using computation_controller macro (#4286)

src/shogun/classifier/AveragedPerceptron.cpp

@@ -71,8 +71,9 @@ bool CAveragedPerceptron::train_machine(CFeatures* data)
 
 	//loop till we either get everything classified right or reach max_iter
 
-	while (!(cancel_computation()) && (!converged && iter < max_iter))
+	while (!converged && iter < max_iter)
 	{
+		COMPUTATION_CONTROLLERS
 		converged=true;
 		SG_INFO("Iteration Number : %d of max %d\n", iter, max_iter);
 

src/shogun/classifier/LPBoost.cpp

@@ -122,8 +122,10 @@ bool CLPBoost::train_machine(CFeatures* data)
 	int32_t num_hypothesis=0;
 	CTime time;
 
-	while (!(cancel_computation()))
+	while (get_max_train_time() <= 0 ||
+	       time.cur_time_diff() <= get_max_train_time())
 	{
+		COMPUTATION_CONTROLLERS
 		int32_t max_dim=0;
 		float64_t violator=find_max_violator(max_dim);
 		SG_PRINT("iteration:%06d violator: %10.17f (>1.0) chosen: %d\n", num_hypothesis, violator, max_dim)
@@ -146,9 +148,6 @@ bool CLPBoost::train_machine(CFeatures* data)
 		solver.optimize(u);
 		//CMath::display_vector(u, num_vec, "u");
 		num_hypothesis++;
-
-		if (get_max_train_time()>0 && time.cur_time_diff()>get_max_train_time())
-			break;
 	}
 	float64_t* lambda=SG_MALLOC(float64_t, num_hypothesis);
 	solver.optimize(u, lambda);

src/shogun/classifier/Perceptron.cpp

@@ -75,8 +75,9 @@ bool CPerceptron::train_machine(CFeatures* data)
 	}
 
 	//loop till we either get everything classified right or reach max_iter
-	while (!(cancel_computation()) && (!converged && iter < max_iter))
+	while (!converged && iter < max_iter)
 	{
+		COMPUTATION_CONTROLLERS
 		converged=true;
 		auto iter_train_labels = train_labels.begin();
 		auto iter_output = output.begin();

src/shogun/classifier/mkl/MKL.cpp

@@ -459,27 +459,29 @@ bool CMKL::train_machine(CFeatures* data)
 	{
 		float64_t* sumw = SG_MALLOC(float64_t, num_kernels);
 
-
-
-		while (true)
+		while (get_max_train_time() <= 0 ||
+		       training_time_clock.cur_time_diff() <= get_max_train_time())
 		{
+			COMPUTATION_CONTROLLERS
 			svm->train();
 
 			float64_t suma=compute_sum_alpha();
 			compute_sum_beta(sumw);
-
-			if((training_time_clock.cur_time_diff()>get_max_train_time ())&&(get_max_train_time ()>0))
-			{
-				SG_SWARNING("MKL Algorithm terminates PREMATURELY due to current training time exceeding get_max_train_time ()= %f . It may have not converged yet!\n",get_max_train_time ())
-				break;
-			}
-
-
 			mkl_iterations++;
-			if (perform_mkl_step(sumw, suma) || cancel_computation())
+			if (perform_mkl_step(sumw, suma))
 				break;
 		}
 
+		if ((training_time_clock.cur_time_diff() > get_max_train_time()) &&
+		    (get_max_train_time() > 0))
+		{
+			SG_SWARNING(
+			    "MKL Algorithm terminates PREMATURELY due to current training "
+			    "time exceeding get_max_train_time ()= %f . It may have not "
+			    "converged yet!\n",
+			    get_max_train_time())
+		}
+
 		SG_FREE(sumw);
 	}
 #ifdef USE_CPLEX

src/shogun/classifier/mkl/MKLMulticlass.cpp

@@ -366,9 +366,9 @@ bool CMKLMulticlass::train_machine(CFeatures* data)
 	int32_t numberofsilpiterations=0;
 	bool final=false;
 
