Promenio nazive ulaza na ML u za treniranje modela.

5 files changed, 65 insertions, 29 deletions

diff --git a/backend/microservice/api/newmlservice.py b/backend/microservice/api/newmlservice.py
index ad848fd9..d84d9567 100644
--- a/backend/microservice/api/newmlservice.py
+++ b/backend/microservice/api/newmlservice.py
@@ -282,8 +282,8 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
     #
     # Podela na test i trening skupove
     #
-    test=paramsExperiment["randomTestSetDistribution"]
-    randomOrder = paramsExperiment["randomOrder"]
+    test=paramsModel["randomTestSetDistribution"]
+    randomOrder = paramsModel["randomOrder"]
     if(randomOrder):
         random=123
     else:
@@ -329,8 +329,8 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
     if(problem_type=='multi-klasifikacioni'):
         #print('multi')
         
-        reg=paramsModel['regularisation'][0]
-        regRate=float(paramsModel['regularisationRate'][0])
+        reg=paramsModel['layers'][0]['regularisation']
+        regRate=float(paramsModel['layers'][0]['regularisationRate'])
         if(reg=='l1'):
             kernelreg=tf.keras.regularizers.l1(regRate)
             biasreg=tf.keras.regularizers.l1(regRate)
@@ -341,12 +341,12 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
             activityreg=tf.keras.regularizers.l2(regRate)
 
         classifier=tf.keras.Sequential()
-        classifier.add(tf.keras.layers.Dense(units=paramsModel['hiddenLayerNeurons'], activation=paramsModel['hiddenLayerActivationFunctions'][0],input_dim=x_train.shape[1], kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#prvi skriveni + definisanje prethodnog-ulaznog
+        classifier.add(tf.keras.layers.Dense(units=paramsModel['layers'][0]['neurons'], activation=paramsModel['layers'][0]['activationFunction'],input_dim=x_train.shape[1], kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#prvi skriveni + definisanje prethodnog-ulaznog
        
         for i in range(paramsModel['hiddenLayers']-1):#ako postoji vise od jednog skrivenog sloja
             ###Kernel
-            reg=paramsModel['regularisation'][i+1]
-            regRate=float(paramsModel['regularisationRate'][i+1])
+            reg=paramsModel['layers'][i+1]['regularisation']
+            regRate=float(paramsModel['layers'][i+1]['regularisationRate'])
             if(reg=='l1'):
                 kernelreg=tf.keras.regularizers.l1(regRate)
                 biasreg=tf.keras.regularizers.l1(regRate)
@@ -356,7 +356,7 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
                 biasreg=tf.keras.regularizers.l2(regRate)
                 activityreg=tf.keras.regularizers.l2(regRate)
 
-            classifier.add(tf.keras.layers.Dense(units=paramsModel['hiddenLayerNeurons'], activation=paramsModel['hiddenLayerActivationFunctions'][i+1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#i-ti skriveni sloj
+            classifier.add(tf.keras.layers.Dense(units=paramsModel['layers'][i+1]['neurons'], activation=paramsModel['layers'][i+1]['activationFunction'],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#i-ti skriveni sloj
         
         classifier.add(tf.keras.layers.Dense(units=5, activation=paramsModel['outputLayerActivationFunction']))#izlazni sloj
 
@@ -386,8 +386,8 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
 
     elif(problem_type=='binarni-klasifikacioni'):
         #print('*************************************************************************binarni')
-        reg=paramsModel['regularisation'][0]
-        regRate=float(paramsModel['regularisationRate'][0])
+        reg=paramsModel['layers'][0]['regularisation']
+        regRate=float(paramsModel['layers'][0]['regularisationRate'])
         if(reg=='l1'):
             kernelreg=tf.keras.regularizers.l1(regRate)
             biasreg=tf.keras.regularizers.l1(regRate)
@@ -398,12 +398,12 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
             activityreg=tf.keras.regularizers.l2(regRate)
         classifier=tf.keras.Sequential()
     
-        classifier.add(tf.keras.layers.Dense(units=paramsModel['hiddenLayerNeurons'], activation=paramsModel['hiddenLayerActivationFunctions'][0],input_dim=x_train.shape[1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#prvi skriveni + definisanje prethodnog-ulaznog
+        classifier.add(tf.keras.layers.Dense(units=paramsModel['layers'][0]['neurons'], activation=paramsModel['layers'][0]['activationFunction'],input_dim=x_train.shape[1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#prvi skriveni + definisanje prethodnog-ulaznog
         
         for i in range(paramsModel['hiddenLayers']-1):#ako postoji vise od jednog skrivenog sloja
             #print(i)
-            reg=paramsModel['regularisation'][i+1]
-            regRate=float(paramsModel['regularisationRate'][i+1])
+            reg=paramsModel['layers'][i+1]['regularisation']
+            regRate=float(paramsModel['layers'][0]['regularisationRate'])
             if(reg=='l1'):
                 kernelreg=tf.keras.regularizers.l1(regRate)
                 biasreg=tf.keras.regularizers.l1(regRate)
@@ -412,12 +412,19 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
                 kernelreg=tf.keras.regularizers.l2(regRate)
                 biasreg=tf.keras.regularizers.l2(regRate)
                 activityreg=tf.keras.regularizers.l2(regRate)
-            classifier.add(tf.keras.layers.Dense(units=paramsModel['hiddenLayerNeurons'], activation=paramsModel['hiddenLayerActivationFunctions'][i+1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#i-ti skriveni sloj
+            classifier.add(tf.keras.layers.Dense(units=paramsModel['layers'][i+1]['neurons'], activation=paramsModel['layers'][i+1]['activationFunction'],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#i-ti skriveni sloj
         
