of demographic data resources. We provide population data to thousands of researchers, policymakers, teachers, and students. All MPC data are available free over the internet." W00t! A Tried And Tested Method - The Longcut. I wonder how long it took to create this. SimpleComplexity: Top 15 Infographic Videos on the Web – Part 2jQuery Visualize Plugin: Accessible Charts & Graphs from Table Elements using HTML 5 Canvas: "We have tested this plugin in the following browsers: IE6, IE7, IE8, Firefox 2, Firefox 3.5, Safari 3 and 4, Opera 9." parseScripts: "'parseScripts' is a simple abstraction you can use the simplify the process of enhancing JavaScript syntax as you so desire." Hee hee hee! Child-like glee. I saw someone say something similar to this somewhere: "What would I be without what I have learned from society? What would I know? What language would I speak? How long would I survive at all?" This thought makes me think of this thought: How fundamental is information?The same conclusion -- the meaning of a book depends on its reader -- holds for another book, DNA. DNA is read by cell's machinery, which is far more complex than the DNA itself. Without a competent reader DNA means nothing. A magnetic tape with a code for a tape driver is meaningless without a driver to read it. It is often said that DNA encodes all that is in a cell. That is not correct [Feitelson][Bethell]. It is a cell that encodes in its own machinery what it is in the cell. Combining parents' DNA does not make an embryo by itself, in a flash of a green light. The combined DNA must be a part of an egg cell, whose division and specialization make an embryo. How Should We Estimate Public Opinion in the States? [PDF] from Jeffery Lax// An article from 2006: "More than a third of the American public suspects that federal officials assisted in the 9/11 terrorist attacks or took no action to stop them so the United States could go to war in the Middle East, according to a new Scripps Howard/Ohio University poll." What's that number look like now? // http://www.goodreads.com/ -- Hm. Scratch another idea. // Detexify2 - LaTeX symbol classifier. Wow. So much cool stuff in the world today. | | Saturday, July 11th, 2009 | | 12:15 pm |
Definitely a failure, but I'll leave it on my wall for a little while. ( Read more... )I guess I should I bother using an RSS reader. For now, I randomly browse the sites I remember liking at one point. Going back to Dolores Labs led me to this: Crowdsifter: More Efficient Content FilteringUsing Crowdsifter with an average of 3.93 workers per image we achieve the same possible minimum error rate as majority voting in raw AMT with 9 workers per image. We do this by controlling worker quality by keeping track of their judgments. And if we have a “expert” evaluated gold standard of what is pornographic, then we can keep track of which workers are doing a good job or a bad job. On non-gold standard images we weight workers’ judgments based on how well we trust their judgment to reflect our standard of porn. Without these controls, majority voting in raw AMT is vulnerable to the many scammers that lurk there.
For images where obscenity is particularly ambiguous, we can allocate more workers. This results in a better sampling of whether an image is obscene. Some images don’t need many judges to accurately determine if they are pornographic. We can determine which images are easily classifiable as porn by sampling a group of workers and checking whether they all agree. Using too many judges per image can become prohibitively costly. It is important to have this scheme so we can dynamically allocate workers. Raw AMT is both wasteful and inefficient, applying many judgments to easy items, while not using enough judgments for hard items. Hmmmmmmm. That led me to Lukas Biewald's link to an article on boosting, which led me to Inductio ex Machina... IJ Good is dead. The Estimation of Probabilities: An Essay on Modern Bayesian Methods. (Whoa. Currently way more expensive on Amazon -- $20 vs $75 or more!) No Free Lunch Theorems -- these sound like they could be important. Are they important? I think I bumped into them before. Peter Norvig: All we want are the facts, ma'am. Tom Mitchell is posting new chapters for Machine Learning online!. Hm, I think I've read this before, but maybe it's worth