svm_c_builder.cpp

ADaM is a data mining and image processing toolkit

		active_size++;
		++k;	// look at the newcomer
	}
}

double Solver_NU::calculate_rho()
{
	int nr_free1 = 0,nr_free2 = 0;
	double ub1 = INF, ub2 = INF;
	double lb1 = -INF, lb2 = -INF;
	double sum_free1 = 0, sum_free2 = 0;

	for(int i=0;i<active_size;i++)
	{
		if(y[i]==+1)
		{
			if(is_lower_bound(i))
				ub1 = min(ub1,G[i]);
			else if(is_upper_bound(i))
				lb1 = max(lb1,G[i]);
			else
			{
				++nr_free1;
				sum_free1 += G[i];
			}
		}
		else
		{
			if(is_lower_bound(i))
				ub2 = min(ub2,G[i]);
			else if(is_upper_bound(i))
				lb2 = max(lb2,G[i]);
			else
			{
				++nr_free2;
				sum_free2 += G[i];
			}
		}
	}

	double r1,r2;
	if(nr_free1 > 0)
		r1 = sum_free1/nr_free1;
	else
		r1 = (ub1+lb1)/2;
	
	if(nr_free2 > 0)
		r2 = sum_free2/nr_free2;
	else
		r2 = (ub2+lb2)/2;
	
	si->r = (r1+r2)/2;
	return (r1-r2)/2;
}

//
// Q matrices for various formulations
//
class SVC_Q: public Kernel
{ 
public:
	SVC_Q(const svm_problem& prob, const svm_parameter& param, const schar *y_)
	:Kernel(prob.l, prob.x, param)
	{
		clone(y,y_,prob.l);
		cache = new Cache(prob.l,(int)(param.cache_size*(1<<20)));
		QD = new Qfloat[prob.l];
		for(int i=0;i<prob.l;i++)
			QD[i]= (Qfloat)(this->*kernel_function)(i,i);
	}
	
	Qfloat *get_Q(int i, int len) const
	{
		Qfloat *data;
		int start;
		if((start = cache->get_data(i,&data,len)) < len)
		{
			for(int j=start;j<len;j++)
				data[j] = (Qfloat)(y[i]*y[j]*(this->*kernel_function)(i,j));
		}
		return data;
	}

	Qfloat *get_QD() const
	{
		return QD;
	}

	void swap_index(int i, int j) const
	{
		cache->swap_index(i,j);
		Kernel::swap_index(i,j);
		swap(y[i],y[j]);
		swap(QD[i],QD[j]);
	}

	~SVC_Q()
	{
		delete[] y;
		delete cache;
		delete[] QD;
	}
private:
	schar *y;
	Cache *cache;
	Qfloat *QD;
};

class ONE_CLASS_Q: public Kernel
{
public:
	ONE_CLASS_Q(const svm_problem& prob, const svm_parameter& param)
	:Kernel(prob.l, prob.x, param)
	{
		cache = new Cache(prob.l,(int)(param.cache_size*(1<<20)));
		QD = new Qfloat[prob.l];
		for(int i=0;i<prob.l;i++)
			QD[i]= (Qfloat)(this->*kernel_function)(i,i);
	}
	
	Qfloat *get_Q(int i, int len) const
	{
		Qfloat *data;
		int start;
		if((start = cache->get_data(i,&data,len)) < len)
		{
			for(int j=start;j<len;j++)
				data[j] = (Qfloat)(this->*kernel_function)(i,j);
		}
		return data;
	}

	Qfloat *get_QD() const
	{
		return QD;
	}

	void swap_index(int i, int j) const
	{
		cache->swap_index(i,j);
		Kernel::swap_index(i,j);
		swap(QD[i],QD[j]);
	}

