void * node_Value(void * void_node)
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
DSL_nodeValue * nodeValue = node->Value();
void * retval = nodeValue;
return retval;
}
void * node_Network(void * void_node)
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
DSL_network * net = node->Network();
void * retval = net;
return retval;
}
vf2D getCPTArray(DSL_network &network, string &childName) {
//DSL_network* net = node->Network(); // node network
int childIdx = network.FindNode(childName.c_str());
DSL_node* node = network.GetNode(childIdx);
DSL_nodeDefinition *def = node->Definition();
const DSL_Dmatrix &cpt = *def->GetMatrix();
DSL_intArray coords;
unsigned int colSize = def->GetNumberOfOutcomes();
unsigned int rowSize = cpt.GetSize() / colSize;
vf2D cpt2(rowSize , vf1D(colSize, 0.0));
unsigned int colIdx = 0;
unsigned int rowIdx = 0;
for (int elemIdx = 0; elemIdx < cpt.GetSize(); elemIdx ++) {
cpt2[rowIdx][colIdx] = cpt[elemIdx];
++colIdx;
if(colIdx == colSize) {
colIdx = 0;
++rowIdx;
}
}
return cpt2;
}
void * node_Definition(void * void_node)
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
DSL_nodeDefinition * nodeDef = node->Definition();
void * retval = nodeDef;
return retval;
}
vf1D getWeights(DSL_network &network, string &childName) {
int childIdx = network.FindNode(childName.c_str());
DSL_node* node = network.GetNode(childIdx);
int handle = node->Handle();
DSL_nodeDefinition *def = node->Definition();
const DSL_Dmatrix &cpt = *def->GetMatrix();
const DSL_intArray &parents = network.GetParents(handle);
int parentCount = parents.NumItems();
DSL_intArray coords;
unsigned int colSize = def->GetNumberOfOutcomes();
unsigned int rowSize = cpt.GetSize() / colSize;
vf1D weights(rowSize, 0.0);
unsigned int rowIdx = 0;
for (int elemIdx = 0; elemIdx < cpt.GetSize(); elemIdx += colSize) {
cpt.IndexToCoordinates(elemIdx, coords);
double mult = 1.0;
for (int parentIdx = 0; parentIdx < parentCount; parentIdx ++) {
DSL_node *parentNode = network.GetNode(parents[parentIdx]);
const DSL_Dmatrix &parent_cpt = *parentNode->Definition()->GetMatrix();
mult *= parent_cpt[coords[parentIdx]];
}
weights[rowIdx++] = mult;
}
return weights;
}
int main(int argc, char* argv[]) {
ios_base::sync_with_stdio(0);
string data_infile = string(argv[1]);
string network_infile = string(argv[2]);
string child_name = string("C1");
if (argv[3] == string("EM") || argv[3] == string("Smile")) {
DSL_network net = LearnParamsEM(data_infile, network_infile, child_name);
int childIdx = net.FindNode(child_name.c_str());
DSL_node* childNode = net.GetNode(childIdx);
if (argv[3] == string("Smile")) {
childNode->ChangeType(DSL_NOISY_MAX);
}
//printCPT(childNode);
} else if (argv[3] == string("Naive")) {
DSL_network net = LearnParamsNaive(data_infile, network_infile, child_name);
//int childIdx = net.FindNode(child_name.c_str());
//DSL_node* childNode = net.GetNode(childIdx);
//printCPT(childNode);
} else if (argv[3] == string("Gauss")) {
DSL_network net = LearnParamsGaussJordan(data_infile, network_infile, child_name);
//int childIdx = net.FindNode(child_name.c_str());
//DSL_node* childNode = net.GetNode(childIdx);
//printCPT(childNode);
} else if (argv[3] == string("NvGJ")) {