-	while (!(cancel_computation()) && !final)
+	while (!final)
 	{
-
+		COMPUTATION_CONTROLLERS
 		//curweights.clear();
 		lpw->computeweights(curweights);
 		weightshistory.push_back(curweights);

src/shogun/classifier/svm/LibLinear.cpp

@@ -326,8 +326,9 @@ void CLibLinear::solve_l2r_l1l2_svc(
 
 	auto pb = progress(range(10));
 	CTime start_time;
-	while (iter < get_max_iterations() && !cancel_computation())
+	while (iter < get_max_iterations())
 	{
+		COMPUTATION_CONTROLLERS
 		if (m_max_train_time > 0 &&
 		    start_time.cur_time_diff() > m_max_train_time)
 			break;
@@ -539,8 +540,9 @@ void CLibLinear::solve_l1r_l2_svc(
 
 	auto pb = progress(range(10));
 	CTime start_time;
-	while (iter < get_max_iterations() && !cancel_computation())
+	while (iter < get_max_iterations())
 	{
+		COMPUTATION_CONTROLLERS
 		if (m_max_train_time > 0 &&
 		    start_time.cur_time_diff() > m_max_train_time)
 			break;
@@ -913,8 +915,9 @@ void CLibLinear::solve_l1r_lr(
 
 	auto pb = progress(range(10));
 	CTime start_time;
-	while (iter < get_max_iterations() && !cancel_computation())
+	while (iter < get_max_iterations())
 	{
+		COMPUTATION_CONTROLLERS
 		if (m_max_train_time > 0 &&
 		    start_time.cur_time_diff() > m_max_train_time)
 			break;

src/shogun/classifier/svm/MPDSVM.cpp

@@ -92,8 +92,9 @@ bool CMPDSVM::train_machine(CFeatures* data)
 	}
 
 	// go ...
-	while (niter++ < maxiter && !cancel_computation())
+	while (niter++ < maxiter)
 	{
+		COMPUTATION_CONTROLLERS
 		int32_t maxpidx=-1;
 		float64_t maxpviol = -1;
 		//float64_t maxdviol = CMath::abs(detas[0]);

src/shogun/classifier/svm/NewtonSVM.cpp

@@ -80,16 +80,11 @@ bool CNewtonSVM::train_machine(CFeatures* data)
 	float64_t obj, *grad=SG_MALLOC(float64_t, x_d+1);
 	float64_t t;
 
-	while (!cancel_computation())
+	while (iter <= num_iter)
 	{
+		COMPUTATION_CONTROLLERS
 		iter++;
 
-		if (iter>num_iter)
-		{
-			SG_PRINT("Maximum number of Newton steps reached. Try larger lambda")
-			break;
-		}
-
 		obj_fun_linear(weights, out, &obj, sv, &size_sv, grad);
 
 #ifdef DEBUG_NEWTON
@@ -211,6 +206,11 @@ bool CNewtonSVM::train_machine(CFeatures* data)
 			break;
 	}
 
+	if (iter > num_iter)
+	{
+		SG_PRINT("Maximum number of Newton steps reached. Try larger lambda")
+	}
+
 #ifdef V_NEWTON
 	SG_PRINT("FINAL W AND BIAS Vector=\n\n")
 	CMath::display_matrix(weights, x_d+1, 1);

src/shogun/classifier/svm/OnlineSVMSGD.cpp

@@ -89,8 +89,9 @@ bool COnlineSVMSGD::train(CFeatures* data)
 		is_log_loss = true;
 
 	int32_t vec_count;
-	for (int32_t e = 0; e < epochs && (!cancel_computation()); e++)
+	for (int32_t e = 0; e < epochs; e++)
 	{
+		COMPUTATION_CONTROLLERS
 		vec_count=0;
 		count = skip;
 		while (features->get_next_example())

src/shogun/classifier/svm/SGDQN.cpp

@@ -129,8 +129,9 @@ bool CSGDQN::train(CFeatures* data)
 	if ((loss_type == L_LOGLOSS) || (loss_type == L_LOGLOSSMARGIN))
 		is_log_loss = true;
 
-	for (int32_t e = 0; e < epochs && (!cancel_computation()); e++)
+	for (int32_t e = 0; e < epochs; e++)
 	{
+		COMPUTATION_CONTROLLERS
 		count = skip;
 		bool updateB=false;
 		for (int32_t i=0; i<num_vec; i++)

src/shogun/classifier/svm/SVMLight.cpp

@@ -647,13 +647,12 @@ int32_t CSVMLight::optimize_to_convergence(int32_t* docs, int32_t* label, int32_
   for (;((iteration<100 || (!mkl_converged && callback) ) || (retrain && (!terminate))); iteration++){
 #else
 