	~ONE_CLASS_Q()
	{
		delete cache;
		delete[] QD;
	}
private:
	Cache *cache;
	Qfloat *QD;
};

class SVR_Q: public Kernel
{ 
public:
	SVR_Q(const svm_problem& prob, const svm_parameter& param)
	:Kernel(prob.l, prob.x, param)
	{
		l = prob.l;
		cache = new Cache(l,(int)(param.cache_size*(1<<20)));
		QD = new Qfloat[2*l];
		sign = new schar[2*l];
		index = new int[2*l];
		for(int k=0;k<l;k++)
		{
			sign[k] = 1;
			sign[k+l] = -1;
			index[k] = k;
			index[k+l] = k;
			QD[k]= (Qfloat)(this->*kernel_function)(k,k);
			QD[k+l]=QD[k];
		}
		buffer[0] = new Qfloat[2*l];
		buffer[1] = new Qfloat[2*l];
		next_buffer = 0;
	}

	void swap_index(int i, int j) const
	{
		swap(sign[i],sign[j]);
		swap(index[i],index[j]);
		swap(QD[i],QD[j]);
	}
	
	Qfloat *get_Q(int i, int len) const
	{
		Qfloat *data;
		int real_i = index[i];
		if(cache->get_data(real_i,&data,l) < l)
		{
			for(int j=0;j<l;j++)
				data[j] = (Qfloat)(this->*kernel_function)(real_i,j);
		}

		// reorder and copy
		Qfloat *buf = buffer[next_buffer];
		next_buffer = 1 - next_buffer;
		schar si = sign[i];
		for(int j=0;j<len;j++)
			buf[j] = si * sign[j] * data[index[j]];
		return buf;
	}

	Qfloat *get_QD() const
	{
		return QD;
	}

	~SVR_Q()
	{
		delete cache;
		delete[] sign;
		delete[] index;
		delete[] buffer[0];
		delete[] buffer[1];
		delete[] QD;
	}
private:
	int l;
	Cache *cache;
	schar *sign;
	int *index;
	mutable int next_buffer;
	Qfloat *buffer[2];
	Qfloat *QD;
};

//
// construct and solve various formulations
//
static void solve_c_svc(
	const svm_problem *prob, const svm_parameter* param,
	double *alpha, Solver::SolutionInfo* si, double Cp, double Cn)
{
	int l = prob->l;
	double *minus_ones = new double[l];
	schar *y = new schar[l];

	int i;

	for(i=0;i<l;i++)
	{
		alpha[i] = 0;
		minus_ones[i] = -1;
		if(prob->y[i] > 0) y[i] = +1; else y[i]=-1;
	}

	Solver s;
	s.Solve(l, SVC_Q(*prob,*param,y), minus_ones, y,
		alpha, Cp, Cn, param->eps, si, param->shrinking);

	double sum_alpha=0;
	for(i=0;i<l;i++)
		sum_alpha += alpha[i];

	if (Cp==Cn)
		info("nu = %f\n", sum_alpha/(Cp*prob->l));

	for(i=0;i<l;i++)
		alpha[i] *= y[i];

	delete[] minus_ones;
	delete[] y;
}

static void solve_nu_svc(
	const svm_problem *prob, const svm_parameter *param,
	double *alpha, Solver::SolutionInfo* si)
{
	int i;
	int l = prob->l;
	double nu = param->nu;

	schar *y = new schar[l];

	for(i=0;i<l;i++)
		if(prob->y[i]>0)
			y[i] = +1;
		else
			y[i] = -1;

	double sum_pos = nu*l/2;
	double sum_neg = nu*l/2;

	for(i=0;i<l;i++)
		if(y[i] == +1)
		{
			alpha[i] = min(1.0,sum_pos);
			sum_pos -= alpha[i];
		}
		else
		{
			alpha[i] = min(1.0,sum_neg);
			sum_neg -= alpha[i];
		}

	double *zeros = new double[l];

	for(i=0;i<l;i++)
		zeros[i] = 0;

	Solver_NU s;
	s.Solve(l, SVC_Q(*prob,*param,y), zeros, y,
		alpha, 1.0, 1.0, param->eps, si,  param->shrinking);
	double r = si->r;

	info("C = %f\n",1/r);

	for(i=0;i<l;i++)
		alpha[i] *= y[i]/r;

	si->rho /= r;
	si->obj /= (r*r);
	si->upper_bound_p = 1/r;
	si->upper_bound_n = 1/r;

	delete[] y;
	delete[] zeros;
}

static void solve_one_class(
	const svm_problem *prob, const svm_parameter *param,
	double *alpha, Solver::SolutionInfo* si)
{
	int l = prob->l;
	double *zeros = new double[l];
	schar *ones = new schar[l];
	int i;