DSL_network netGJ = LearnParamsGaussJordan(data_infile, network_infile, child_name);
DSL_network netNaive = LearnParamsNaive(data_infile, network_infile, child_name);
DSL_network netEM = LearnParamsEM(data_infile, network_infile, child_name);
DSL_network netOriginal = OpenNetwork(network_infile);
vf1D weights = getWeights(netOriginal, child_name);
vf2D cptGJ = getCPTArray(netGJ, child_name);
vf2D cptNaive = getCPTArray(netNaive, child_name);
vf2D cptEM = getCPTArray(netEM, child_name);
vf2D cptOriginal = getCPTArray(netOriginal, child_name);
cout << "Gauss:" << EuclidianDistance( cptOriginal, cptGJ, weights) << endl;
cout << "Naive:" << EuclidianDistance( cptOriginal, cptNaive, weights) << endl;
cout << "EM:" << EuclidianDistance( cptOriginal, cptEM, weights) << endl;
//cout << "Euclidian distance (orig vs GJ): " << EuclidianDistance(original, netGJ, child_name) << endl;
//cout << "Euclidian distance (orig vs Naive): " << EuclidianDistance(original, netNaive, child_name) << endl;
//int childIdx = net.FindNode(child_name.c_str());
}
return 0;
}
void fillCPTWithOR(vf2D &noisyORParameters) {
probsSize = childDimension * sumParentDimensions + childDimension; //childDimension[rows] * sumParentDimensions[cols] + leak_column
theProbs.SetSize(probsSize);
int probsIdx = 0; //index of cpt vector
int paramsRowIdx = 0; //index of minimal_parameter array
int paramsColIdx = 0; //index of minimal_parameter array
for (int parentIdx = 0; parentIdx < numberOfParents; ++parentIdx) { // for each column
for(int parentStateIdx = 0; parentStateIdx < parentDimensions[parentIdx] ; ++parentStateIdx) { //for each subset of columns
double sumColumn = 0.0;
for(int childStateIdx = 0; childStateIdx < childDimension; ++childStateIdx) { //for each row
//if (parentStateIdx == distinguishedStates[parentIdx]) { //if it's the distinguished state
if (parentStateIdx == parentDimensions[parentIdx] - 1) { //if it's the distinguished state
if (childStateIdx == childDimension - 1) { // if it's both distinguished for child and parent
theProbs[probsIdx++] = 1;
} else {
sumColumn += theProbs[probsIdx++] = 0;
}
} else {
if (childStateIdx == childDimension - 1) { // if it's the child's distinguished state
theProbs[probsIdx++] = 1.0 - sumColumn;
} else { // standard parameter
sumColumn += theProbs[probsIdx++] = noisyORParameters[paramsRowIdx][paramsColIdx];
++paramsRowIdx;
if (paramsRowIdx == (childDimension - 1)) {
++paramsColIdx;
paramsRowIdx = 0;
}
}
}
}
}
}
double sumColumn = 0.0;
for (int c_idx = 0; c_idx < childDimension; ++c_idx) { // LEAK parameters
if (c_idx == childDimension - 1) {
theProbs[probsIdx++] = 1.0 - sumColumn;
} else {
sumColumn += theProbs[probsIdx++] = noisyORParameters[paramsRowIdx][paramsColIdx];
++paramsRowIdx;
if (paramsRowIdx == (childDimension - 1)) {
++paramsColIdx;
paramsRowIdx = 0;
}
}
}
childNode->Definition()->SetDefinition(theProbs);
childNode->Definition()->GetMatrix()->Normalize();
}
void printCPT(DSL_node *node) {
DSL_network* net = node->Network(); // node network
int handle = node->Handle();
DSL_nodeDefinition *def = node->Definition();
const DSL_Dmatrix &cpt = *def->GetMatrix();
const DSL_idArray &outcomes = *def->GetOutcomesNames();
const DSL_intArray &parents = net->GetParents(handle);
