func main() {
rand.Seed(time.Now().UnixNano())
positives := []svm.Sample{
{V: []float64{0.5, 0}},
{V: []float64{0.4, 0.1}},
{V: []float64{0.3, 0.2}},
{V: []float64{0.5, 0.9}},
{V: []float64{0.6, 1}},
{V: []float64{0.4, 0.85}},
{V: []float64{0.5, 0.5}},
}
negatives := []svm.Sample{
{V: []float64{0, 0.5}},
{V: []float64{0.1, 0.4}},
{V: []float64{0.9, 0.5}},
{V: []float64{1, 0.6}},
{V: []float64{0, 0.46}},
}
problem := &svm.Problem{
Positives: positives,
Negatives: negatives,
Kernel: svm.PolynomialKernel(1, 2),
}
printProblem(problem)
printSolution("random linear", svm.RandomlySolveLinear(problem, 100000, 5), problem)
subgradientSolver := &svm.SubgradientSolver{
Tradeoff: 0.001,
StepSize: 0.1,
Steps: 100000,
}
printSolution("subgradient descent", subgradientSolver.Solve(problem), problem)
gradientSolver := &svm.GradientDescentSolver{
Tradeoff: 0.001,
Timeout: time.Minute,
}
printSolution("gradient descent", gradientSolver.Solve(problem), problem)
}
func main() {
rand.Seed(time.Now().UnixNano())
positives := []svm.Sample{
svm.Sample{V: []float64{0, 0}},
svm.Sample{V: []float64{0, 0.3}},
svm.Sample{V: []float64{0.3, 0}},
svm.Sample{V: []float64{0.1, 0.1}},
}
negatives := []svm.Sample{
svm.Sample{V: []float64{1, 1}},
svm.Sample{V: []float64{1, 0.7}},
svm.Sample{V: []float64{0.7, 1}},
svm.Sample{V: []float64{0.9, 0.9}},
}
problem := &svm.Problem{
Positives: positives,
Negatives: negatives,
Kernel: svm.LinearKernel,
}
solution := svm.RandomlySolveLinear(problem, 100000, 5)
fmt.Println("Solution from random solver:", solution)
subgradientSolver := &svm.SubgradientSolver{
Tradeoff: 0.001,
Steps: 10000,
StepSize: 0.001,
}
solution = subgradientSolver.Solve(problem)
fmt.Println("Solution from subgradient solver:", solution)
gradientSolver := &svm.GradientDescentSolver{
Tradeoff: 0.0001,
Timeout: time.Minute,
}
solution = gradientSolver.Solve(problem).Linearize()
fmt.Println("Solution from gradient descent solver:", solution)
}