Mike Driver
ā
Stranger Things
Discoholic šŖ©
Sade Olutola
Origami Around
Cosmic Funnies
almost home
Kiana Khansmith
Game of Thrones Daily
wallacepolsom
d e v o n
hello vonnie
tannertan36
JVL
taylor price
macklin celebrini has autism
$LAYYYTER
seen from United States
seen from Indonesia
seen from Pakistan
seen from United States
seen from United States
seen from Italy
seen from United States
seen from TĆ¼rkiye
seen from Venezuela
seen from Venezuela
seen from Germany
seen from United States
seen from Finland
seen from Venezuela
seen from China
seen from United States
seen from Canada
seen from United States
seen from Malaysia
seen from United States
@more-whales
I generally do not read unsolicited theory proposals sent to me. It's entirely pointless to send them to me. Do not do it. There is no turn of phrase that will change anything about it. The moment you send me your theory you have already disqualified. Because I cannot think of any reason an actual researcher would do that. A real researcher would send their breakthrough insights straight to Ed Witten.
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
Free to watch ā¢ No registration required ā¢ HD streaming
The original Star Wars trilogy tells the story of a ragtag band of misfits, many of whom are adept at manipulating a force pervading everyday matter, who ally to mount an insurrection against the established order and help destroy a giant, partially built beam machine. The history of American solid state physics, as chronicled in these chapters, followed much the same plot.
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
Free to watch ā¢ No registration required ā¢ HD streaming
itās been years and I am still not over this
It turns out that my dictation software might prefer my mediocre webcam microphone to a dedicated condenser mic. Thereās a very audible difference in quality, but I guess that the webcam in failing to record sound-in-general very faithfully manages to reject a lot of background noise (and maybe confusing detail in my voice itself), at least enough so that it passes the automatic microphone āqualityā check while the other fails every time. It kind of makes sense given that the webcam mic has one very specific purpose and limited resources. But when it comes to actually taking dictation, it doesnāt seem to do any better, at least on a first impression. Iād like to do more serious experiments on this.
So if you have any texts you want to hear me read out loud, Iām taking suggestions.
The results are in. Speaking clearly and slowly but still relatively naturally, the two microphones give statistically indistinguishable error rates of 10-15% (with a sample size of about 300 words, which is all Iām going to count up by hand). Apparently this is typical, and I donāt know how anyone puts up with itālike, you can expect to get an error in the process of correcting the error you expect to get in each sentence.
If I speak at half that speed, pausing between words and emphasizing consonants that seem to give it trouble (pretty much all of them), theĀ bad (āgoodā) mic gets an error rate of ~7.5%, and the good one is down to about half that. So NatSpeak doesnāt even know what it wants. (Well, but the 95% confidence intervals still overlap.)
Itās really obnoxious to have to talk like that, but at least itās somewhat workable. (The software asks me to speak naturally, claiming that it will learn. This has not been my experience. Maybe I should still note that the profile for the good mic has had a little more training than the other.)
No great AI comedy, but my favorite error is probably āKrakatoaā for āgravitasā. Most impressive success might be āThomistsā.
I finally upgraded to NaturallySpeaking 15, the 2016 deep learning version (my old version had come out in 2012).Ā So far I have less than a 2% error rateĀ with no training at all.. Still interacts with tumbler weirdly, though less so (seems to occasionally double spaces aand periods oh and apparently other characters).
I just updated to the latest AutoCAD and had to post this. Iām sorry
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
Free to watch ā¢ No registration required ā¢ HD streaming
The thing about the Sokal 2 thing is that, out of the seven accepted articles, three of them had fabricated data. So the only conclusion I can really draw from those three is that these three journals donāt look too hard to see if data is fabricated:
Gender, Place and Culture
Sex Roles
Sexuality and Culture
If you want another list of journals that donāt look too closely at whether the results are real, you could look at the list of retracted papers by Anil Potti.