	int n = (int)(param->nu*prob->l);	// # of alpha's at upper bound

	for(i=0;i<n;i++)
		alpha[i] = 1;
	if(n<prob->l)
		alpha[n] = param->nu * prob->l - n;
	for(i=n+1;i<l;i++)
		alpha[i] = 0;

	for(i=0;i<l;i++)
	{
		zeros[i] = 0;
		ones[i] = 1;
	}

	Solver s;
	s.Solve(l, ONE_CLASS_Q(*prob,*param), zeros, ones,
		alpha, 1.0, 1.0, param->eps, si, param->shrinking);

	delete[] zeros;
	delete[] ones;
}

static void solve_epsilon_svr(
	const svm_problem *prob, const svm_parameter *param,
	double *alpha, Solver::SolutionInfo* si)
{
	int l = prob->l;
	double *alpha2 = new double[2*l];
	double *linear_term = new double[2*l];
	schar *y = new schar[2*l];
	int i;

	for(i=0;i<l;i++)
	{
		alpha2[i] = 0;
		linear_term[i] = param->p - prob->y[i];
		y[i] = 1;

		alpha2[i+l] = 0;
		linear_term[i+l] = param->p + prob->y[i];
		y[i+l] = -1;
	}

	Solver s;
	s.Solve(2*l, SVR_Q(*prob,*param), linear_term, y,
		alpha2, param->C, param->C, param->eps, si, param->shrinking);

	double sum_alpha = 0;
	for(i=0;i<l;i++)
	{
		alpha[i] = alpha2[i] - alpha2[i+l];
		sum_alpha += fabs(alpha[i]);
	}
	info("nu = %f\n",sum_alpha/(param->C*l));

	delete[] alpha2;
	delete[] linear_term;
	delete[] y;
}

static void solve_nu_svr(
	const svm_problem *prob, const svm_parameter *param,
	double *alpha, Solver::SolutionInfo* si)
{
	int l = prob->l;
	double C = param->C;
	double *alpha2 = new double[2*l];
	double *linear_term = new double[2*l];
	schar *y = new schar[2*l];
	int i;

	double sum = C * param->nu * l / 2;
	for(i=0;i<l;i++)
	{
		alpha2[i] = alpha2[i+l] = min(sum,C);
		sum -= alpha2[i];

		linear_term[i] = - prob->y[i];
		y[i] = 1;

		linear_term[i+l] = prob->y[i];
		y[i+l] = -1;
	}

	Solver_NU s;
	s.Solve(2*l, SVR_Q(*prob,*param), linear_term, y,
		alpha2, C, C, param->eps, si, param->shrinking);

	info("epsilon = %f\n",-si->r);

	for(i=0;i<l;i++)
		alpha[i] = alpha2[i] - alpha2[i+l];

	delete[] alpha2;
	delete[] linear_term;
	delete[] y;
}

//
// decision_function
//
struct decision_function
{
	double *alpha;
	double rho;	
};

decision_function svm_train_one(
	const svm_problem *prob, const svm_parameter *param,
	double Cp, double Cn)
{
	double *alpha = Malloc(double,prob->l);
	Solver::SolutionInfo si;
	switch(param->svm_type)
	{
		case C_SVC:
			solve_c_svc(prob,param,alpha,&si,Cp,Cn);
			break;
		case NU_SVC:
			solve_nu_svc(prob,param,alpha,&si);
			break;
		case ONE_CLASS:
			solve_one_class(prob,param,alpha,&si);
			break;
		case EPSILON_SVR:
			solve_epsilon_svr(prob,param,alpha,&si);
			break;
		case NU_SVR:
			solve_nu_svr(prob,param,alpha,&si);
			break;
	}

	info("obj = %f, rho = %f\n",si.obj,si.rho);