         classifier.add(tf.keras.layers.Dense(units=1, activation=paramsModel['outputLayerActivationFunction']))#izlazni sloj
 
         classifier.compile(loss =paramsModel["lossFunction"] , optimizer = opt , metrics =paramsModel['metrics'])
 
+        print('AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA')
+        print(x_train)
+        print('AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA')
+        print(y_train)
+        print('AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA')
+
+
         history=classifier.fit(x_train, y_train, epochs = paramsModel['epochs'],batch_size=float(paramsModel['batchSize']),callbacks=callback(x_test, y_test,paramsModel['_id']))
         hist=history.history
         y_pred=classifier.predict(x_test)
@@ -434,8 +441,8 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
         return filepath,hist
 
     elif(problem_type=='regresioni'):
-        reg=paramsModel['regularisation'][0]
-        regRate=float(paramsModel['regularisationRate'][0])
+        reg=paramsModel['layers'][0]['regularisation']
+        regRate=float(paramsModel['layers'][0]['regularisationRate'])
         if(reg=='l1'):
             kernelreg=tf.keras.regularizers.l1(regRate)
             biasreg=tf.keras.regularizers.l1(regRate)
@@ -446,12 +453,12 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
             activityreg=tf.keras.regularizers.l2(regRate)
         classifier=tf.keras.Sequential()
     
-        classifier.add(tf.keras.layers.Dense(units=paramsModel['hiddenLayerNeurons'], activation=paramsModel['hiddenLayerActivationFunctions'][0],input_dim=x_train.shape[1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#prvi skriveni + definisanje prethodnog-ulaznog
+        classifier.add(tf.keras.layers.Dense(units=paramsModel['layers'][0]['neurons'], activation=paramsModel['layers'][0]['activationFunction'],input_dim=x_train.shape[1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#prvi skriveni + definisanje prethodnog-ulaznog
         
         for i in range(paramsModel['hiddenLayers']-1):#ako postoji vise od jednog skrivenog sloja
             #print(i)
-            reg=paramsModel['regularisation'][i+1]
-            regRate=float(paramsModel['regularisationRate'][i+1])
+            reg=paramsModel['layers'][i+1]['regularisation']
+            regRate=float(paramsModel['layers'][i+1]['regularisationRate'])
             if(reg=='l1'):
                 kernelreg=tf.keras.regularizers.l1(regRate)
                 biasreg=tf.keras.regularizers.l1(regRate)
@@ -461,7 +468,7 @@ def train(dataset, paramsModel,paramsExperiment,paramsDataset,callback):
                 biasreg=tf.keras.regularizers.l2(regRate)
                 activityreg=tf.keras.regularizers.l2(regRate)
 
-            classifier.add(tf.keras.layers.Dense(units=paramsModel['hiddenLayerNeurons'], activation=paramsModel['hiddenLayerActivationFunctions'][i+1],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#i-ti skriveni sloj
+            classifier.add(tf.keras.layers.Dense(units=paramsModel['layers'][i+1]['neurons'], activation=paramsModel['layers'][i+1]['activationFunction'],kernel_regularizer=kernelreg, bias_regularizer=biasreg, activity_regularizer=activityreg))#i-ti skriveni sloj
         
         classifier.add(tf.keras.layers.Dense(units=1),activation=paramsModel['outputLayerActivationFunction'])
 
diff --git a/frontend/src/app/_elements/column-table/column-table.component.ts b/frontend/src/app/_elements/column-table/column-table.component.ts
index 13828a2c..b99a3be0 100644
--- a/frontend/src/app/_elements/column-table/column-table.component.ts
+++ b/frontend/src/app/_elements/column-table/column-table.component.ts
@@ -223,7 +223,7 @@ export class ColumnTableComponent implements AfterViewInit {
       this.experiment.name = selectedName;
       //napravi odvojene dugmice za save i update -> za update nece da se otvara dijalog za ime
       this.experimentService.addExperiment(this.experiment).subscribe((response) => {
-        this.experiment = response;
+        this.experiment._id = response._id;
         this.okPressed.emit();
       });
     });
diff --git a/frontend/src/app/_elements/folder/folder.component.ts b/frontend/src/app/_elements/folder/folder.component.ts
index a99c6a9d..d5a7a85c 100644
--- a/frontend/src/app/_elements/folder/folder.component.ts
+++ b/frontend/src/app/_elements/folder/folder.component.ts
@@ -53,8 +53,9 @@ export class FolderComponent implements AfterViewInit {
 
