def smoothed_saliency(ind, colors, probs):
lab = rgb2lab(colors[None].astype(np.uint8)).squeeze()
#lab_dist = np.square(lab[...,None] - lab.T).sum(1)
lab_dist = squareform(pdist(lab, 'sqeuclidean'))
s = (lab_dist * probs).sum(1)
s = (s - s.min()) / (s.max() - s.min())
m = lab.shape[0] // 4
dist, nn = NearestNeighbors(m).fit(lab).kneighbors()
T = dist.sum(1)
sp = ((T[:, None] - dist) * s[nn]).sum(1) / ((m - 1) * T)
return sp
def load_data():
"""
Function to load the data
:param size: total number of points to get
:param num_lm: number of points which will be landmarks
:return: the batch loader, landmark points, labels, batched data without landmark points,
data organized in graphs, neighborhood graph for the landmarks, original data, original labels,
neighborhood graphs for non-landmarks
"""
global batch_size, num_batches
# import data
data, labels = original_clean()
test_data = data[:test_size, :]
test_labels = labels[:test_size]
data = data[test_size:, :]
# make landmarks with points with most neighbors
N = NearestNeighbors(
n_neighbors=k_start).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
num_connections = N.sum(
axis=0).argsort()[::-1] # see how many neighbors each point has
top_landmarks_idxs = num_connections[:num_lm] # sort in descending order
land_marks = data[top_landmarks_idxs, :] # pick the top ones
data = np.delete(data, top_landmarks_idxs, axis=0) # delete the landmarks
# find the nearest landmarks for the landmarks
landmark_neighbors = NearestNeighbors(n_neighbors=k_lm).fit(
land_marks).kneighbors_graph(land_marks).todense()
# break data into batches, create empty holders
batch_loader = np.zeros((num_batches, batch_size + num_lm, n))
batch_graph = np.zeros(
(num_batches, batch_size + num_lm, batch_size + num_lm))
# create the full neighborhood graph for each batch
for i in range(num_batches):
holder = data[batch_size * i:batch_size * (i + 1)]
# find the nearest landmarks for the rest of the points
holder_graph = NearestNeighbors(n_neighbors=k_other).fit(
land_marks).kneighbors_graph(holder).todense()
for j in range(batch_size): # copy over the holder graph
for l in range(num_lm):
if holder_graph[j, l] == 1:
batch_graph[i, j, l + batch_size] = 1
batch_graph[i, l + batch_size, j] = 1
for j in range(num_lm): # copy over landmark neighbors
for l in range(j, num_lm):
if landmark_neighbors[j, l] == 1 and j != l:
batch_graph[i, j + batch_size, l + batch_size] = 1
batch_graph[i, l + batch_size, j + batch_size] = 1
holder = np.concatenate((holder, land_marks))
batch_loader[i] = holder
batch_size += num_lm # adjust the batch size
return batch_loader, data, batch_graph, landmark_neighbors, test_data, test_labels, land_marks
def load_data(size, num_lm):
global divisor, m, n, batch_size, set_random
# import data
data = pd.read_csv('pol_data.csv', delimiter=',').values[:, :-1]
