Page MenuHomec4science
Files F77493624

utils.py
No OneTemporary
Actions

File Metadata

Created
Wed, Aug 14, 19:40

utils.py
View Options

import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import torch
cuda = torch.cuda.is_available()
from sklearn.manifold import TSNE
import pandas as pd
colors = [ 'purple','orange','green','red', 'blue', 'cyan']
mnist_classes = ['P1', 'P2', 'P3', 'P4','P5','P6']
marker= ["d","s","*",">","X","o"]
plt.rcParams['xtick.labelsize']=16
plt.rcParams['ytick.labelsize']=16
plt.rcParams["legend.markerscale"] = 2.1
def extract_embeddings(dataloader, model):
with torch.no_grad():
model.eval()
embeddings = np.zeros((len(dataloader.dataset), 512))
labels = np.zeros(len(dataloader.dataset))
k = 0
for images, target in dataloader:
if cuda:
images = images.cuda()
embeddings[k:k+len(images)] = model(images).data.cpu().numpy()
labels[k:k+len(images)] = target.numpy()
k += len(images)
return embeddings, labels
def TSNEplot(z_run,test_labels,graph_name,perplexity):
output = z_run
#array of latent space, features fed rowise
target = test_labels
#groundtruth variable
print('target shape: ', target.shape)
print('output shape: ', output.shape)
print('perplexity: ',perplexity)
group=target
group = np.ravel(group)
RS=np.random.seed(1974)
tsne = TSNE(n_components=3, random_state=RS, perplexity=perplexity)
tsne_fit = tsne.fit_transform(output)
return tsne,tsne_fit,target
def plot_embeddings(embeddings, targets,name,train_dataset, xlim=None, ylim=None):
plt.figure(figsize=(9,6))
for i in range(len(train_dataset.classes)):
inds = np.where(targets==i)[0]
plt.scatter(embeddings[inds,0], embeddings[inds,1], alpha=1, color=colors[i],marker=marker[i])
if xlim:
plt.xlim(xlim[0], xlim[1])
if ylim:
plt.ylim(ylim[0], ylim[1])
plt.legend(mnist_classes)
graph_title = "Feature space distribution " + str(name)
plt.xlabel ('Dimension 1', labelpad=10,fontsize = 15)
plt.ylabel ('Dimension 2', labelpad=10,fontsize = 15)
plt.savefig(graph_title, bbox_inches='tight',dpi=600)
plt.show()
def Three_embeddings(embeddings, targets,graph_name,ang, xlim=None, ylim=None):
group=targets
df2 = pd.DataFrame(group)
df2.columns = ['Categorical']
df2=df2['Categorical'].replace(0,'P1')
df2 = pd.DataFrame(df2)
df2=df2['Categorical'].replace(1,'P2')
df2 = pd.DataFrame(df2)
df2=df2['Categorical'].replace(2,'P3')
df2 = pd.DataFrame(df2)
df2=df2['Categorical'].replace(3,'P4')
df2 = pd.DataFrame(df2)
df2=df2['Categorical'].replace(4,'P5')
df2 = pd.DataFrame(df2)
df2=df2['Categorical'].replace(5,'P6')
group = pd.DataFrame(df2)
group=group.to_numpy()
group = np.ravel(group)
x1=embeddings[:, 0]
x2=embeddings[:, 1]
x3=embeddings[:, 2]
df = pd.DataFrame(dict(x=x1, y=x2,z=x3, label=group))
groups = df.groupby('label')
uniq = list(set(df['label']))
uniq=np.sort(uniq)
#uniq=["0","1","2","3"]
fig = plt.figure(figsize=(12,6), dpi=100)
fig.set_facecolor('white')
ax = plt.axes(projection='3d')
ax.grid(False)
ax.view_init(azim=ang)#115
marker= ["d","s","*",">","X","o"]
color = [ 'purple','orange','green','red', 'blue', 'cyan']
ax.set_facecolor('white')
ax.w_xaxis.pane.fill = False
ax.w_yaxis.pane.fill = False
ax.w_zaxis.pane.fill = False
ax.xaxis.set_pane_color((1.0, 1.0, 1.0, 0.0))
ax.yaxis.set_pane_color((1.0, 1.0, 1.0, 0.0))
ax.zaxis.set_pane_color((1.0, 1.0, 1.0, 0.0))
# make the grid lines transparent
ax.xaxis._axinfo["grid"]['color'] = (1,1,1,0)
ax.yaxis._axinfo["grid"]['color'] = (1,1,1,0)
ax.zaxis._axinfo["grid"]['color'] = (1,1,1,0)
graph_title = "Feature space distribution"
j=0
for i in uniq:
print(i)
indx = group == i
a=x1[indx]
b=x2[indx]
c=x3[indx]
ax.plot(a, b, c ,color=color[j],label=uniq[j],marker=marker[j],linestyle='',ms=7)
j=j+1
plt.xlabel ('Dimension 1', labelpad=20,fontsize = 15)
plt.ylabel ('Dimension 2', labelpad=20,fontsize = 15)
ax.set_zlabel('Dimension 3',labelpad=20,fontsize = 15)
plt.title(str(graph_title),fontsize = 15)
plt.legend(markerscale=20)
plt.locator_params(nbins=6)
plt.legend(loc='upper left',frameon=False)
plt.savefig(graph_name, bbox_inches='tight',dpi=400)
plt.show()
return ax,fig

Event Timeline

c4science · Help