Made some slight update because I didn’t think about the keys not being in order. Might give the wrong predicted value orders in predictedValueArray
3 Likes
Fixed some bugs that I didn’t notice. Update the distributed computing version.
5 Likes
If you guys wanted to access fullDataVector upon receiving predicted values for Distributed Computing, I have added in that. Just replace the Distributed computing version.
Not sure why I had left that purposely.
3 Likes
New update for Distributed Computing Version:
- Whenever a client joins or leaves, all clients will watch a new set of random clients. Makes it more difficult for the hackers to figure out which client controls the hacker’s data. Also, it allows the clients have balanced number of watchers.
All you need is to replace that particular version.
3 Likes
Added timer to check when the last time server received an update from client. This is applicable for distributed computing only.
2 Likes
Fixed some bugs with distributed version when it comes to anomaly detector.
2 Likes
Also guys , do me a favour, if you guys find any bugs, please report any bugs using the link I have given you under “Links” section. It seems like it is difficult for me to find all issues on my own with the distributed computing version.
1 Like
please do not post ChatGPT on the devforum
1 Like
How could you tell? Seem normal stuff to me.
I’m honored that someone thinks my responses to this resource is AI-like, but it only takes a high school diploma to learn how to read and write proper English feedback.
1 Like
So, this is a good server resource. But, how can you say it will help with client-sided-only stuff like “Aimbots” which are either camera-based for FPS or mouse-based for TPS shooters? How would you analyze the client-sided behaviors of the two and store/record the data that way? The way I see it, this does seem really good for server-sided physics-based stuff currently?
For aim bot, just move the updateDataVector() function (from DataCollector script in original or custom version) to a client and make it fire remote event so that the server receives inputs from the client.
Anyways, it’s just a suggestion and you have the freedom to change the source code so you can make it work as you like. I only provide the template.
1 Like
Great! I got it working just now on your original source for the velocity check on the server. To me, doing checks like this on the client should yield more accuracy since you won’t have to account for latency. Also, that’s what I was thinking about doing with this - 1st person and 3rd person aimbot analyzing.
I’m still just getting used to this deep-learning library. My idea is to use some basic aimbot scripts to test this and have it record anomalies when i lock onto a target and have it record regular data otherwise.
Was going to maybe use mouse.Move() for TPS and CurrentCamera:Getpropertycontextchanged(“CFrame”) for FPS camera aimbots. Not sure how it’ll go. If I do go through with this I can say how it goes, but maybe someone with a mind meant for long-analysis would pick up the tab.
1 Like
You’re asking me to do the aim bot version for the anti-cheat?
I guess I could do that since machine learning engineers are rare around here.
Well awesome! So yeah like… even in terms of performance, having it based on those connections i mentioned would be the best. Since you wanna collect data when necessary if you dont need to do it per frame.
1 Like
mark rober has a video on it, i encourage you to watch it:
2 Likes
Does this module work with profile service?
If you mean the profile service is integrated inside the library, then no. You have to do that yourself.