int parentCount = parents.NumItems();
DSL_intArray coords;
// for (int elemIdx = 0; elemIdx < cpt.GetSize(); elemIdx ++) { for (int parentIdx = 0; parentIdx < parentCount; parentIdx ++){ } }
for (int elemIdx = 0; elemIdx < cpt.GetSize(); elemIdx ++) {
string name = "";
cpt.IndexToCoordinates(elemIdx, coords);
//cout << "P(" << node->GetId() << " = " << outcomes[coords[parentCount]] << " | ";
for (int parentIdx = 0; parentIdx < parentCount; parentIdx ++) {
DSL_node *parentNode = net->GetNode(parents[parentIdx]);
if(elemIdx == 0) {
cout << parentNode->GetId()<< " ";
if(parentIdx == parentCount-1) {
cout<< node->GetId() <<endl;
}
}
const DSL_idArray &parentStates = *parentNode->Definition()->GetOutcomesNames();
//cout << parentNode->GetId() << " = " << parentStates[coords[parentIdx]];
name += parentStates[coords[parentIdx]];
name += " ";
}
name += outcomes[coords[parentCount]];
cout << name << " " << cpt[elemIdx] << endl;
//cout << ") = " << cpt[elemIdx] << endl;
}
}
cptMap get_cptmap(DSL_node *node) {
DSL_network* net = node->Network(); // node network
int handle = node->Handle();
DSL_nodeDefinition *def = node->Definition();
const DSL_Dmatrix &cpt = *def->GetMatrix();
const DSL_idArray &outcomes = *def->GetOutcomesNames();
const DSL_intArray &parents = net->GetParents(handle);
int parentCount = parents.NumItems();
DSL_intArray coords;
cptMap cptmap;
for (int elemIdx = 0; elemIdx < cpt.GetSize(); elemIdx ++) {
cpt.IndexToCoordinates(elemIdx, coords);
//cout << "P(" << node->GetId() << " = " << outcomes[coords[parentCount]] << " | ";
keyMap km;
for (int parentIdx = 0; parentIdx < parentCount; parentIdx ++) {
// if (parentIdx > 0) cout << ", ";
DSL_node *parentNode = net->GetNode(parents[parentIdx]);
const DSL_idArray &parentStates = *parentNode->Definition()->GetOutcomesNames();
km[string(parentNode->GetId())] = string(parentStates[coords[parentIdx]]);
//cout << parentNode->GetId() << " = " << parentStates[coords[parentIdx]];
}
//printMap(km);
if (cptmap.count(km) == 0){
valMap vm;
vm[string(outcomes[coords[parentCount]])] = cpt[elemIdx];
cptmap[km] = vm;
} else {
cptmap[km][string(outcomes[coords[parentCount]])] = cpt[elemIdx];
}
//cout << ") = " << cpt[elemIdx] << endl;
}
return cptmap;
}
void node_SetName(void * void_node, char * name )
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
node->Info().Header().SetName(name);
}
int node_Handle(void * void_node)
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
return node->Handle();
}
void node_GetName(void * void_node, char * retval)
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
sprintf( retval, "%s", node->Info().Header().GetName() );
}
int node_ChangeType(void * void_node, int newType)
{
DSL_node * node = reinterpret_cast<DSL_node*>(void_node);
return node->ChangeType(newType);
}
LearningInfo(string data_infile, string network_infile, string child_name) {
if (dataSet.ReadFile(data_infile.c_str()) != DSL_OKAY) {
cout << "Cannot read data file... exiting." << endl;
exit(1);
}
if (originalNet.ReadFile(network_infile.c_str(), DSL_XDSL_FORMAT) != DSL_OKAY) {
cout << "Cannot read network... exiting." << endl;
exit(1);
}
string err;
if (dataSet.MatchNetwork(originalNet, matches, err) != DSL_OKAY) {
cout << "Cannot match network... exiting." << endl;