     if (this.signalRService.hubConnection) {
       this.signalRService.hubConnection.on("NotifyDataset", (dName: string, dId: string) => {
-        this.refreshFiles(dId);
-
+        if (this.type == FolderType.Dataset) {
+          this.refreshFiles(dId);
+        }
       });
     } else {
       console.warn("Dataset-Load: No connection!");
diff --git a/frontend/src/app/_pages/experiment/experiment.component.html b/frontend/src/app/_pages/experiment/experiment.component.html
index 83c45405..2b32db81 100644
--- a/frontend/src/app/_pages/experiment/experiment.component.html
+++ b/frontend/src/app/_pages/experiment/experiment.component.html
@@ -37,12 +37,12 @@
         </div>
         <div #steps id="step_3" class="step-content">
             <div class="step-content-inside">
-                <app-folder #folderModel [type]="FolderType.Model" [forExperiment]="experiment" [startingTab]="TabType.NewFile" [tabsToShow]="[TabType.MyModels]" (okPressed)="goToPage(3)"></app-folder>
+                <app-folder #folderModel [type]="FolderType.Model" [forExperiment]="experiment" [startingTab]="TabType.NewFile" [tabsToShow]="[TabType.MyModels]" (okPressed)="goToPage(3); trainModel();" (selectedFileChanged)="setModel($event)"></app-folder>
             </div>
         </div>
         <div #steps id="step_4" class="step-content">
             <div class="step-content-inside">
-                <app-metric-view></app-metric-view>
+                <app-metric-view #metricView></app-metric-view>
             </div>
         </div>
     </div>
diff --git a/frontend/src/app/_pages/experiment/experiment.component.ts b/frontend/src/app/_pages/experiment/experiment.component.ts
index 3c8d8651..c4d6063c 100644
--- a/frontend/src/app/_pages/experiment/experiment.component.ts
+++ b/frontend/src/app/_pages/experiment/experiment.component.ts
@@ -10,6 +10,8 @@ import { ModelsService } from 'src/app/_services/models.service';
 import Model from 'src/app/_data/Model';
 import Dataset from 'src/app/_data/Dataset';
 import { ColumnTableComponent } from 'src/app/_elements/column-table/column-table.component';
+import { SignalRService } from 'src/app/_services/signal-r.service';
+import { MetricViewComponent } from 'src/app/_elements/metric-view/metric-view.component';
 
 @Component({
   selector: 'app-experiment',
@@ -26,11 +28,11 @@ export class ExperimentComponent implements AfterViewInit {
   experiment: Experiment;
   dataset?: Dataset;
   @ViewChild("folderDataset") folderDataset!: FolderComponent;
-  @ViewChild("folderModel") folderModel!: FolderComponent;
   @ViewChild(ColumnTableComponent) columnTable!: ColumnTableComponent;
+  @ViewChild("folderModel") folderModel!: FolderComponent;
+  @ViewChild("metricView") metricView!: MetricViewComponent;
 
-
-  constructor(private experimentsService: ExperimentsService, private modelsService: ModelsService) {
+  constructor(private experimentsService: ExperimentsService, private modelsService: ModelsService, private signalRService: SignalRService) {
     this.experiment = new Experiment("exp1");
   }
 
@@ -43,7 +45,11 @@ export class ExperimentComponent implements AfterViewInit {
   }
 
   trainModel() {
-    this.modelsService.trainModel((<Model>this.folderModel.selectedFile)._id, this.experiment._id).subscribe(() => { console.log("pocelo treniranje") })
+    if (!this.modelToTrain) {
+      Shared.openDialog('Greška', 'Morate odabrati konfiguraciju neuronske mreže');
+    } else {
+      this.modelsService.trainModel(this.modelToTrain._id, this.experiment._id).subscribe(() => { console.log("pocelo treniranje") });
+    }
   }
 
   stepHeight = this.calcStepHeight();
@@ -65,8 +71,23 @@ export class ExperimentComponent implements AfterViewInit {
     this.stepsContainer.nativeElement.addEventListener('scroll', (event: Event) => {
       Shared.emitBGScrollEvent(this.stepsContainer.nativeElement.scrollTop);
     });
+
+    if (this.signalRService.hubConnection) {
+      this.signalRService.hubConnection.on("NotifyEpoch", (mName: string, mId: string, stat: string, totalEpochs: number, currentEpoch: number) => {
+        console.log(this.modelToTrain?._id, mId);
+        if (this.modelToTrain?._id == mId) {
+          stat = stat.replace(/'/g, '"');
+          //console.log('JSON', this.trainingResult);
+          this.history.push(JSON.parse(stat));
+          this.metricView.update(this.history);
+        }
+      });
+
+    }
   }
 
+  history: any[] = [];
+
   updatePageIfScrolled() {
     if (this.scrolling) return;
     const currentPage = Math.round(this.stepsContainer.nativeElement.scrollTop / this.stepHeight)
@@ -130,4 +151,11 @@ export class ExperimentComponent implements AfterViewInit {
 
     this.columnTable.loadDataset(this.dataset);
   }
+
+  modelToTrain?: Model;
+
+  setModel(model: FolderFile) {
+    const m = <Model>model;
+    this.modelToTrain = m;
+  }
 }