NLMSvariants/doc/Messprotokoll.md

1. CPP-NLMS-Testreihe

1.1 Veränderung der Lernrate

Paramter:

			- Input: 500.000 Pixel (-n 500000)
			- WindowSize: 5 (Standart Value)
			- Lernrate von 0.1 bis 1.0 (increased by 0.1)
			- -i cathedral.ppm 
			- random seed generator : 3 (-s 3)

1. Local Mean

Lernrate	meanerror	variance
0.1	0.020861	373.760784
0.2	-0.015231	1261.386282
0.3	-0.054507	2574.244199
0.4	-0.088173	4229.104809
0.5	-0.116079	6169.762185
0.6	-0.139313	8360.777760
0.7	-0.158889	10783.351638
0.8	-0.175600	13432.957565
0.9	-0.190101	16318.529330
1.0	-0.202958	19463.062580

2. direct Predecessor

Lernrate	meanerror	devitation
0.1	0.024560	608.448633
0.2	-0.030306	2007.513911
0.3	-0.084634	3959.201147
0.4	-0.126798	6328.009325
0.5	-0.157719	9057.241991
0.6	-0.179736	12137.852301
0.7	-0.194805	15592.808948
0.8	-0.204488	19471.279015
0.9	-0.210085	23849.267714
1.0	-0.212681	28835.466521

3. differential Predecessor

Lernrate	meanerror	devitation
0.1	0.046246	237.604268
0.2	0.058362	838.182243
0.3	0.060553	1834.829054
0.4	0.060390	3227.671286
0.5	0.060858	5016.829374
0.6	0.063120	7202.687465
0.7	0.067694	9785.843689
0.8	0.074840	12767.049840
0.9	0.084710	16147.248088
1.0	0.097443	19927.694764

4. NLMS

Lernrate	meanerror	devitation
0.1	-0.865235	51.897994
0.2	-0.916315	67.761425
0.3	-0.950006	82.291526
0.4	-0.975096	96.554001
0.5	-0.995089	111.236008
0.6	-1.011618	126.820482
0.7	-1.025538	143.713619
0.8	-1.037385	162.334219
0.9	-1.047550	183.192878
1.0	-1.056358	206.993456

1.2 Veränderung der WindowSize

Paramter:

			- Input: 500.000 Pixel (-n 500000)
			- WindowSize: -w {1,3,5, 8, 10, 15, 20, 25, 50, 100}
			- Lernrate: -l 0.6
			- -i cathedral.ppm 
			- random seed generator : 3 (-s 3)

Tabelle beinhaltet jeweils den 'mean error' der jeweiligen Variante

Windowsize	local Mean	direct Predecessor	differential Predecessor	NLMS
1	0.330125	0.330131	0.330131	-1.063652
3	0.444605	0.412069	0.413190	-1.089547
5	-0.139313	-0.179736	0.063120	-1.011618
8	0.181180	0.042986	0.221078	-0.873373
10	0.279242	0.188497	0.291145	-0.811906
15	0.215564	0.222018	0.265400	-0.696719
20	0.273161	0.383155	0.194116	-0.615447
25	0.213731	0.276315	0.185892	-0.576340
50	0.467920	0.463972	0.188099	-0.479407
100	-0.331102	-0.412149	-0.018204	-0.506369

---

2. CS-NLMS-Testreihe

Parameter:

			-N:4000 Pixel
			-M: 5
			-Learnrate: 0.1 bis 1.0 (increased by 0.1)
			-Image: boats.y.png
			--> Gewichte sind random, daher nicht reproduzierbar!

1. local mean

Lernrate	Average error	Variance of the error
0.1	-0,0223811013862385	5337,82462718957
0.2	-0,0622259728578006	7363,56898611733
0.3	-0,0298976210637875	9614,92806843726
0.4	0,0849844894447272	12088,2433018213
0.5	0,299055989866592	14819,6805234785
0.6	0,585283074155977	17874,0914703503
0.7	0,979106442634296	21337,9978781998
0.8	1,46391408202542	25322,8680485397
0.9	2,05872807071837	29964,9719994441
1,0	2,79656379441662	35436,9340804565

2. direct predecessor

Lernrate	Average error	Variance of the error
0.1	0,151005103097187	5471,5876773699
0.2	0,606943328690361	7729,18121268243
0.3	1,14390553779214	10465,2060324616
0.4	1,76810718228813	13695,8876087231
0.5	2,5193083296817	17431,0433982945
0.6	3,38144695588288	21696,5188824818
0.7	4,36367842697108	26566,2513099756
0.8	5,51257464442153	32193,8755290981
0.9	6,83374986915135	38850,115776936
1.0	8,38129514988831	46958,0983013679

3. differential predecessor

Lernrate	Average error	Variance of the error
0.1	-0,0160622191672049	4750,33040669542
0.2	0,00640864710641419	6041,46065665648
0.3	0,0432263456982662	7412,09833871273
0.4	0,106075295986603	8891,97578539296
0.5	0,174689052473554	10532,273473301
0.6	0,241331954473674	12394,8177286223
0.7	0,359973579958283	14552,6708886431
0.8	0,494261683495002	17102,3682973733
0.9	0,671346960651145	20166,0023256068
1.0	0,892045554010017	23907,2852270274