PNAS
Lancet Oncology
Nature Medicine
JAMA
Journal of Clinical Oncology
Blood
New England Journal of Medicine
PLoS One
Clinical Cancer Research
The Nature MedicineĀ one was published 2006. It was retracted in 2011. Hereās the retraction statement:
Nature Medicine:
We wish to retract this article because we have been unable to reproduce certain crucial experiments showing validation of signatures for predicting response to chemotherapies, including docetaxel and topotecan. Although we believe that the underlying approach to developing predictive signatures is valid, a corruption of several validation data sets precludes conclusions regarding these signatures. As these results are fundamental to the conclusions of the paper, we formally retract the paper. We deeply regret the impact of this action on the work of other investigators.
Why wasnāt it retracted following the letter to Nature Medicine by Coombes Wang and Baggerly, which they published in 2007? They showed that you canāt get any of the results in the paper by following the data analysis methodology described in the paper. Some quotes:
Baggerly, Wang and Coombes:
1. We cannot reproduce their selection of cell lines.Ā ā¦
2. The lists of genes initially reported in the supplementary information on the Nature Medicine website are wrong because of an āoff-by-oneā indexing error. ā¦Ā
3. Using their software and lists of cell lines, we reproduced their published heatmaps for six out of seven drugs. (We could not reproduce the heatmap for cytoxan.) However, after correcting for the off-by-one error, we matched the reported gene lists exactly for only three out of seven drugs (Supplementary Report 9). The other lists contain outliers. ā¦Ā
4. For docetaxel, their software yields only 31 of their 50 reported genes. ā¦
5. Their software does not maintain the independence of training and test sets, and the test data alter the model. ā¦
6. The interaction between point 4 (accidentally including genes from Chang et al.2 whose expression levels separate the test set responders from nonresponders) and point 5 (combining training and testing data when choosing gene weights) can produce 'better than chanceā predictions in the wrong direction. ā¦
I think the reason, I think what happened between 2007 when this letter was published and 2011 when the paper was retracted, is that it was revealed that Anil Potti had lied about being a Rhodes scholar on his job application. That was 2010.
I donāt think thereās any way to conclude from Sokal 2 that Gender, Place and Culture is a stupid journal but Nature Medicine is fine. They both publish made-up results and donāt look too hard at them. (and if you want to conclude anything else about a journal, you have to submit an article without made-up data, like the Sokal2 article published by Fat Studies, which was very silly.)
Starting around 1999 Jan Hendrik Schƶn published a series of high-profile (Science, Nature...) papers in condensed matter physics with fabricated data. The peer review system, responsibilities of coauthors, and general academic environment werenāt (arenāt) designed to catch someone acting in bad faith. Any criticism he got was just another idea for a āmeasurementā addressing the criticism. (It took about two years for people to start noticing, and largely then only because he was replotting the same fabricated data with different labels.)
Doug Natelson published a detailed retrospective on the Schƶn scandal that I linked a couple weeks ago, which seems relevant again. The retrospective has a lot of interesting detail from someone in the same field who overlapped with Schƶn as a postdoc at Lucent Technologies.
(Again, also interesting are the report from Lucentās investigation, and the comments on Natelsonās blog where he posted it.)
As I mentioned previously, I've realized in recent weeks that many current students out there may never have heard of Jan Hendrik Schƶn, and that seems wrong, a missed opportunity for a cautionary tale about responsible conduct of research. Ā It's also a story that gives a flavor of the time and touches on other issues still current today - faddishness and competitiveness in top-level science, the allure of glossy publications, etc.
See also the report from Lucentās investigation, and the comments on Prof Natelsonās blog where he posted this.
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
Free to watch ā¢ No registration required ā¢ HD streaming
How a hidden coding error fueled a seven-year dispute between two of condensed matterās top theorists.
In his 20-minute presentation, attended by many of the biggest names in condensed-matter theory, Chandler was essentially declaring that the Princeton team had gotten it wrong. āIt was a matter of people saying, āWho are you going to believe, Chandler or Debenedetti?āā recalls Angell. āAnd Chandler carried the bigger stick.ā
Over the next seven years, the perplexing discrepancy would ignite a bitter conflict, with junior scientists caught in the crossfire. At stake were not only the reputations of the two groups but also a peculiar theory that sought to explain some of waterās deepest and most enduring mysteries. Earlier this year, the dispute was finally settled. And as it turns out, the entire ordeal was the result of botched code.
