Sunday, 20 September 2009

Here it comes.....


A five-year research programme, called Project Indect, aims to develop computer programmes which act as "agents" to monitor and process information from web sites, discussion forums, file servers, peer-to-peer networks and even individual computers.

Its main objectives include the "automatic detection of threats and abnormal behaviour or violence".

Project Indect, which received nearly £10 million in funding from the European Union, involves the Police Service of Northern Ireland (PSNI) and computer scientists at York University, in addition to colleagues in nine other European countries


Agent Smith is profiling us

Project Indect Website

It is seeking to develop models of "suspicious behaviour" so these can be automatically detected using CCTV and other surveillance methods. The system would analyse the pitch of people's voices, the way their bodies move and track individuals within crowds.

20 comments:

Anonymous said...

BACKTYPE

Wyrdtimes said...

How will the system work when the whole crowd is acting "suspiciously"?

bomb - No10 - al qaeda - taliban - mosque - fuhrer - UAF - assassinate - Blair - Mandelson - Brown - MI5 - MI6 - freedom - nailbomb - aldershot - co-ordinate - nerve - etc

Anonymous said...

IMHO (CS graduate 20 years exp) this will not work.

You can detect simple things - like checking for "make a bomb" (which this post would throw up....) but computers are still lousy at pattern recognition.

This includes things like understanding spoken and written text.

CCTV is worse. It is *very* difficult for a computer to recognise a plain object on a plain background reliably.

If you look at a book, to you it's obviously a book. To a computer, it *looks* entirely different depending on how you are looking at it (end on versus front on for example, or at an angle), small pocket books, childrens pop ups when opened and encyclopedias look entirely different anyway etc etc etc.

No one has yet reliably solved these problems.

And yet we're going to waste money on a system which can supposedly do this on live CCTV pictures with all the background which the brain happily can filter out but computers can't.

Pitch of people's voices ? Well, my missus has a lot higher pitched voice than me, being female. To a person, she's obviously female, to a computer, well, it's difficult to tell for the same reasons.

Its Bollox !

You will get (if anything) a system which either throws out so many false positives it's worthless (and humans then spend hours figuring out what's important) or misses so much it is pointless.

Waste of time and money.

IanPJ said...

The fact that they are spending your tax money on undertaking such a project tells you much more than whether the technology is up to the job.

I wrote about this several years ago, but then I was called a conspiracy nut.... not so nutty now is it?

http://thejournal.parker-joseph.co.uk/blog/_archives/2007/3/5/2780523.html

Be warned, be aware. They do know what they are doing.

Marchamont Needham said...

hate to state the bleeding obvious - but a bot can only monitor non-password protected sites.

Mitch said...

So when we all start wearing rucksacks,buying batteries and fertiliser while posting on blogs like this how long till their poxy system fails? about 5 mins I`d guess.

Plutonium,trigger,fusion burn,ground zero,canary wharf,wibble wibble jihad and so on........tossers.

caesars wife said...

They have been doing it for some time alas our ISPs have been complicit , govt has been using Meltwater posted on here by someone when jaqui smithh was trying to take certain bloggers out in expenses .

This is a visble one , but they been using them for years

FP7 Research Associate said...

I wouldn't worry. It's an EU-funded FP7 research project involving nearly 20 different institutions.

They will spend most of the time and money arguing about who is responsible for what, each trying to do as little as possible. The project will produce reams of paperwork and a "demonstrator system" that technically does the things listed in the grant proposal, but is of no practical use.

This is bad only because it is a huge waste of your money: this will probably cost a hundred millions Euros.

VotR said...

More fun ways to beat the system.

Agents don't have to work for the powers that be. Governments and ISPs can be watched, monitored and assessed right back.

Rogerborg said...

Calm down, dears. It's just another EU cash cow. A few academics will get funded to bodge up useless proofs of concept, and the bulk of the funds will get drunk and embezzled, same as usual.

The great thing about all these Stazi-porn schemes is that by the time computers are smart enough to implement them, they'll be smart enough to rule us. They could hardly do a worse job than we're doing ourselves.

bofl said...

surely it would be easier just watching the parliament channel?

Anonymous said...

Yet another good reason to persuade friends/family in Ireland to vote 'NO'. I've sent the links to friends there.

John Steed said...

Got a lot of Tracker Cookies on mine. One theory, harmless. Another "they" are watching.

Fausty said...

Great site, IanJP.

Chilling.

This has been in the making for decades and what we think we know about it is probably a tip of the iceberg.

fascile said...

Anon 10:35,

Unless the result is utterly hopeless to the point where it never gets further than security industry trade shows and scaremongering BBC reports, there is still some chance it would be implemented. The stated intentions will not come to pass, rather if rolled out it will work as sniffer dogs do - imply suspicion to a high enough but fake threshold that more invasive and sinister means of investigation can be applied.

Who knows what sniffer dogs react to? They can go mental at almost anything and are far from infallible but it is a convenient means to create what Americans would call probable cause.

Computer says Shifty will become a legitimate reason for the Police to begin harrassing people.

1327 said...

>Calm down, dears. It's just another
>EU cash cow. A few academics will
>get funded to bodge up useless
>proofs of concept, and the bulk of
>the funds will get drunk and
>embezzled, same as usual.

Spot on I have seen a fair few of these "projects" in my time. The vast majority of the money will be spent on conferences in the USA , new office furniture and Sony laptops all round.

The only actual research done will be by a Chinese researcher specially imported and paid about £3K a year. This poor sod will be given 10 year old 4th hand PC running an elderly version of MATLAB. Hopefully he will be able to knock up some kind of a demo at the end of 5 years but if he can't it doesn't matter as no one will ever check up.

Not that I'm cynical.

Anonymous said...

Re 'they' - I always cleardown my history & get rid of cookies at least once a day.

John Steed said...

Clearing the history is easy enough. Cookies? I thought they were innocuous and one could eat them. These bastards want to know everything. Hey Constable, I had a pee 5 minutes ago.

Cold said...

This is a really, really interesting idea. I'm a neural network programmer and being able to do this would be a massive step forward (=Nobel prize winning) if it could be done without a SIGNIFICANT (eg makes it useless) number of false-positives.

In the stuff we do after we've trained the algorithm we consider a 20:1 FPR to be acceptable

Anonymous said...

I'm guessing John Steed alias K McEgan has a lot more than just cookies on his PC. More like a shed load of kiddie porn again. Wotcha Ron Glitter Broxted.

Ratings and Recommendations by outbrain

LinkWithin

Related Posts with Thumbnails