Skip to content
Snippets Groups Projects
Commit 300fe29b authored by Baptiste Bauvin's avatar Baptiste Bauvin
Browse files

Faster with old code

parent 21abc7b0
No related branches found
No related tags found
No related merge requests found
Show whitespace changes
Inline Side-by-side
multiview_platform/MonoMultiViewClassifiers/Monoview/Additions/QarBoostUtils.py
+ 1
− 140
View file @ 300fe29b
......@@ -246,7 +246,7 @@ class ColumnGenerationClassifierQar(BaseEstimator, ClassifierMixin, BaseBoost):
self.example_weights_.append(self.example_weights)
if self.random_start:
first_voter_index = self.random_state.choice(
self.get_possible(y_kernel_matrix, y))
np.where(np.sum(y_kernel_matrix, axis=0)>0)[0])
else:
first_voter_index, _ = self._find_best_weighted_margin(
y_kernel_matrix)
......@@ -374,8 +374,6 @@ class ColumnGenerationClassifierQar(BaseEstimator, ClassifierMixin, BaseBoost):
if self.c_bound_sol:
return np.sum(hypothese) > 0
else:
# ones_matrix = np.zeros(y.shape)
# ones_matrix[hypothese.reshape(y.shape) < 0] = 1
print(np.average(hypothese.reshape(y.shape), weights=self.example_weights))
quit()
weighted_margin = np.average(hypothese.reshape(y.shape), weights=self.example_weights)#ondes matrix, axis=0
......@@ -392,10 +390,6 @@ class ColumnGenerationClassifierQar(BaseEstimator, ClassifierMixin, BaseBoost):
def _find_new_voter(self, y_kernel_matrix, y):
"""Here, we solve the two_voters_mincq_problem for each potential new voter,
and select the one that has the smallest minimum"""
c_bounds = []
possible_sols = []
indices = []
causes = []
m = y_kernel_matrix.shape[0]
weighted_previous_sum = np.multiply(y,
self.previous_vote.reshape((m, 1)))
......@@ -432,7 +426,6 @@ class ColumnGenerationClassifierQar(BaseEstimator, ClassifierMixin, BaseBoost):
return sols[best_hyp_index], best_hyp_index
def make_masked_c_bounds(self, sols, bad_margins):
c_bounds = self.compute_c_bounds(sols)
trans_c_bounds = self.compute_c_bounds(sols + 1)
......@@ -519,135 +512,3 @@ class ColumnGenerationClassifierQar(BaseEstimator, ClassifierMixin, BaseBoost):
interpretString += "\n\n The bound was not respected"
return interpretString
# def _solve_one_weight_min_c(self, next_column, y, margin_old, m):
# """Here we solve the min C-bound problem for two voters using one weight only and return the best weight
# No precalc because longer ; see the "derivee" latex document for more precision"""
# # zero_diag = np.ones((m, m)) - np.identity(m)
# weighted_previous_sum = np.multiply(y,
# self.previous_vote.reshape(
# (m, 1)))
# if self.c_bound_sol:
# weighted_next_column = next_column.reshape((m, 1))
# else:
# weighted_next_column = np.multiply(next_column.reshape((m, 1)),
# self.example_weights.reshape((m, 1)))
#
# self.B1 = np.sum(2 * weighted_next_column * weighted_previous_sum)
# self.B0 = np.sum(weighted_previous_sum ** 2)
# self.B2 = m
#
# # M2 = np.sum(np.multiply(
# # np.matmul(weighted_next_column, np.transpose(weighted_next_column)),
# # zero_diag))
# # plif = np.sum(np.multiply(weighted_next_column, y))**2 - m
# # print(M2, plif, "plouf")
# # M1 = np.sum(np.multiply(np.matmul(weighted_previous_sum,
# # np.transpose(weighted_next_column)) +
# # np.matmul(weighted_next_column,
