As a beginner, i was trying to simply compute the dot product of two matrices using theano.
my code is very simple.
import theano
import theano.tensor as T
import numpy as np
from theano import function
def covarience(array):
input_array=T.matrix('input_array')
deviation_matrix = T.matrix('deviation_matrix')
matrix_filled_with_1s=T.matrix('matrix_filled_with_1s')
z = T.dot(input_array, matrix_filled_with_1s)
identity=np.ones((len(array),len(array)))
f=function([array,identity],z)
# print(f)
covarience(np.array([[2,4],[6,8]]))
but the problem is each time i run this code , i get error message like "TypeError: Unknown parameter type: "
Can anyone tell me whats wrong with my code?
You cannot pass numpy array to theano function, theano functions can only be defined by theano.tensor variables. So you can always define computations with interaction of tensor/symbolic variables, and to perform actual computation on values/real data you can use functions, it doesn't make sense to define theano function itself with numpy array.
This should work:
import theano
import theano.tensor as T
import numpy as np
a = T.matrix('a')
b = T.matrix('b')
z = T.dot(a, b)
f = theano.function([a, b], z)
a_d = np.asarray([[2, 4], [6, 8]], dtype=theano.config.floatX)
b_d = np.ones(a_d.shape, dtype=theano.config.floatX)
print(f(a_d, b_d))
Related
I am trying to make sense of the generic_filter1d() function in scipy.ndimage. My understanding is that this function would allow a function being applied to a 1d array by 'extracting' a subset of this array and applying a function that returns a single value (analogous to the generic_filter()?). Is this correct? I have the following simple code:
def test(x,y):
return np.sum(x/y)
and
import numpy as np
from scipy.ndimage import generic_filter1d
dummy_data= np.arange(10)
size = 5
y= np.linspace(0.1, size*0.1, size)
rstl = generic_filter1d(dummy_data, test, filter_size= size, extra_arguments =(y,))
And it fails. It says that I have an extra argument.
I have a dataset which I have to process in such a way that it works with a convolutional neural network of PyTorch (I'm completely new to PyTorch). The data is stored in a dataframe with a column for pictures (28 x 28 ndarrays with int32 entries) and a column with its class labels. The pixels of the images merely adopt values +1 and -1 (since it is simulation data of a classical 2d Ising Model). The dataframe looks like this.
I imported the following (a lot of this is not relevant for now, but I included everything for completeness. "data_loader" is a custom py file.):
import numpy as np
import matplotlib.pyplot as plt
import data_loader
import pandas as pd
import torch
import torchvision.transforms as T
from torchvision.utils import make_grid
from torch.nn import Module
from torch.nn import Conv2d
from torch.nn import Linear
from torch.nn import MaxPool2d
from torch.nn import ReLU
from torch.nn import LogSoftmax
from torch import flatten
from sklearn.metrics import classification_report
import time as time
from torch.utils.data import DataLoader, Dataset
Then, I want to get this in the correct shape in order to make it useful for PyTorch. I do this by defining the following class
class MetropolisDataset(Dataset):
def __init__(self, data_frame, transform=None):
self.data_frame = data_frame
self.transform = transform
def __len__(self):
return len(self.data_frame)
def __getitem__(self,idx):
if torch.is_tensor(idx):
idx = idx.tolist()
label = self.data_frame['label'].iloc[idx]
image = self.data_frame['image'].iloc[idx]
image = np.array(image)
if self.transform:
image = self.transform(image)
return (image, label)
I call instances of this class as:
train_set = MetropolisDataset(data_frame = df_train,
transform = T.Compose([
T.ToPILImage(),
T.ToTensor()]))
validation_set = MetropolisDataset(data_frame = df_validation,
transform = T.Compose([
T.ToPILImage(),
T.ToTensor()]))
test_set = MetropolisDataset(data_frame = df_test,
transform = T.Compose([
T.ToPILImage(),
T.ToTensor()]))
The problem does not yet arise here, because I am able to read out and show images from these instances of the above defined class.
