stefan
/
NLMSvariants
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/FBRDNLMS/NLMSvariants

This commit is contained in:
kbecke23 2018-06-11 15:52:49 +02:00
parent 8da6a017b2 486bc83403
commit 59cf6e19a5
4 changed files with 29 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
bin/README.md
View File

@ -60,7 +60,8 @@ Logs are created by default:
+ each nlms function.Predicted value, input value and error value as well as square mean and deviation + each nlms function.Predicted value, input value and error value as well as square mean and deviation
+ updated weights for each nlms calculation + updated weights for each nlms calculation
`CPP_NLMS -i <inputfile> -g true` prints a graphical overview to `graphResults.html`. The output is an __SVG__ embedded in html. `CPP_NLMS -i <inputfile> -g true` prints a graphical overview to `graphResults.html` using the `graphResults_template.html` in the same directory as the binary.
`CPP_NLMS -i <inputfile> -g <pathToTemplate>` does the same, only the `graphResults_template.html` is handed from a diffrent path. The output is an __SVG__ embedded in html for both.
![alt text](https://github.com/FBRDNLMS/NLMSvariants/blob/master/img/lms_graphing.png "ANSI_C/C++ bin, generate graphical output ") ![alt text](https://github.com/FBRDNLMS/NLMSvariants/blob/master/img/lms_graphing.png "ANSI_C/C++ bin, generate graphical output ")