# # np.transpose(weighted_previous_sum))
# # , zero_diag))
# # M0 = np.sum(np.multiply(np.matmul(weighted_previous_sum,
# # np.transpose(weighted_previous_sum)),
# # zero_diag))
# # self.A2 = self.B2 + M2
# # self.A1 = self.B1 + M1
# # self.A0 = self.B0 + M0
#
# # C2 = (M1 * self.B2 - M2 * self.B1)
# # C1 = 2 * (M0 * self.B2 - M2 * self.B0)
# # C0 = M0 * self.B1 - M1 * self.B0
#
# margin_new = np.sum(weighted_next_column)
# # margin_old = np.sum(weighted_previous_sum)
#
# self.A2 = margin_new**2
# self.A1 = 2*margin_new*margin_old
# self.A0 = margin_old**2
#
# C2 = (self.A1 * self.B2 - self.A2 * self.B1)
# C1 = 2 * (self.A0 * self.B2 - self.A2 * self.B0)
# C0 = self.A0 * self.B1 - self.A1 * self.B0
#
#
# det = C1*C1-4*C2*C0
#
# if C2 == 0:
# if C1 == 0:
# return ['break', "the derivate was constant"]
# else:
# sols = np.array([float(C0) / C1])
# elif det < 0:
# return ['break', "no real roots"]
# elif det == 0 :
# sols = np.array([-C1/(2*C2)])
# else:
# # print('yes')
# sols = np.array([(-C1 + math.sqrt(C1 * C1 - 4 * C2 * C0)) / (2 * C2)])
#
# is_acceptable, sol = self._analyze_solutions_one_weight(sols)
# if is_acceptable:
# return np.array([sol])
# else:
# return ["break", sol]
#
# def _analyze_solutions_one_weight(self, sols):
# """"We just check that the solution found by np.roots is acceptable under our constraints
# (real, a minimum and over 0)"""
# if sols.shape[0] == 2:
# best_sol = self._best_sol(sols)
# if best_sol > 0:
# if best_sol == sols[0]:
# print("moins")
# return True, best_sol
# else:
# return False, "sol < 0"
# elif sols.shape[0] == 1:
# if self._cbound(sols[0]) < self._cbound(sols[0] + 1) and sols[0] > 0:
# return True, sols[0]
# else:
# return False, "sol was max"
# else:
# return False, "no sol"
# elif sols.shape[0] == 2 and sols[0] > 0 and sols[1] > 1:
# best_sol = self._best_sol(sols)
# elif np.greater(sols, np.zeros(2)).any():
# return self._analyze_solutions_one_weight(np.array([np.max(sols)]))
# else:
# return False, "no solution were found"
#
# if isinstance(best_sol, complex):
# return False, "the sol was complex"
# else:
# return True, best_sol
#
# else:
# for hypothese_index, hypothese in enumerate(
# weighted_hypothesis.transpose()):
# if (
# hypothese_index not in self.chosen_columns_
# or self.twice_the_same) \
# and set(self.chosen_columns_) != {hypothese_index} \
# and self._is_not_too_wrong(hypothese, y):
# w = self._solve_one_weight_min_c(hypothese, y, margin_old, m)
# if w[0] != "break":
# c_bounds.append(self._cbound(w[0]))
# possible_sols.append(w)
# indices.append(hypothese_index)
# else:
# causes.append(w[1])
# if not causes:
# causes = ["no feature was better than random and acceptable"]
# if c_bounds:
# min_c_bound_index = ma.argmin(c_bounds)
# self.c_bounds.append(c_bounds[min_c_bound_index])
# selected_sol = possible_sols[min_c_bound_index]
# self.margins.append(self.margin(selected_sol))
# self.disagreements.append(self.disagreement(selected_sol))
# selected_voter_index = indices[min_c_bound_index]
# return selected_sol, selected_voter_index #selected_sol/(1+selected_sol)
# else:
# return "break", " and ".join(set(causes))
\ No newline at end of file
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment