You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+33-1Lines changed: 33 additions & 1 deletion
Original file line number
Diff line number
Diff line change
@@ -2,7 +2,36 @@
2
2
3
3
A Generative Algorithm AI implementation on a 2D racing simulator "Java Racer" (https://github.com/boraakoguz/JavaRacer)
4
4
5
+
## Features
6
+
The algorithm features:
7
+
- Elitism
8
+
- Cross-over succession depending on the success of the parents
9
+
- Mutation
10
+
- Advanced mutation techniques to improve variation
11
+
- A roll-back mechanism in order to prevent bad training.
12
+
13
+
Project features:
14
+
- Custom map/track design
15
+
- Save mechanism
16
+
- Live data on the screen
17
+
18
+
Live Data meanings:
19
+
- Best Fit: The most points scored by an agent of the previous generation
20
+
- Gen. No: The number of generations
21
+
- Population Size: Number of agents in the population
22
+
- Instruction Size: The length of the instruction array used by the agents. Gradually expands as training continues
23
+
- Last Increase Points: The maximum points scored by an agent at the previous Instruction array expansion. Serves as a benchmark and controlls the roll-back system, advanced mutation system and stops Instruction increase in case of failure to pass the benchmark.
Saving is done by writing essential data on the save.txt file in the source foulder. Simply deleting or renaming the file will open up a fresh save and create a new save.txt file on the folder.
52
+
Changing the data on the file is not advised and may cause errors.
0 commit comments