another look: A Solution to Plato's Problem: The Latent Semantic Analysis Theory of Acquisition, Induction and Representation of Knowledge -- from The Book That Changed My Life. Graphical demonstration of Jensen's inequality. Helping myself remember: A function g(x) is called convex if g(p * x + (1 - p) * y) <= p * g(x) + (1 - p) * g(y) for all x and y and 0 < p < 1. As a function of p (p * x + (1 - p) * y, p * g(x) + (1 - p) * g(y)) gives the line segment interpolating between (x, g(x)) and (y, g(y)). So, a convex function is one for which the function is below the line segment joining any two points on the function. Or I guess if you say S = {(x, y): y >= g(x)}, then g is convex if p * z1 + (1 - p) * z2 is in S for all z1, z2 in S and 0 < p < 1. Well, I guess I've not added anything. Anyway, after looking at Casella and Berger, a non-visual proof for E[g(X)] >= g(E[X]). Say that a + b*x is a tangent line to g(x) at E[X]. Then g(x) >= a + b*x for all x, which means E[g(X)] >= E[a + b*X] = a + b*E[X] = g(E[X]). Tada. On Discriminative vs. Generative classifiers: A comparison of logistic regression and naive BayesComment on "On Discriminative vs. Generative Classifiers" [PDF]Discussion on Generative vs. Discriminative modelsCity Public Star Party: Saturday 11th of July 2009 08:30:00 PM$9.99 Daily from Fri., July 10 until Thu., July 16; Price: $8-$10.50 // Introduction to Cybernetics by W. Ross Ashby -- the full book online! From AI or IA?. Paul Meehl's Clinical versus Statistical Prediction [PDF] online! The Earth Is Round (p < 0.05) Also: The Difference Between "Significant" and "Not Significant" is not Itself Statistically Significant And also: Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology [PDF] by Paul Meehl: Since the null hypothesis is quasi-always false, tables summarizing research
in terms of patterns of "significant differences" are little more than complex, causally
uninterpretable outcomes of statistical power functions. Multiple paths to estimating
numerical point values ("consistency tests") are better, even if approximate with rough
tolerances; and lacking this, ranges, orderings, second-order differences, curve peaks and valleys, and function forms should be used. // Woollett's Maxima By Example | | Tuesday, July 7th, 2009 | | 7:19 pm |
| | Sunday, July 5th, 2009 | | 2:22 pm |
Thursday I tore apart a box. Friday I bought some more sticky tack. I actually did have an idea for what I wanted. This isn't really it: ( Read more... )Nearly 2:30! I planned to accomplish a lot more this weekend than I have, but I guess I've accomplished some things I didn't plan on too. // Hm. It is better to smile. // "Wir starteten in La Guardia, New York, mit dreistündiger Verspätung infolge Schneestürmen." starten -- to start, infolge -- due to (We started in La Guardia, New York, with a three hour delay to snow storms.) "Ich richtete mich sofort zum Schlafen, es war Nacht." (I set about falling asleep right away. It was night.) | | Saturday, July 4th, 2009 | | 7:55 pm |
| | Thursday, July 2nd, 2009 | | 7:13 pm |
Informing Ourselves To Death "We no longer know, as the Middle Ages did, where we come from, and where we are going, or why." Thinking you know something and knowing it are not the same thing! I guess I make a fool of myself by not being charitable, but I'm short on sleep and so not charitable. "Did Iraq invade Kuwait because of a lack of information? If a hideous war should ensue between Iraq and the U.S., will it happen because of a lack of information? If children die of starvation in Ethiopia, does it occur because of a lack of information? Does racism in South Africa exist because of a lack of information? If criminals roam the streets of New York City, do they do so because of a lack of information?" Yes. Yes. Yes. Yes. Yes. "If you and your spouse are unhappy together, and end your marriage in divorce, will it happen because of a lack of information? If your children misbehave and bring shame to your family, does it happen because of a lack of information? If someone in your family has a mental breakdown, will it happen because of a lack of information?" Yes. Yes. Yes. Yes. I think if you don't answer "yes" to any of these, you have a have certain lack of imagination. Now for a nap. | | Tuesday, June 30th, 2009 | | 12:07 am |
|
[ << Previous 20 ]
|