	// output SVs

	int nSV = 0;
	int nBSV = 0;
	for(int i=0;i<prob->l;i++)
	{
		if(fabs(alpha[i]) > 0)
		{
			++nSV;
			if(prob->y[i] > 0)
			{
				if(fabs(alpha[i]) >= si.upper_bound_p)
					++nBSV;
			}
			else
			{
				if(fabs(alpha[i]) >= si.upper_bound_n)
					++nBSV;
			}
		}
	}

	info("nSV = %d, nBSV = %d\n",nSV,nBSV);

	decision_function f;
	f.alpha = alpha;
	f.rho = si.rho;
	return f;
}

//
// svm_model
//
struct svm_model
{
	svm_parameter param;	// parameter
	int nr_class;		// number of classes, = 2 in regression/one class svm
	int l;			// total #SV
	svm_node **SV;		// SVs (SV[l])
	double **sv_coef;	// coefficients for SVs in decision functions (sv_coef[n-1][l])
	double *rho;		// constants in decision functions (rho[n*(n-1)/2])
	double *probA;          // pariwise probability information
	double *probB;

	// for classification only

	int *label;		// label of each class (label[n])
	int *nSV;		// number of SVs for each class (nSV[n])
				// nSV[0] + nSV[1] + ... + nSV[n-1] = l
	// XXX
	int free_sv;		// 1 if svm_model is created by svm_load_model
				// 0 if svm_model is created by svm_train
};

// Platt's binary SVM Probablistic Output: an improvement from Lin et al.
void sigmoid_train(
	int l, const double *dec_values, const double *labels, 
	double& A, double& B)
{
	double prior1=0, prior0 = 0;
	int i;

	for (i=0;i<l;i++)
		if (labels[i] > 0) prior1+=1;
		else prior0+=1;
	
	int max_iter=100; 	// Maximal number of iterations
	double min_step=1e-10;	// Minimal step taken in line search
	double sigma=1e-3;	// For numerically strict PD of Hessian
	double eps=1e-5;
	double hiTarget=(prior1+1.0)/(prior1+2.0);
	double loTarget=1/(prior0+2.0);
	double *t=Malloc(double,l);
	double fApB,p,q,h11,h22,h21,g1,g2,det,dA,dB,gd,stepsize;
	double newA,newB,newf,d1,d2;
	int iter; 
	
	// Initial Point and Initial Fun Value
	A=0.0; B=log((prior0+1.0)/(prior1+1.0));
	double fval = 0.0;

	for (i=0;i<l;i++)
	{
		if (labels[i]>0) t[i]=hiTarget;
		else t[i]=loTarget;
		fApB = dec_values[i]*A+B;
		if (fApB>=0)
			fval += t[i]*fApB + log(1+exp(-fApB));
		else
			fval += (t[i] - 1)*fApB +log(1+exp(fApB));
	}
	for (iter=0;iter<max_iter;iter++)
	{
		// Update Gradient and Hessian (use H' = H + sigma I)
		h11=sigma; // numerically ensures strict PD
		h22=sigma;
		h21=0.0;g1=0.0;g2=0.0;
		for (i=0;i<l;i++)
		{
			fApB = dec_values[i]*A+B;
			if (fApB >= 0)
			{
				p=exp(-fApB)/(1.0+exp(-fApB));
				q=1.0/(1.0+exp(-fApB));
			}
			else
			{
				p=1.0/(1.0+exp(fApB));
				q=exp(fApB)/(1.0+exp(fApB));
			}
			d2=p*q;
			h11+=dec_values[i]*dec_values[i]*d2;
			h22+=d2;
			h21+=dec_values[i]*d2;
			d1=t[i]-p;
			g1+=dec_values[i]*d1;
			g2+=d1;
		}

		// Stopping Criteria
		if (fabs(g1)<eps && fabs(g2)<eps)
			break;

		// Finding Newton direction: -inv(H') * g
		det=h11*h22-h21*h21;
		dA=-(h22*g1 - h21 * g2) / det;
		dB=-(-h21*g1+ h11 * g2) / det;
		gd=g1*dA+g2*dB;


		stepsize = 1; 		// Line Search
		while (stepsize >= min_step)
		{
			newA = A + stepsize * dA;
			newB = B + stepsize * dB;

			// New function value
			newf = 0.0;
			for (i=0;i<l;i++)
			{
				fApB = dec_values[i]*newA+newB;
				if (fApB >= 0)
					newf += t[i]*fApB + log(1+exp(-fApB));
				else
					newf += (t[i] - 1)*fApB +log(1+exp(fApB));
			}
			// Check sufficient decrease
			if (newf<fval+0.0001*stepsize*gd)
			{
				A=newA;B=newB;fval=newf;