labels = np.asarray([0,1,1,1,1,1,0,0,1,0,0,1,1,0,0,1,1,0,1,1,1,1,1,1,1,1,1,0,1,0,1,1,0,1,0,0,0,0,1,1,1,1,1,1,1,1,1,1,0,1,0,1,0,1,0,0,0,0,0,1,0,1,1,1,0,0,0,1,1,1,0,1,0,1,1,1,1,1,0,1,1,0,0,0,1,0,0,1,0,1,1,1,1,1,1,1,1,1,0,1,1,1,1,1,1,0,0,1,1,1,0,1,0,1,1,1,0,1,0,0,0,0,0,1,0,0,1,1,1,1,1,0,0,0,0,0,1,1,1,0,0,0,1,1,1,0,1,0,1,0,1,1,1,0,0,0,1,0,1,1,1,1,0,1,1,0,0,1,1,1,0,1,1,1,1,0,1,1,1,1,1,1,1,1,1,1,1,0,1,0,0,1,1,1,0,1,0,1,1,1,1,1,1,0,1,0,0,1,1,1,0,1,1,0,1,1,0,1,1,1,1,1,0,0,0,1,1,0,0,0,0,1,0,1,0,1,1,1,0,0,1,0,1,1,1,1,0,0,1,0,0,1,0,1,1,0,0,1,1,1,1,1,1,1,1,0,0,1,1,1,1,1,0,0,1,0,0,0,0,1,0,0,0,1,1,1,1,1,1,1,1,1,1,1,0,0,1,1,1,0,1,0,0,0,0,0,1,0,1,0,1,1,0,0,0,1,1,1,1,1,1,1,1,0,1,1,0,1,1,0,1,1,1,1,0,1,1,1,0,0,1,1,0,1,0,0,0,1,0,1,0,1,0,1,0,0,0,1,0,1,1,1,0,0,1,1,1,1,0,1,1,1,1,0,1,1,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,0,0,1,1,1,1,1,0,0,0,0,0,0,0,1,1,1,0,0,1,0,0,1,1,0,1,1,1,0,1,1,1,1,1,1,0,1,1,0,1,0,0])
test_data = data[size:, :]
test_labels = labels[size:]
print(size)
data = data[:size, :]
labels = labels[:size]
m = np.size(data, 0)
n = np.size(data, 1)
# data = normalize(data)
# make landmarks, select x random points in the data set
land_marks = np.empty((num_lm, n))
top_landmarks_idxs = []
if set_random:
for i in range(num_lm):
index = random.randint(0, m - i)
land_marks[i] = data[index]
data = np.delete(data, index, axis=0)
labels = np.delete(labels, index, axis=0)
else:
N = NearestNeighbors(n_neighbors=k_start).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
num_connections = N.sum(axis=0).argsort()[::-1]
top_landmarks_idxs = num_connections[:num_lm]
land_marks = data[top_landmarks_idxs, :]
data = np.delete(data, top_landmarks_idxs, axis=0)
landmark_neighbors = NearestNeighbors(n_neighbors=k_lm).fit(land_marks).kneighbors_graph(land_marks).todense()
divisor = int(size / batch_size) # review this line
batch_loader = np.zeros((divisor, batch_size + num_lm, n))
batch_graph = np.zeros((divisor, batch_size + num_lm, batch_size + num_lm))
for i in range(divisor):
holder = data[batch_size * i: batch_size * (i + 1)]
holder_graph = NearestNeighbors(n_neighbors=k_other).fit(land_marks).kneighbors_graph(holder).todense()
for j in range(batch_size): # copy over the holder graph
for l in range(num_lm):
if holder_graph[j, l] == 1:
batch_graph[i, j, l + batch_size] = 1
batch_graph[i, l + batch_size, j] = 1
for j in range(num_lm): # copy over landmark neighbors
for l in range(j, num_lm):
if landmark_neighbors[j, l] == 1 and j != l:
batch_graph[i, j + batch_size, l + batch_size] = 1
batch_graph[i, l + batch_size, j + batch_size] = 1
holder = np.concatenate((holder, land_marks))
batch_loader[i] = holder
batch_size += num_lm
return batch_loader, land_marks, labels, data, batch_graph, top_landmarks_idxs, test_data, test_labels, landmark_neighbors
def load_data(size, num_lm):
global divisor, m, n, batch_size, set_random
# import data
data, labels = sklearn.datasets.make_swiss_roll(size)
data, labels = shuffle(data, labels)
m = np.size(data, 0)
n = np.size(data, 1)
data = normalize(data)
saveLabels = labels
saveData = data
# make landmarks, select x random points in the data set
land_marks = np.empty((num_lm, n))
top_landmarks_idxs = []
if set_random:
for i in range(num_lm):
index = random.randint(0, m - i)
land_marks[i] = data[index]
data = np.delete(data, index, axis=0)
labels = np.delete(labels, index, axis=0)
else:
N = NearestNeighbors(
