[MaloModo]

Originally Posted by RedSword If anyone is wondering, it is for the games CSS and HL2DM. Note that I tried to connect to your first server... Connecting to 95.142.67.85:27095... STEAM validation rejected And you say you need 100 players; but shouldn't you get some pros in that 100 (or else your neural would give false positive) ? Good luck with your thesis, Red post scriptum : my ping is too high; so I guess you only target Europeans

Yes get some "online" pros in there...don't want any "false" false positives. I am sorry but what a stupid comment. Is anyone naive enough nowadays to actually believe this drivel? And IMVHO wouldn't you be getting pretty "biased" data from "pros" around here? Sampling bias and all that

I am not sure of your exact end game but this reminds me of SB Oculus that was abandoned years ago I believe. http://forums.eventscripts.com/viewtopic.php?t=19996

Anyhow good luck. Certainly no expert but always had a certain curiosity with neural networks. I believe they have been used with some success for stock trading.

[SWAT_88]

Originally Posted by Scrumilation Might want to post it to reddit, if you are lucky your servers will be filled and you will have data for the rest of your life :p

Thanks for the hint
However, I've already collected the data, and I don't have enough time for a second test run. But thanks anyway, I appreciate it.

Originally Posted by MaloModo Yes get some "online" pros in there...don't want any "false" false positives. I am sorry but what a stupid comment. Is anyone naive enough nowadays to actually believe this drivel? And IMVHO wouldn't you be getting pretty "biased" data from "pros" around here? Sampling bias and all that I am not sure of your exact end game but this reminds me of SB Oculus that was abandoned years ago I believe. http://forums.eventscripts.com/viewtopic.php?t=19996 Anyhow good luck. Certainly no expert but always had a certain curiosity with neural networks. I believe they have been used with some success for stock trading.

I've collected data from more than 100 players, I'm sure I've also got data from some "pros"

I've already successfully trained some neural networks for pattern recognition, so I think this should work also with movement and aiming data.
If it doesn't work, I can also use other techniques, like "symbolic regression" or "random forest", etc.

PS:
Can you please remove the IP address from your post?
I've already finished data recording, and we should remove all server advertisements.

Best regards

[SWAT_88]

Originally Posted by RedSword I've no idea if you'll be showing statistics in your thesis, but if so, even if it's in German, those can be, sometimes, and/or partially, understood. For instances graphs or tables; you just need to translate the header. I would probably not read it (or well I'd try and get tired after one paragraph, as I don't really speak German), but I'd be interested in giving a look at it (skimming through), at least for those graph/tables; and possibly google translate if needs be.

I've uploaded my thesis and all the diagrams, heat maps, as well as the source code to a shared folder using Google Drive.
The thesis is written in English, so you don't have to use Google Translate

The link to the thesis is https://goo.gl/OzwM2e
The link to the source code is https://goo.gl/r2MKse

I've also posted some general information about my thesis in the first post of this thread here.

Have a nice day

[asherkin]

Sweet! Lots of awesome work in there and an impressive conclusion.

[SWAT_88]

Originally Posted by asherkin Sweet! Lots of awesome work in there and an impressive conclusion.

Thanks, I'm glad that my work is useful for somebody

By the way, I've added you and the other moderators to the "Acknowledgments" page.
I highly appreciate your support. Without you and the SourceMod forums, it would not have been possible to get sufficient data from more than 100 players in two weeks.

[Potato Uno]

I read like the first 10 pages until it became too technical for my brain to understand.

Looks like great work and a massive amount of time and effort went into it.

[asherkin]

Originally Posted by Potato Uno I read like the first 10 pages until it became too technical for my brain to understand. Looks like great work and a massive amount of time and effort went into it.

You should have a peruse through the figures in the appendix if you're interested but not in the technicals.

[RedSword]

Thanks for sharing your thesis. The graphs are interesting to look at.

During the time your server ran, what were the most players you had ? And did you evaluate how expensive your data gathering was ? If someone would like to replicate your environment, could they do it on a popular server with a high number of players ?

I also notice that no pool of only-hacking players were used. Rerunning the experiment with those would be interesting ("hacking only" server, lol).

Thanks again for sharing !

[SWAT_88]

Originally Posted by RedSword Thanks for sharing your thesis. The graphs are interesting to look at. During the time your server ran, what were the most players you had ? And did you evaluate how expensive your data gathering was ? If someone would like to replicate your environment, could they do it on a popular server with a high number of players ? I also notice that no pool of only-hacking players were used. Rerunning the experiment with those would be interesting ("hacking only" server, lol). Thanks again for sharing !

Sorry for the delay. Busy weekend.
I had about 12 players or so at peak times, but with no significant change in CPU usage.

Although the plugin is recording the stats of every player, it is very light-weight.
So the plugin is as expensive (or inexpensive) as an average plugin for killstats.

[Ade]

while i appreciate the technical interest in hl2dm and briefly raising the player number for it, i have a question for the author of the thesis:
has this yielded any attempts to make a new anticheat system for these 2 games?