-  for (; ((!cancel_computation()) &&
-	      ((iteration < 3 || (!mkl_converged && callback)) ||
-	       (retrain && (!terminate))));
+  for (; (iteration < 3 || (!mkl_converged && callback)) ||
+	     (retrain && (!terminate));
 	   iteration++)
   {
 #endif
-
+	  COMPUTATION_CONTROLLERS
 	  if(use_kernel_cache)
 		  kernel->set_time(iteration);  /* for lru cache */
 

src/shogun/machine/KernelMachine.cpp

@@ -357,13 +357,9 @@ SGVector<float64_t> CKernelMachine::apply_get_outputs(CFeatures* data)
 				                  ? num_vectors
 				                  : (thread_num + 1) * step;
 
-#ifdef WIN32
 				for (int32_t vec = start; vec < end; vec++)
-#else
-				for (int32_t vec = start; vec < end && !cancel_computation();
-				     vec++)
-#endif
 				{
+					COMPUTATION_CONTROLLERS
 					pb.print_progress();
 
 					ASSERT(kernel)
@@ -386,11 +382,6 @@ SGVector<float64_t> CKernelMachine::apply_get_outputs(CFeatures* data)
 			}
 			pb.complete();
 		}
-
-#ifndef WIN32
-		if (cancel_computation())
-			SG_INFO("prematurely stopped.           \n")
-#endif
 	}
 
 	SG_DEBUG("leaving %s::apply_get_outputs(%s at %p)\n",
@@ -516,12 +507,10 @@ SGVector<float64_t> CKernelMachine::apply_locked_get_output(
 		int32_t start = thread_num * step;
 		int32_t end =
 		    (thread_num == num_threads) ? num_inds : (thread_num + 1) * step;
-#ifdef WIN32
+
 		for (int32_t vec = start; vec < end; vec++)
-#else
-		for (int32_t vec = start; vec < end && !cancel_computation(); vec++)
-#endif
 		{
+			COMPUTATION_CONTROLLERS
 			pb.print_progress();
 			index_t index = indices[vec];
 			ASSERT(kernel)
@@ -542,13 +531,7 @@ SGVector<float64_t> CKernelMachine::apply_locked_get_output(
 			}
 		}
 	}
-
-#ifndef WIN32
-	if (cancel_computation())
-		SG_INFO("prematurely stopped.\n")
-	else
-#endif
-		pb.complete();
+	pb.complete();
 
 	return output;
 }

src/shogun/multiclass/KNN.cpp

@@ -128,8 +128,9 @@ SGMatrix<index_t> CKNN::nearest_neighbors()
 	auto pb = progress(range(n), *this->io);
 
 	//for each test example
-	for (int32_t i = 0; i < n && (!cancel_computation()); i++)
+	for (int32_t i = 0; i < n; i++)
 	{
+		COMPUTATION_CONTROLLERS
 		pb.print_progress();
 
 		//lhs idx 0..num train examples-1 (i.e., all train examples) and rhs idx i
@@ -211,8 +212,9 @@ CMulticlassLabels* CKNN::classify_NN()
 	auto pb = progress(range(num_lab), *this->io);
 
 	// for each test example
-	for (int32_t i = 0; i < num_lab && (!cancel_computation()); i++)
+	for (int32_t i = 0; i < num_lab; i++)
 	{
+		COMPUTATION_CONTROLLERS
 		pb.print_progress();
 
 		// get distances from i-th test example to 0..num_m_train_labels-1 train examples

src/shogun/transfer/multitask/LibLinearMTL.cpp

@@ -251,8 +251,9 @@ void CLibLinearMTL::solve_l2r_l1l2_svc(const liblinear_problem *prob, double eps
 
 	auto pb = progress(range(10));
 	CTime start_time;
-	while (iter < max_iterations && !cancel_computation())
+	while (iter < max_iterations)
 	{
+		COMPUTATION_CONTROLLERS
 		if (m_max_train_time > 0 && start_time.cur_time_diff() > m_max_train_time)
 			break;