41
src/ansi_c_implementation/NLMSvariants.c
View File

@ -194,7 +194,7 @@ Variant (1/3), substract local mean.
*/ */
void localMean ( mldata_t *mlData, point_t points[] ) { void localMean ( mldata_t *mlData, point_t points[] ) {
double *localWeights = (double *) malloc ( sizeof(double) * mlData->windowSize + 1); double *localWeights = (double *) malloc ( sizeof(double) * mlData->windowSize + 1);
localWeights = mlData->weights; memcpy(localWeights, mlData->weights, sizeof(double) * sizeof(mlData->windowSize) );
char fileName[50]; char fileName[50];
const unsigned xErrorLength = mlData->samplesCount; const unsigned xErrorLength = mlData->samplesCount;
@ -227,7 +227,7 @@ void localMean ( mldata_t *mlData, point_t points[] ) {
xError[xCount] = xActual - xPredicted; // Get error value xError[xCount] = xActual - xPredicted; // Get error value
xSquared = 0.0; xSquared = 0.0;
for (i = 1; i < _arrayLength; i++) { // Get xSquared for (i = 1; i < _arrayLength; i++) { // Get xSquared
xSquared += pow(xSamples[xCount - i] - xMean, 2); xSquared += (xSamples[xCount - i] - xMean) * (xSamples[xCount - i] - xMean);
} }
if ( xSquared == 0.0 ) { // Otherwise returns Pred: -1.#IND00 in some occassions if ( xSquared == 0.0 ) { // Otherwise returns Pred: -1.#IND00 in some occassions
xSquared = 1.0; xSquared = 1.0;
@ -235,9 +235,9 @@ void localMean ( mldata_t *mlData, point_t points[] ) {
for ( i = 1; i < _arrayLength; i++ ) { // Update weights for ( i = 1; i < _arrayLength; i++ ) { // Update weights
localWeights[i] = localWeights[i - 1] + mlData->learnrate * xError[xCount] // Substract localMean localWeights[i] = localWeights[i - 1] + mlData->learnrate * xError[xCount] // Substract localMean
* ( (xSamples[xCount - i] - xMean) / xSquared ); * ( (xSamples[xCount - i] - xMean) / xSquared );
fprintf( fp9, "%lf\n", localWeights[i] ); fprintf( fp9, "%lf;", localWeights[i] );
} }
fprintf(fp9, "\n");
fprintf(fp4, "%d\t%f\t%f\t%f\n", xCount, xPredicted, xActual, xError[xCount]); // Write to logfile fprintf(fp4, "%d\t%f\t%f\t%f\n", xCount, xPredicted, xActual, xError[xCount]); // Write to logfile
points[xCount].xVal[1] = xCount; // Save points so graph can be build later on points[xCount].xVal[1] = xCount; // Save points so graph can be build later on
@ -253,12 +253,14 @@ void localMean ( mldata_t *mlData, point_t points[] ) {
double deviation = 0.0; double deviation = 0.0;
for (i = 1; i < xErrorLength; i++) { // Mean square for (i = 1; i < xErrorLength; i++) { // Mean square
deviation += pow(xError[i] - mean, 2); deviation += (xError[i] - mean) * (xError[i] - mean);
} }
deviation /= xErrorLength; // Deviation deviation /= xErrorLength; // Deviation
printf("mean square err: %lf, variance: %lf\t\tlocal Mean\n", mean, deviation); printf("mean square err: %lf, variance: %lf\t\tlocal Mean\n", mean, deviation);
fprintf(fp4, "\nQuadratische Varianz(x_error): %f\nMittelwert:(x_error): %f\n\n", deviation, mean); // Write to logfile fprintf(fp4, "\nQuadratische Varianz(x_error): %f\nMittelwert:(x_error): %f\n\n", deviation, mean); // Write to logfile
fclose(fp4); fclose(fp4);
free(localWeights);
} }
/* /*
@ -272,12 +274,12 @@ substract direct predecessor
====================================================================================================== ======================================================================================================
*/ */
void directPredecessor( mldata_t *mlData, point_t points[]) { void directPredecessor( mldata_t *mlData, point_t points[]) {
double *localWeights = ( double * ) malloc ( sizeof(double) * mlData->windowSize + 1 ); double *localWeights = (double * ) malloc ( sizeof(double) * mlData->windowSize + 1 );
localWeights = mlData->weights; memcpy(localWeights, mlData->weights, sizeof(double) * sizeof(mlData->windowSize));
char fileName[512]; char fileName[512];
const unsigned xErrorLength = mlData->samplesCount; const unsigned xErrorLength = mlData->samplesCount;
double xError[xErrorLength]; double xError[xErrorLength];
unsigned xCount = 0, i; unsigned xCount = 0, i;
double xActual = 0.0; double xActual = 0.0;
double xPredicted = 0.0; double xPredicted = 0.0;
@ -303,7 +305,8 @@ void directPredecessor( mldata_t *mlData, point_t points[]) {
double xSquared = 0.0; double xSquared = 0.0;
for (i = 1; i < _arrayLength; i++) { for (i = 1; i < _arrayLength; i++) {
xSquared += pow(xSamples[xCount - 1] - xSamples[xCount - i - 1], 2); // substract direct predecessor xSquared += (xSamples[xCount - 1] - xSamples[xCount - i - 1])
* (xSamples[xCount - 1] - xSamples[xCount - i - 1]); // substract direct predecessor
} }
if ( xSquared == 0.0 ) { // Otherwise returns Pred: -1.#IND00 in some occassions if ( xSquared == 0.0 ) { // Otherwise returns Pred: -1.#IND00 in some occassions
xSquared = 1.0; xSquared = 1.0;
@ -327,12 +330,14 @@ void directPredecessor( mldata_t *mlData, point_t points[]) {
for (i = 1; i < xErrorLength; i++) { for (i = 1; i < xErrorLength; i++) {
deviation += pow(xError[i] - mean, 2); // Mean square deviation += (xError[i] - mean) * (xError[i] - mean); // Mean square
} }
deviation /= xErrorLength; // Deviation deviation /= xErrorLength; // Deviation
printf("mean square err: %lf, variance: %lf\t\t\tdirect Predecessor\n", mean, deviation); printf("mean square err: %lf, variance: %lf\t\t\tdirect Predecessor\n", mean, deviation);
fprintf(fp3, "\nQuadratische Varianz(x_error): %f\nMittelwert:(x_error): %f\n\n", deviation, mean); fprintf(fp3, "\nQuadratische Varianz(x_error): %f\nMittelwert:(x_error): %f\n\n", deviation, mean);
fclose(fp3); fclose(fp3);
free(localWeights);
} }
/* /*
@ -347,10 +352,11 @@ differential predecessor.
*/ */
void differentialPredecessor ( mldata_t *mlData, point_t points[] ) { void differentialPredecessor ( mldata_t *mlData, point_t points[] ) {
double *localWeights = (double *) malloc ( sizeof(double) * mlData->windowSize + 1 ); double *localWeights = (double *) malloc ( sizeof(double) * mlData->windowSize + 1 );
localWeights = mlData->weights; memcpy(localWeights, mlData->weights, sizeof(double) * sizeof(mlData->windowSize));
const unsigned xErrorLength = mlData->samplesCount;
const unsigned xErrorLength = mlData->samplesCount;
char fileName[512]; char fileName[512];
double xError[xErrorLength]; double xError[xErrorLength];
unsigned xCount = 0, i; unsigned xCount = 0, i;
double xPredicted = 0.0; double xPredicted = 0.0;
@ -377,7 +383,8 @@ void differentialPredecessor ( mldata_t *mlData, point_t points[] ) {
double xSquared = 0.0; double xSquared = 0.0;
for (i = 1; i < _arrayLength; i++) { for (i = 1; i < _arrayLength; i++) {
xSquared += pow(xSamples[xCount - i] - xSamples[xCount - i - 1], 2); // Substract direct predecessor xSquared += (xSamples[xCount - i] - xSamples[xCount - i - 1])
* (xSamples[xCount - i] - xSamples[xCount - i -1]); // Substract direct predecessor
} }
if ( xSquared == 0.0 ) { // Otherwise returns Pred: -1.#IND00 in some occassions if ( xSquared == 0.0 ) { // Otherwise returns Pred: -1.#IND00 in some occassions
xSquared = 1.0; xSquared = 1.0;
@ -404,12 +411,14 @@ void differentialPredecessor ( mldata_t *mlData, point_t points[] ) {
for (i = 1; i < xErrorLength; i++) { // Mean square for (i = 1; i < xErrorLength; i++) { // Mean square
deviation += pow(xError[i] - mean, 2); deviation += (xError[i] - mean) * (xError[i] - mean);;
} }
deviation /= xErrorLength; deviation /= xErrorLength;
printf("mean square err: %lf, variance: %lf\t\t\tdifferential Predecessor\n", mean, deviation); printf("mean square err: %lf, variance: %lf\t\t\tdifferential Predecessor\n", mean, deviation);
fprintf(fp6, "\nQuadratische Varianz(x_error): %f\nMittelwert:(x_error): %f\n\n", deviation, mean); fprintf(fp6, "\nQuadratische Varianz(x_error): %f\nMittelwert:(x_error): %f\n\n", deviation, mean);
fclose(fp6); fclose(fp6);
free(localWeights);
} }
/* /*

