Difference between revisions of "Oculus/Version History"

Revision as of 16:28, 27 August 2017

I realised that with too many inputs and small number of data NN was learning too slow.

I decreased the number of data and it started to learn faster.

I made a system with 3 predictors which use different attributes and got the sum of the result.

Ranking:55

1.2 - 26.08.2017
- Anti-Ram added.
- Some multi-wave added.
- Fixed infinite loop in MEA.
- Fixed the "battlefield width/height = 0" bug.

Ranking:63

1.1 - 26.08.2017
- Added Forward Wall MEA to Movement.
- Added Time Since Deceleration to Gun.

Ranking:80

__________________________________________________________________________________________

1.0 - 26.08.2017
- Initial release.
- Movement uses Lateral Velocity, Lateral Acceleration, Bullet Float Time
- Gun uses Lateral Velocity, Lateral Acceleration, Bullet Float Time, Forward Wall MEA
- Ranking: 98

@@ Line 1: / Line 1: @@
 * '''1.2''' - 26.08.2017
 **Made the movement learn faster with same accuracy:
-** ** I realised that with too many inputs and small number of data NN was learning too slow.**I splitted the predictors into 3 parts.
+: I realised that with too many inputs and small number of data NN was learning too slow.
+: I decreased the number of data and it started to learn faster.
+: I made a system with 3 predictors which use different attributes and got the sum of the result.
+**Added precise bot width danger to surfing.
 Ranking:55
+* '''1.2''' - 26.08.2017
+** Anti-Ram added.
+** Some multi-wave added.
+** Fixed infinite loop in MEA.
+** Fixed the "battlefield width/height = 0" bug.
+Ranking:63
+**Wow I found an error both in gun and movement.
 * '''1.1''' - 26.08.2017
-** Forward Wall [[MEA]] Infinite Loop fix
-** Forward Wall [[MEA]] normalisation fix
 ** Added Forward Wall MEA to Movement.
-** Added Time Since Deceleration to Gun
+** Added Time Since Deceleration to Gun.
 Ranking:80