n_neighbors=k_start).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
num_connections = N.sum(axis=0).argsort()[::-1]
top_landmarks_idxs = num_connections[:num_lm]
land_marks = data[top_landmarks_idxs, :]
data = np.delete(data, top_landmarks_idxs, axis=0)
landmark_neighbors = NearestNeighbors(n_neighbors=k_lm).fit(
land_marks).kneighbors_graph(land_marks).todense()
divisor = int(size / batch_size)
batch_loader = np.zeros((divisor, batch_size + num_lm, n))
batch_graph = np.zeros((divisor, batch_size + num_lm, batch_size + num_lm))
for i in range(divisor):
holder = data[batch_size * i:batch_size * (i + 1)]
holder_graph = NearestNeighbors(n_neighbors=k_other).fit(
land_marks).kneighbors_graph(holder).todense()
for j in range(batch_size): # copy over the holder graph
for l in range(num_lm):
if holder_graph[j, l] == 1:
batch_graph[i, j, l + batch_size] = 1
batch_graph[i, l + batch_size, j] = 1
for j in range(num_lm): # copy over landmark neighbors
for l in range(j, num_lm):
if landmark_neighbors[j, l] == 1 and j != l:
batch_graph[i, j + batch_size, l + batch_size] = 1
batch_graph[i, l + batch_size, j + batch_size] = 1
holder = np.concatenate((holder, land_marks))
batch_loader[i] = holder
batch_size += num_lm
return batch_loader, land_marks, labels, data, batch_graph, top_landmarks_idxs, saveData, saveLabels
def load_data():
global batch_size, divisor
data, labels = original_clean()
test_data = data[300:, :]
test_labels = labels[300:]
data = data[:300, :]
labels = labels[:300]
N = NearestNeighbors(
n_neighbors=k_start).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
num_connections = N.sum(axis=0).argsort()[::-1]
top_landmarks_idxs = num_connections[:num_lm]
land_marks = data[top_landmarks_idxs, :]
data = np.delete(data, top_landmarks_idxs, axis=0)
landmark_neighbors = NearestNeighbors(n_neighbors=k_lm).fit(
land_marks).kneighbors_graph(land_marks).todense()
divisor = int(size / batch_size)
batch_loader = np.zeros((divisor, batch_size + num_lm, n))
batch_graph = np.zeros((divisor, batch_size + num_lm, batch_size + num_lm))
for i in range(divisor):
holder = data[batch_size * i:batch_size * (i + 1)]
holder_graph = NearestNeighbors(n_neighbors=k_other).fit(
land_marks).kneighbors_graph(holder).todense()
for j in range(batch_size): # copy over the holder graph
for l in range(num_lm):
if holder_graph[j, l] == 1:
batch_graph[i, j, l + batch_size] = 1
batch_graph[i, l + batch_size, j] = 1
for j in range(num_lm): # copy over landmark neighbors
for l in range(j, num_lm):
if landmark_neighbors[j, l] == 1 and j != l:
batch_graph[i, j + batch_size, l + batch_size] = 1
batch_graph[i, l + batch_size, j + batch_size] = 1
holder = np.concatenate((holder, land_marks))
batch_loader[i] = holder
batch_size += num_lm
return batch_loader, land_marks, labels, data, batch_graph, top_landmarks_idxs, test_data, test_labels, landmark_neighbors
def fit(self, data, k):
"""
The method to fit an MVU model to the data.
:param data: The data to which the model will be fitted.
:param k: The number of neighbors to fix.
:return: Embedded Gramian: The Gramian matrix of the embedded data.