2
src/ansi_c_implementation/README.md
View File

@ -29,7 +29,7 @@ There are a bunch of options you can predefine but do not have to. The only para
| -w | Size of M (window). | 5 | | -w | Size of M (window). | 5 |
| -c | Choose RGB color channel, green has least noise. | green | | -c | Choose RGB color channel, green has least noise. | green |
| -l | Learnrate of machine learning.| 0.4 | | -l | Learnrate of machine learning.| 0.4 |
| -g | include graph building. Choose for amount of input data lower than 1200. If the template is located in another folder use its path otherwise use true.| none| | -g | include graph building. Choose for amount of input data lower than 1200. If the template is located in another folder use its path otherwise use true. Do not use whitespace in path to folder.| none|
| -s | Seed randomizing weights. Choose for repoducability. | time(NULL)| | -s | Seed randomizing weights. Choose for repoducability. | time(NULL)|
This code is ANSI aka C89 compatible. No POSIX, C99, C11 or GNU libs, because it had to be windows compatible . There are way easier methods like getline() for file parsing or getopt() as an args parser, because of compatibility reasons things have been kept simple. This code is ANSI aka C89 compatible. No POSIX, C99, C11 or GNU libs, because it had to be windows compatible . There are way easier methods like getline() for file parsing or getopt() as an args parser, because of compatibility reasons things have been kept simple.

2
src/cpp_implementation/README.md
View File

@ -28,6 +28,6 @@ There are a bunch of options you can predefine but do not have to. The only para
| -w | Size of M (window). | 5 | | -w | Size of M (window). | 5 |
| -c | Choose RGB color channel, green has least noise. | green | | -c | Choose RGB color channel, green has least noise. | green |
| -l | Learnrate of machine learning. | 0.4 | | -l | Learnrate of machine learning. | 0.4 |
| -g | include graph building. Choose for amount of input data lower than 1200. Choose path if template is located in another folder, else use true.| none| | -g | include graph building. Choose for amount of input data lower than 1200. Choose path if template is located in another folder, else use true. Do not use whitespace in path.| none|
| -s | Seed randomizing weights. Choose for repoducability. | time(NULL)| | -s | Seed randomizing weights. Choose for repoducability. | time(NULL)|