def wda_pre(data_train_center, data_test_center, y_label, direction, reg=100):
P0 = np.random.randn(data_train_center.shape[1], direction)
P0 /= np.sqrt(np.sum(P0**2, 0, keepdims=True))
Pwda, projwda = wda(data_train_center,
y_label,
direction,
reg,
k=10,
maxiter=30,
P0=P0)
data_wda_fit = projwda(data_train_center)
data_wda_test = projwda(data_test_center)
return data_wda_fit, data_wda_test
def fit(self, fd_train: FDataGrid, y_train: np.array):
"""Fits the classifier using median of each class of the training samples
Arguments :
- fd_train : training data - FDataGrid
- y_train : training labels - ndarray of shape (n_samples,)"""
X_train = np.transpose(fd_train.data_matrix, axes=(0,2,1))
# compute mean of each class and label them
mean_outliers = np.mean(X_train[np.where(y_train==1)[0]], axis=0).transpose()
mean_inliers = np.mean(X_train[np.where(y_train==0)[0]], axis=0).transpose()
train = np.concatenate((mean_outliers, mean_inliers))
train_lab = np.zeros(len(train))
train_lab[0:len(mean_outliers)] = 1
# project sampels
P0 = np.random.randn(train.shape[1], self.p)
P0 /= np.sqrt(np.sum(P0**2, 0, keepdims=True))
self.Pwda, self.projwda = wda(train, train_lab, self.p, self.reg, k=10, maxiter=100, P0=P0)
train_proj = self.projwda(train)
# fit classifier
self.clf.fit(train_proj, train_lab)
#%% plot samples
pl.figure(1)
pl.scatter(xt[:, 0], xt[:, 1], c=ys, marker='+', label='Source samples')
pl.legend(loc=0)
pl.title('Discriminant dimensions')
#%% plot distributions and loss matrix
p = 2
reg = 1
k = 10
maxiter = 100
P, proj = wda(xs, ys, p, reg, k, maxiter=maxiter)
#%% plot samples
xsp = proj(xs)
xtp = proj(xt)
pl.figure(1, (10, 5))
pl.subplot(1, 2, 1)
pl.scatter(xsp[:, 0], xsp[:, 1], c=ys, marker='+', label='Projected samples')
pl.legend(loc=0)
pl.title('Projected training samples')
pl.subplot(1, 2, 2)
pl.scatter(xtp[:, 0], xtp[:, 1], c=ys, marker='+', label='Projected samples')
#%% Compute FDA p = 2 Pfda, projfda = fda(xs, ys, p) ############################################################################## # Compute Wasserstein Discriminant Analysis # ----------------------------------------- #%% Compute WDA p = 2 reg = 1e0 k = 10 maxiter = 100 Pwda, projwda = wda(xs, ys, p, reg, k, maxiter=maxiter) ############################################################################## # Plot 2D projections # ------------------- #%% plot samples xsp = projfda(xs) xtp = projfda(xt) xspw = projwda(xs) xtpw = projwda(xt) pl.figure(2)
#%% Compute FDA p = 2 Pfda, projfda = fda(xs, ys, p) ############################################################################## # Compute Wasserstein Discriminant Analysis # ----------------------------------------- #%% Compute WDA p = 2 reg = 1e0 k = 10 maxiter = 100 Pwda, projwda = wda(xs, ys, p, reg, k, maxiter=maxiter) ############################################################################## # Plot 2D projections # ------------------- #%% plot samples xsp = projfda(xs) xtp = projfda(xt) xspw = projwda(xs) xtpw = projwda(xt) pl.figure(2)
