Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
Math

+ - BayesianBracket 2013: A Bayesian March Madness Analysis->

Submitted by
SocratesJedi
SocratesJedi writes "Last year, I wrote about the fun I was having using graph theory to predict NCAA tournament outcomes. This year, I instead modeled both the game outcome predictor and the entire tournament in the framework of Bayesian statistics. In addition to each team’s Win-Loss record, individual player statistics were also used as part of the predictor. Validation on a random sample of 200 regular-season games suggests about 75% per-game accuracy. Interestingly, Bayesian analysis suggests that the most likely tournament outcome isn’t necessarily the best bracket choice, if prediction of later rounds are preferentially weighted during scoring.

To do all of this, I used GeNIe and SMILE, two user-friendly, open-source Bayesian statistics packages that might be of interest to some Slashdotters.

[Disclosure: I am NOT affiliated with GeNIe or SMILE in any way; I just thought they were useful!]."

Link to Original Source
Medicine

+ - Got an Insanely Great Condom for Bill & Melinda Gates?

Submitted by theodp
theodp (442580) writes "GeekWire reports that the Gates Foundation needs your help to design the next-gen rubber to help prevent STDs and unwanted pregnancy. As part of the foundation’s Grand Challenges Explorations initiative, people who have ideas to improve the use and effectiveness of condoms are encouraged to apply for a research grant, ideally to create a condom 'that men would prefer to no condom.' Last July, Melinda Gates pledged $560 million for contraception initiatives to address the health and social problems brought on by high rates of unplanned pregnancy."
Facebook

+ - 'Catch Me if You Can' Con Man Warns of Facebook Fraud

Submitted by
Hugh Pickens writes
Hugh Pickens writes writes "Christopher Williams reports that Frank Abagnale, the celebrated con man, confidence trickster, check forger, impostor, and escape artist portrayed in the Steven Spielberg film 'Catch Me If You Can,' warns that data posted on Facebook is an open invitation to identity thieves. "If you tell me your date of birth and where you're born [on Facebook] I'm 98 per cent [of the way] to stealing your identity," says Abagnale who escaped from police custody twice, once from a taxiing airliner and once from a US federal penitentiary, before he was 21 years old. "Never state your date of birth and where you were born [on personal profiles], otherwise you are saying 'come and steal my identity'." Abagnale, who now works as a security consultant, was the target of a US federal manhunt in the 1960s as he posed as an airline pilot, doctor and attorney to steal millions of dollars. “What I did 40 years ago as a teenage boy is 4,000 times easier now,” says Abagnale who urged Facebook members to educate themselves and their children about the risks of giving away personal information online. “I have three sons on [Facebook]. I totally understand why people like it. But like every technology you have to teach children, it is an obligation of society to teach them how to use it carefully.”"

Comment: Re:Would Someone Explain This? (Score 5, Interesting) 626

by SocratesJedi (#42918871) Attached to: French Police Unsure Which Twin To Charge In Sexual Assaults
It sounds a little implausible, but perhaps I am unaware of the forensic issues. Due to massive improvements in DNA sequencing, it costs less than $10,000 to acquire a full genome (see https://www.genome.gov/sequencingcosts/ ). So, back-of-the-envelope:

(a) $20k to acquire both genomes, plus
(b) some computational effort to identify interesting DNA polymorphisms ($0 - $1000 ???), plus
(c) PCR'ing out and sequencing of a region of the crime-scene DNA (cheap; less than $100).

So $22k, not counting labor costs?

IAAMB (I am a molecular biologist), but not a forensic one. Maybe it just doesn't work that way. Anyone have other information?

Comment: Re:Good first step (Score 1) 197

by SocratesJedi (#41712627) Attached to: Randomly Generated Math Article Accepted By 'Open-Access' Journal
Just thinking off the top of my head here: Perhaps you could set it up to enumerate theorems, construct a 'dependence' graph to see (i.e. "Theorem 2354 relies on Theorems 54, 272,1102 and 2208") and publish an automated paper on any "bottleneck" theorem that is required to demonstrate many future theorems? Or, failing that, if you can prove things in multiple ways, looking for theorems that reduce the number of dependencies on earlier axioms/theorems.

Comment: Re:Make it illegal (Score 1) 1199

by SocratesJedi (#41574677) Attached to: Hiring Smokers Banned In South Florida City
Thanks for your good reply to GP. However, you should always try to bother with citations, in case others are reading the thread. Here are some citations:

Comment: Re:what "take advantage"? (Score 1) 135

by SocratesJedi (#41407473) Attached to: Another EUSecWest NFC Trick: Ride the Subway For Free
Central authentication is probably overkill. Most travelers probably embark at the same handful of stations every time. I imagine that if fare cards were set up to store metadata plus an HMAC (to prevent tampering with the payload) and stations configured to alarm if the same metadata was ever seen twice locally, at that station, that it would eliminate most fraud. Under that scheme any given card-payload could only be used once per station. (And, n.b. that the inclusion of a HMAC prevents arbitrary card payload choice, since only the station authority can issue a valid HMAC for a given metadata payload).

I guess attackers could still swap known-valid metadata+payload information online to use at multiple stations, but at that point the cost of simply allowing the tiny fraction of abusers to win is probably less than the cost of building out a bunch of infrastructure.
Security

+ - Framesniffing attack against SharePoint and Linkedin-> 1

Submitted by stonedyak
stonedyak (267348) writes "Context Information Security has highlighted a weakness in Internet Explorer, Chrome and Safari web browsers that enables remote attackers to steal sensitive information held on private Microsoft SharePoint sites, as well as mine data from other public websites such as LinkedIn. In these Framesniffing Attacks, a hidden HTML frame is used to load a target website inside the attacker's malicious webpage to read information about the content and structure of the framed pages. The attack bypasses browser security restrictions that are meant to prevent webpages directly reading the contents of 3rd party sites loaded in frames."
Link to Original Source

+ - Using graph theory to predict NCAA tournament outcomes->

Submitted by
SocratesJedi
SocratesJedi writes "Like many technically-minded people, I don't have a lot of time to keep up with sports. Nevertheless, trying to predict the outcome of the NCAA men’s basketball tournament is a fun activity to share with friends, family and colleagues. This year, I abandoned my usual strategy of quasi-randomly choosing teams and instead modeled the win-loss history of all Division I teams as a weighted network. The network included information from 5242 games played during the 2011-2012 season. From this, teams came be ranked using tools from graph theory and those rankings can be used to predict tournament outcomes. Without any a priori information, this method accurately identified all the #1 seeds in the top 5 best teams. It also predicts that at least one underdog, Belmont (#14 seed), will reach the Elite Eight. Although the ultimate test will be how well it predicts tournament outcomes, initial benchmarks suggest 70-80% accuracy would not be unreasonable."
Link to Original Source

Comment: Re:Like the cat (Score 1) 324

by SocratesJedi (#38936189) Attached to: $100,000 Prize: Prove Quantum Computers Impossible
Hrm? Entangled particles can exhibit correlations in some measurements that would (apparently) require them to exchange information faster-than-light. However, these correlations are only observable when measurements from distant locations are subsequently compared and - as far as I am aware - can not be used to actually send a message faster-than-light. No information is recoverable by observing only one data set. It is a complicated issue.

Comment: Re:increased response time (Score 2, Insightful) 113

by SocratesJedi (#38163440) Attached to: Robots To Patrol South Korean Prisons
Are you kidding me? All human lives are valuable, without exception. Any other belief is, frankly, uncivilized and reeks of a primitive us-versus-them mentality. What's worse is that prisoners are explicitly under the protection of the state. If an unarmed prisoner is injured in an act of violence, it ought to be interpreted as a total fuck-up and a warden had better lose his job.

Comment: Re:Why is this tagged "medicine"? (Score 1) 841

by SocratesJedi (#37971028) Attached to: Why Do So Many College Science Majors Drop Out?
Medicine has a strong connection to science. Most of the major initial contribution to the life sciences were made by physician-scientists. Having studied medicine (I am an MD/PhD student), I can tell you that it is essentially impossible to "memorize textbooks and regurgitate on command" without building a mental model of the underlying biology or physiology. While there is a strong need to build a base of knowledge, there is also a continuing need to be able to critically evaluate the scientific literature. I would say that any medical program that doesn't promote critical thinking and scientific literacy is a program in need of reform. My experience with the basic sciences faculty, however, has been that they spend a fair amount of time thinking about how to best train students to be critically evaluate scientific ideas.

Comment: Re:It's not at all addictive (Score 1) 920

by SocratesJedi (#37891780) Attached to: The White House Responds To We the People Petition
This is an empirical question. A quick Google search reveals this study on withdrawal in daily marijuana users: Marijuana abstinence effects in marijuana smokers maintained in their home environment (PDF link). Bottom line is that clinically significant withdrawal symptoms were observed in that population.

Comment: Re:I stopped reading the responses after... (Score 1) 920

by SocratesJedi (#37891102) Attached to: The White House Responds To We the People Petition
I couldn't find find a really good source for addictive potential in the literature (which is not to say there isn't a good source).

However, you may find this article in the Lancet (Pubmed link) to be of interest. The study is "Development of a rational scale to assess the harm of drugs of potential misuse". One criterion used was abuse potential. Long story short: (a) cannabis ranked as a middle-of-the-road substance in terms of harms, and (b) legal classification of drugs in the UK does not correlate well with degree of harms.

What this country needs is a dime that will buy a good five-cent bagel.

Working...