def computeDecisionBoundary(self, w_full, lam, stepSize=0.01, maxFunctionCalls=10000, printSummary=True,
plot=False, plotIter=False, useGradientCriterion=False, tol=1e-4):
gd = self.constructGradientDescentObject(lam)
gd.stepSize = stepSize
storeIterValues=False
if plotIter:
storeIterValues=True
sol = gd.computeMin(w_full, maxFunctionCalls=maxFunctionCalls, printSummary=printSummary,
storeIterValues=storeIterValues, tol=tol,
useGradientCriterion=useGradientCriterion)
w_star = sol[0];
w_star_normalized = 1/np.linalg.norm(w_star)*w_star
if printSummary:
print "--- Classification Summary ---"
print "w_full = " + str(w_star)
print "w_full normalized = " + str(w_star_normalized)
print "norm of w_full = " + str(np.linalg.norm(w_star))
print "lambda = " + str(lam)
self.classificationErrorRate(w_star, verbose=True)
print "------------------"
print ""
if plot:
self.plotData(w_star)
if plotIter:
gd.plotIterValues()
return w_star
def computeDecisionBoundary(self,
w_full,
lam,
stepSize=0.01,
maxFunctionCalls=10000,
printSummary=True,
plot=False,
plotIter=False,
useGradientCriterion=False,
tol=1e-4):
gd = self.constructGradientDescentObject(lam)
gd.stepSize = stepSize
storeIterValues = False
if plotIter:
storeIterValues = True
sol = gd.computeMin(w_full,
maxFunctionCalls=maxFunctionCalls,
printSummary=printSummary,
storeIterValues=storeIterValues,
tol=tol,
useGradientCriterion=useGradientCriterion)
w_star = sol[0]
w_star_normalized = 1 / np.linalg.norm(w_star) * w_star
if printSummary:
print "--- Classification Summary ---"
print "w_full = " + str(w_star)
print "w_full normalized = " + str(w_star_normalized)
print "norm of w_full = " + str(np.linalg.norm(w_star))
print "lambda = " + str(lam)
self.classificationErrorRate(w_star, verbose=True)
print "------------------"
print ""
if plot:
self.plotData(w_star)
if plotIter:
gd.plotIterValues()
return w_star
