Illegal Immigration & the Known Unknowns - Computerphile

Searching for the known unknown. Data mining & illegal immigration. Professor Uwe Aickelin explains how to mine data you don't even have.

This video was filmed and edited by Sean Riley.

[ociemitchell]

I like how the lorries are "drifting" around the curves!

[thomaskn]

5:28 Lorries drifting around the curve.

[fuppetti]

An effective solution would be filling up every lorry with nitrous oxide, that way you could hear the immigrants within laughing.

[nathanperkins]

This is Vehicular Profiling. Imagine how the vehicles feel.

[stumbling]

2:43 Customs official opens the back of a suspicious lorry to find only a tree.
Tree:  I'm an asylum tree.  Please let me in.  They want to cut me down!

[RiverWyvrn]

Very interesting, didn't know modeling had these kinds of applications

[maitland]

I found this video to be quite frightening, actually: a discussion of all the technology used to catch human beings with no thought or mention of the plight of the human beings themselves. Gave me chills. I'm a scientist so not a stranger to the idea of abstracting a problem and modeling, and of course the technology is interesting, but I don't like to see us disregarding the humanity of our subjects or the complexity of our social issues.

[biocode]

They can start checking everybody for limited intervals to improve their training data so that it won't have false negatives

[Ensorcle]

Firstly, Prof Aickelin's accent is madness. Second, this is a SUPER cool way of utilizing data. I would love to do work like this.

[dixie_rekd]

still think this guy has the best accent ever.

[joelproko]

Somewhere there you spoke of the probability of a certain check finding a person that is on a lorry being part of the simulation. How do you determine how likely it is that the person will be found by this check?
I guess testing different orderings of the various tests might give you a clue, but even then it's only a relative efficiency. Or do you test it by sending a thousand lorries that you know have people hidden but the local people don't through the check you're testing? If so, do the people manning the checks know/realize they're being tested?

[DampeSN]

Couldn't you just search lorries at random and compare that data to lorries picked doing the different techniques and get the correct sense of success rates from that? You'd know which techniques are better than random and by how much. And you can also test combinations of techniques. Regardless, the unknown of how many people are hiding can be estimated from random searches. It isn't an unknown then.

[harryritchie]

I know it's probably a little too mathematical but, could you consider making a video about the different techniques in more detail? I found this really interesting and it seems like a very obscure optimisation problem. Maybe run through a simplified idea of supervised and unsupervised techniques?? I'd like to see the different roles these techniques play in interesting examples or real research!

Great video :)

[parachrematistique]

Am I the only one who realizes that under the interesting model, we're talking about human beings?

[Wout]

I'm quite surprised they made a video on this, frankly. Sure, nice CS and all, but at least where I live the subject of Calais border control is considered controversial (or so I'd say, at least). There have already been many deaths surrounding this issue and although I won't say we're (fully) responsible for those, I'd argue it does make the issue more complex than finding an efficient sorting algorithm, something with a formally defined and superficial solution. If you simply want to make a video about these "known unknowns" which happened to be with this context, a short content disclaimer in the introduction would have been in place IMO. Well, my thoughts at least.

[austinb]

7:09 tell that to the Los Angeles airport

[kelseystilson]

Why even make the graphical model? Why not just run the simulation without watching all the cars go though?

[lineikatabs]

wait, did he say "cosmic rays"?! are we actually using COSMIC RAYS as an x-ray? i didn't know that and my mind is blown!

[kamebwoi]

You know a topic is fascinating when everybodies reaction is to chime in with their own ideas for solving the problem :D

[hellterminator]

After watching this video I still have no idea how we estimate the known unknowns.