exit(1);
}
for(unsigned int i=0 ; i < matches.size() ; ++i) {
matchNetToData[matches[i].node] = matches[i].column;
matchDataToNet[matches[i].column] = matches[i].node;
}
childIdx = originalNet.FindNode(child_name.c_str());
childNode = originalNet.GetNode(childIdx);
if (childNode->Definition()->GetType() != (DSL_CHANCE | DSL_DISCRETE | DSL_NOISY_MAX) ) {
cout << "Child should be a NoisyMAX... exiting" << endl;
// ewentualnie zmienic na noisy-max ręcznie
exit(1);
}
childMAXDefinition = new DSL_noisyMAX(*(childNode->Definition()));
DSL_intArray &parents = originalNet.GetParents(childNode->Handle());
numberOfParents = parents.NumItems();
parentIndices = vector<int>(numberOfParents, 0);
for(int i=0; i<numberOfParents; ++i)
parentIndices[i] = parents[i];
childDimension = childNode->Definition()->GetNumberOfOutcomes();
parentDimensions = vector<int>(numberOfParents, 0);
sumParentDimensions = 0;
parentOutcomesStrengths = vector<DSL_intArray>(numberOfParents);
minimalNumberOfParameters = 1; // minimal number of unique parameters to calculate (count leak right away)
for(int parentIdx = 0 ; parentIdx < numberOfParents ; ++parentIdx) {
DSL_node *parentNode = originalNet.GetNode(parentIndices[parentIdx]);
sumParentDimensions += (parentDimensions[parentIdx] = parentNode->Definition()->GetNumberOfOutcomes()); //parent dimension is equal to the number of outcomes
parentOutcomesStrengths[parentIdx] = childMAXDefinition->GetParentOutcomeStrengths(parentIdx);
//for (int stateIdx=0 ; stateIdx < parentDimensions[parentIdx] ; ++stateIdx)
// cout << parentOutcomesStrengths[parentIdx][stateIdx] << " ";
//cout << endl;
minimalNumberOfParameters += parentDimensions[parentIdx] - 1; // (each parent dimension reduced by one) because we don't count distinguished states of parents
distinguishedStates[parentIdx] = parentOutcomesStrengths[parentIdx][parentDimensions[parentIdx] - 1];
}
int sumOffset = 0;
parameterRowOffset = vi1D(numberOfParents + 1, 0); // +1 so we know the offset for LEAK column
for(int parentIdx = 0; parentIdx < numberOfParents ; ++parentIdx) {
parameterRowOffset[parentIdx] = sumOffset;
sumOffset += parentDimensions[parentIdx] - 1;
}
parameterRowOffset[numberOfParents] = sumOffset;
parametersRowLength = minimalNumberOfParameters;
minimalNumberOfParameters *= (childDimension - 1); // number of unique rows, last row is always 1.0 - sum
// DEBUG(minimalNumberOfParameters);
// DEBUG(childDimension);
// DEBUGV(parentDimensions);
// DEBUGV(parameterRowOffset);
//for(int j=0; j< 7 ; ++j) {
// DSL_datasetVarInfo vi = ds.GetVariableInfo(j);
// cout << "discreete:" << vi.discrete << " id:" << vi.id << endl << " missingInt:" << vi.missingInt << " mF:" << vi.missingFloat << "snames:"<< endl;
// for(int i=0;i<vi.stateNames.size(); ++i)
// cout << vi.stateNames[i]<< " ";
// cout <<endl;
//}
//for(int i = 0; i < ds.GetNumberOfRecords(); ++i) {
// vector<int> row(ds.GetNumberOfVariables(), 0);
// int sum_ones = 0;
// for(int j = 0; j < ds.GetNumberOfVariables(); ++j) {
// sum_ones += (row[j] = ds.GetInt(j,i));
// }
//}
//vector<int> rd = ds.GetIntData(0);
//cout <<"RDSize:"<<rd.size()<< endl;
//for(int i=0;i<rd.size();++i) {
// cout << vi.stateNames[rd[i]] << endl;
//}
//
}