"""
# Number of data points in the set
n = data.shape[0]
# Set the seed
np.random.seed(self.seed)
# Calculate the nearest neighbors of each data point and build a graph
N = NearestNeighbors(n_neighbors=k).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
# Sort the neighbor graph to find the points with the most connections
num_connections = N.sum(axis=0).argsort()[::-1]
# Separate the most popular points
top_landmarks_idxs = num_connections[:self.landmarks]
top_landmarks = data[top_landmarks_idxs, :]
# Compute the nearest neighbors for all of the landmarks so they are all connected
L = NearestNeighbors(n_neighbors=3).fit(top_landmarks).kneighbors_graph(top_landmarks).todense()
L = np.array(L)
# The data without the landmarks
new_data_idxs = [x for x in list(range(n)) if x not in top_landmarks_idxs]
new_data = np.delete(data, top_landmarks_idxs, axis=0)
# Construct a neighborhood graph where each point finds its closest landmark
l = NearestNeighbors(n_neighbors=3).fit(top_landmarks).kneighbors_graph(new_data).todense()
l = np.array(l)
# Reset N to all 0's
N = np.zeros((n, n))
# Add all of the intra-landmark connections to the neighborhood graph
for i in range(self.landmarks):
for j in range(self.landmarks):
if L[i, j] == 1.:
N[top_landmarks_idxs[i], top_landmarks_idxs[j]] = 1.
# Add all of the inter-landmark connections to the neighborhood graph
for i in range(n - self.landmarks):
for j in range(self.landmarks):
if l[i, j] == 1.:
N[new_data_idxs[i], top_landmarks_idxs[j]] = 1.
# Save the neighborhood graph to be accessed latter
self.neighborhood_graph = N
# To check for disconnected regions in the neighbor graph
lap = laplacian(N, normed=True)
eigvals, _ = np.linalg.eig(lap)
for e in eigvals:
if e == 0. and self.solver_iters is None:
raise DisconnectError("DISCONNECTED REGIONS IN NEIGHBORHOOD GRAPH. "
"PLEASE SPECIFY MAX ITERATIONS FOR THE SOLVER")
# Declare some CVXPy variables
# Gramian of the original data
P = cp.Constant(data.dot(data.T))
# The projection of the Gramian
Q = cp.Variable((n, n), PSD=True)
# Initialized to zeros
Q.value = np.zeros((n, n))
# A shorter way to call a vector of 1's
ONES = cp.Constant(np.ones((n, 1)))
# A variable to keep the notation consistent with the Berkley lecture
T = cp.Constant(n)
# Declare placeholders to get rid of annoying warnings
objective = None
constraints = []
# Wikipedia Solution
if self.equation == "wikipedia":
objective = cp.Maximize(cp.trace(Q))
constraints = [Q >> 0, cp.sum(Q, axis=1) == 0]
for i in range(n):
for j in range(n):
if N[i, j] == 1:
constraints.append((P[i, i] + P[j, j] - P[i, j] - P[j, i]) -
(Q[i, i] + Q[j, j] - Q[i, j] - Q[j, i]) == 0)
# UC Berkley Solution
if self.equation == "berkley":
objective = cp.Maximize(cp.multiply((1 / T), cp.trace(Q)) -
cp.multiply((1 / (T * T)), cp.trace(cp.matmul(cp.matmul(Q, ONES), ONES.T))))
constraints = [Q >> 0, cp.sum(Q, axis=1) == 0]
for i in range(n):
for j in range(n):
if N[i, j] == 1.:
constraints.append(Q[i, i] - 2 * Q[i, j] + Q[j, j] -
(P[i, i] - 2 * P[i, j] + P[j, j]) == 0)
# Solve the problem with the SCS Solver
problem = cp.Problem(objective, constraints)
# FIXME The solvertol syntax is unique to SCS
problem.solve(solver=self.solver,
eps=self.solver_tol,
max_iters=self.solver_iters,
warm_start=self.warm_start)
return Q.value
def load_data(size, num_lm):
"""
Function to load the data
:param size: total number of points to get
:param num_lm: number of points which will be landmarks
:return: the batch loader, landmark points, labels, batched data without landmark points,
data organized in graphs, neighborhood graph for the landmarks, original data, original labels,
neighborhood graphs for non-landmarks
"""
global divisor, m, n, batch_size, set_random
# import data
data, labels = sklearn.datasets.make_swiss_roll(size)
# data, labels = shuffle(data, labels)
m = np.size(data, 0)
n = np.size(data, 1)
data = normalize(data)
save_labels = labels
save_data = data
# make landmarks, select x random points in the data set
land_marks = np.empty((num_lm, n))
top_landmarks_idxs = []
if set_random:
# select x random points in the data set
for i in range(num_lm):
index = random.randint(0, m - i)
land_marks[i] = data[index]
data = np.delete(data, index, axis=0)
labels = np.delete(labels, index, axis=0)
else:
# pick the points with the most neighbors
N = NearestNeighbors(
n_neighbors=k_start).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
num_connections = N.sum(axis=0).argsort()[::-1]
top_landmarks_idxs = num_connections[:num_lm]
land_marks = data[top_landmarks_idxs, :]
data = np.delete(data, top_landmarks_idxs, axis=0)
# find the nearest landmarks for the landmarks
landmark_neighbors = NearestNeighbors(n_neighbors=k_lm).fit(
land_marks).kneighbors_graph(land_marks).todense()
# break data into batches
divisor = int(size / batch_size)
batch_loader = np.zeros((divisor, batch_size + num_lm, n))
batch_graph = np.zeros((divisor, batch_size + num_lm, batch_size + num_lm))
for i in range(divisor):
holder = data[batch_size * i:batch_size * (i + 1)]
# find the nearest landmarks for the rest of the points
holder_graph = NearestNeighbors(n_neighbors=k_other).fit(
land_marks).kneighbors_graph(holder).todense()
for j in range(batch_size): # copy over the holder graph
for l in range(num_lm):
if holder_graph[j, l] == 1:
batch_graph[i, j, l + batch_size] = 1
batch_graph[i, l + batch_size, j] = 1
for j in range(num_lm): # copy over landmark neighbors
for l in range(j, num_lm):
if landmark_neighbors[j, l] == 1 and j != l:
batch_graph[i, j + batch_size, l + batch_size] = 1
batch_graph[i, l + batch_size, j + batch_size] = 1
holder = np.concatenate((holder, land_marks))
batch_loader[i] = holder
batch_size += num_lm # adjust the batch size
return batch_loader, land_marks, labels, data, batch_graph, top_landmarks_idxs, save_data, save_labels, landmark_neighbors
def load_data(size, num_lm):
global divisor, m, n, batch_size, set_random
# import data
data, labels = sklearn.datasets.make_swiss_roll(size)
data, labels = shuffle(data, labels)
m = np.size(data, 0)
n = np.size(data, 1)
data = normalize(data)
saveLabels = labels
saveData = data
# make landmarks, select x random points in the data set
land_marks = np.empty((num_lm, n))
top_landmarks_idxs = []
if set_random:
for i in range(num_lm):
index = random.randint(0, m - i)
land_marks[i] = data[index]
data = np.delete(data, index, axis=0)
labels = np.delete(labels, index, axis=0)
else:
N = NearestNeighbors(
n_neighbors=k_start).fit(data).kneighbors_graph(data).todense()
N = np.array(N)
num_connections = N.sum(axis=0).argsort()[::-1]
top_landmarks_idxs = num_connections[:num_lm]
land_marks = data[top_landmarks_idxs, :]
data = np.delete(data, top_landmarks_idxs, axis=0)
labels = np.delete(labels, top_landmarks_idxs, axis=0)
landmark_neighbors = NearestNeighbors(n_neighbors=k_lm).fit(
land_marks).kneighbors_graph(land_marks).todense()
divisor = int(size / batch_size)
batch_loader = np.zeros((divisor, batch_size + num_lm, n))
batch_graph = np.zeros((divisor, batch_size + num_lm, batch_size + num_lm))
for i in range(divisor):
holder = data[batch_size * i:batch_size * (i + 1)]
holderLabel = labels[batch_size * i:batch_size *
(i + 1)] #For ShortCircuit Detection
holder_graph = NearestNeighbors(n_neighbors=k_other).fit(
land_marks).kneighbors_graph(holder).todense()
for j in range(batch_size): # copy over the holder graph
for l in range(num_lm):
if holder_graph[j, l] == 1:
batch_graph[i, j, l + batch_size] = 1
batch_graph[i, l + batch_size, j] = 1
for j in range(num_lm): # copy over landmark neighbors
for l in range(j, num_lm):
if landmark_neighbors[j, l] == 1 and j != l:
batch_graph[i, j + batch_size, l + batch_size] = 1
batch_graph[i, l + batch_size, j + batch_size] = 1
holder = np.concatenate((holder, land_marks))
fig = plt.figure() #For Short Circuit Detection
ax = plt.axes(projection='3d')
landmarkplaceholder = []
for j in range(num_lm):
landmarkplaceholder.append(batch_size + j)
ax.scatter(holder[range(batch_size), 0],
holder[range(batch_size), 1],
holder[range(batch_size), 2],
c=holderLabel)
ax.scatter(holder[landmarkplaceholder, 0],
holder[landmarkplaceholder, 1],
holder[landmarkplaceholder, 2],
c="Red",
marker="^",
alpha=1)
for o in range(batch_size):
for j in range(batch_size, batch_size + num_lm):
if batch_graph[i][o][j] > 0:
ax.plot([holder[o][0], holder[j][0]],
[holder[o][1], holder[j][1]],
[holder[o][2], holder[j][2]],
c='Red',
alpha=0.5)
plt.show() #End of Short Circuit Detection Code
batch_loader[i] = holder
batch_size += num_lm
return batch_loader, land_marks, labels, data, batch_graph, top_landmarks_idxs, saveData, saveLabels
def over_sampling(self):
if self.k + 1 > self.n_train_less:
print(
'Expected n_neighbors <= n_samples, but n_samples = {}, n_neighbors = {}, '
'has changed the n_neighbors to {}'.format(
self.n_train_less, self.k + 1, self.n_train_less))
self.k = self.n_train_less - 1
data_less_filter = []
num_maj_filter = []
length_less = len(self.train_less)
num_maj = number_maj(self.train[:, 1:], self.train_less[:, 1:],
self.tp_less, self.train[:, 0])
for m in range(len(num_maj)):
if num_maj[m] < self.k:
data_less_filter.append(self.train_less[m])
num_maj_filter.append(num_maj[m])
self.train_less = np.array(data_less_filter)
distance_more, nn_array_more = NearestNeighbors(
n_neighbors=self.k + 1).fit(self.train_more[:, 1:]).kneighbors(
self.train_less[:, 1:], return_distance=True)
distance_less, nn_array = NearestNeighbors(n_neighbors=self.k + 1).fit(
self.train_less[:, 1:]).kneighbors(self.train_less[:, 1:],
return_distance=True)
distance_less = distance_less.sum(axis=1)
distance_more = distance_more.sum(axis=1)
