LL MLKAD - Killaura detector based on machine learning 0.2.4

What's this
As the title says, this plugin uses machine learning to detect players who use Killaura.

Usage
Download, copy the plugin (.dll file) to the server plugins folder. The debug file (.pdb) can be moved to the plugins folder optionally.

Features
Compared to traditional Killaura detection plugins, this plugin does not rely on cps methods for detection, so players with fast click speed will not be misjudged.
Machine learning can help us discover many potential features of Killaura, so the accuracy of this plugin's judgment will be very high.
This plugin insists on being open source and free, which means we can get help from the community, and collect more data to improve the plugin. Over time, the detection rate of this plugin will get higher and higher.

Performance
On a test platform with 2.20GHz, using Release compilation (O2 optimization), each detection takes about 39400 nanoseconds (0.0000394 seconds).


How it works
Killaura attacks creatures around the player, not the one in front of the player. Therefore, it is possible to detect whether a player is using Killaura by checking whether the player’s line of sight is aimed at the target when attacking creatures.
This plugin collects the angle between the player's line of sight and the monster, and the distance between the player and the monster, to collect the player's attack characteristics.
Then, use DNN neural network for learning and classification.

In the figure, distance and angle are x and y values respectively, blue dots are normal attack characteristics, red dots are Killaura (Horion) characteristics.
For more information, please visit the LiteLoader blog.

Note
Currently, this plugin is in the development stage, and the model is not reliable. Please do not rely too much on the detection results of this plugin.
If you find any false positives, please go to the Github Issues to give feedback. Your feedback is our motivation!