distance = distance_less / distance_more
# print(distance)
density = 1 / distance # calculate density
density = list(
map(lambda x: min(100, x),
density)) # Control the maximum density range at 100
# The density is sorted below, and the minority samples are also sorted in order of density.
density_sorted = sorted(range(len(density)),
key=lambda a: density[a],
reverse=True) # sorted
data_resorted = []
density_sorted_data = []
num_sorted = []
for i in range(len(self.train_less)):
data_resorted.append(self.train_less[density_sorted[i]])
density_sorted_data.append(density[density_sorted[i]])
num_sorted.append(num_maj_filter[density_sorted[i]])
density = np.array(density_sorted_data)
cluster_big_density = []
cluster_small_density = []
cluster_big_data = []
cluster_small_data = []
cluster_big_num = []
cluster_small_num = []
cluster = k_means(X=density.reshape((len(density), 1)), n_clusters=2)
for i in range(cluster[1].shape[0]):
if cluster[1][i] != cluster[1][i + 1]: # Partition cluster
cluster_big_density = density[:i + 1]
cluster_big_data = np.array(data_resorted)[:i + 1, :]
cluster_big_num = num_sorted[:i + 1]
cluster_small_density = density[i + 1:]
cluster_small_data = np.array(data_resorted)[i + 1:, :]
cluster_small_num = num_sorted[i + 1:]
break
# If there is only one point in a cluster, do not divide the cluster
if len(cluster_big_data) < 2 or len(cluster_small_data) < 2:
cluster_big_data = np.array(data_resorted)
cluster_big_density = density
cluster_big_num = num_sorted
flag = 1 # if flag==1 only run big cluster once
else:
flag = 2
sum_0 = 0
sum_1 = 0
# Calculate weight
for p in range(len(cluster_big_num)):
sum_0 += (5 - cluster_big_num[p]) / self.k + 1
for p in range(len(cluster_small_num)):
sum_0 += (5 - cluster_small_num[p]) / self.k + 1
ratio = [] # save the every cluster's totol weight
ratio.append(sum_0)
ratio.append(sum_1)
wight = [5 / 6, 4 / 6, 3 / 6, 2 / 6, 1 / 6]
kk = self.k
diff = len(self.train_more
) - length_less # the number of samples need to synthesize
totol_less = len(self.train_less)
for i in range(flag):
if i == 0: # big cluster
density = cluster_big_density
self.n_train_less = len(cluster_big_data)
self.train_less = cluster_big_data
maj_num_ab = cluster_big_num
else: # small cluster
density = cluster_small_density
self.n_train_less = len(cluster_small_data)
self.train_less = cluster_small_data
maj_num_ab = cluster_small_num
self.k = min(
len(self.train_less) - 1,
kk) # if len(self.train_less)<k,set k =len(self.train_less)
# The number of sample points that need to be inserted at each point
if flag == 1:
number_synthetic = int(
len(self.train_more) / self.IR - len(self.train_less))
else:
if i == 0:
number_synthetic = int(
(len(self.train_less) / totol_less) * diff)
len_big = number_synthetic
else:
number_synthetic = diff - len_big
# Calculate how many points should be inserted for each sample
N = list(
map(lambda x: int((x / ratio[i]) * number_synthetic), wight))
self.reminder = number_synthetic - sum(N)
self.num = 0
neighbors = NearestNeighbors(n_neighbors=self.k + 1).fit(
self.train_less[:, 1:])
nn_array = neighbors.kneighbors(self.train_less[:, 1:],
return_distance=False)
self.synthetic = np.zeros((number_synthetic, self.n_attrs - 1))
for p in range(self.train_less.shape[0]):
self._populate(p, nn_array[p][1:], number_synthetic, N,
maj_num_ab)
label_synthetic = np.array([self.tp_less] *
number_synthetic).reshape(
(number_synthetic, 1))
np.random.seed(self.random_state)
synthetic_dl = self.synthetic
synthetic_dl = np.hstack(
(label_synthetic, synthetic_dl)) # class column
data_res = synthetic_dl
if i == 0:
return_data = np.vstack((copy.deepcopy(self.train), data_res))
if flag == 1:
return return_data
self.new_index = 0
else:
return_data = np.vstack((copy.deepcopy(return_data), data_res))
return return_data