Then, as far as I found out, it is necessary to let this go through the DataLoader of PyTorch, which I do as follows:
batch_size = 64
train_dl = DataLoader(train_set, batch_size, shuffle=True, num_workers=3, pin_memory=True)
validation_dl = DataLoader(validation_set, batch_size, shuffle=True, num_workers=3, pin_memory=True)
test_dl = DataLoader(test_set, batch_size, shuffle=True, num_workers=3, pin_memory=True)
However, if I want to use these instances of the DataLoader, simply nothing happens. I neither get an error, nor the computation seems to get anywhere. I tried to run a CNN but it does not seem to compute anything. Something else I tried was to show some sample images with the code provided by this article, but the same issue occurs. The sample code is:
def show_images(images, nmax=10):
fig, ax = plt.subplots(figsize=(8, 8))
ax.set_xticks([]); ax.set_yticks([])
ax.imshow(make_grid((images.detach()[:nmax]), nrow=8).permute(1, 2, 0))
def show_batch(dl, nmax=64):
for images in dl:
show_images(images, nmax)
break
show_batch(test_dl)
It seems that there is some error in the implementation of my MetropolisDataset class or with the DataLoader itself. How could this problem be solved?
As mentioned in the comments, the problem was partly solved by setting num_workers to zero since I was working in a Jupyter notebook, as answered here. However, this left open one further problem that I got errors when I wanted to apply the DataLoader to run a CNN. The issue was then that my data did consist of int32 numbers instead of float32. I do not include further codes, because this was related directly to my data - however, the issue was (as very often) merely a wrong datatype.
I got the following error:
<lambdifygenerated-1>:2: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.return numpy.array((A1exp(-1/2(x - xc1)**2/sigma1**2), 0, 0))
Here I have just one model but this code is written for model combination in fitting by the lmfit Please kindly let me know about it.
import matplotlib.pyplot as plt
import numpy as np
import sympy
from sympy.parsing import sympy_parser
import lmfit
gauss_peak1 = sympy_parser.parse_expr('A1*exp(-(x-xc1)**2/(2*sigma1**2))')
gauss_peak2 = 0
exp_back = 0
model_list = sympy.Array((gauss_peak1, gauss_peak2, exp_back))
model = sum(model_list)
print(model)
model_list_func = sympy.lambdify(list(model_list.free_symbols), model_list)
model_func = sympy.lambdify(list(model.free_symbols), model)
np.random.seed(1)
x = np.linspace(0, 10, 40)
param_values = dict(x=x, A1=2, sigma1=1, xc1=2)
y = model_func(**param_values)
yi = model_list_func(**param_values)
yn = y + np.random.randn(y.size)*0.4
plt.plot(x, yn, 'o')
plt.plot(x, y)
lm_mod = lmfit.Model(model_func, independent_vars=('x'))
res = lm_mod.fit(data=yn, **param_values)
res.plot_fit()
plt.plot(x, y, label='true')
plt.legend()
plt.show()
lmfit.Model takes a model function that is a Python function. It parses the function arguments and expects those to be the Parameters for the model.
I don't think using sympy-created functions will do that. Do you need to use sympy here? I don't see why. The usage here seems designed to make the code more complex than it needs to be. It seems you want to make a model with a Gaussian-like peak, and a constant(?) background. If so, why not do
from lmfit.Models import GaussianModel, ConstantModel
model = GaussianModel(prefix='p1_') + ConstantModel()
params = model.make_params(p1_amplitude=2, p1_center=2, p1_sigma=1, c=0)
That just seems way easier to me, and it is very easy to add a second Gaussian peak to that model.
But even if you have your own preferred mathematical expression, don't use that as a sympy string, use it as Python code:
def myfunction(x, A1, xc1, sigma1):
return A1*exp(-(x-xc1)**2/(2*sigma1**2))
and then
from lmfit import Model
mymodel = Model(myfunction)
params = mymodel.guess(A1=2, xc1=2, sigma1=1)
In short: sympy is an amazing tool, but lmfit does not use it.
I am trying to create a simple image classification tool.
I would like the code below to work with classifying images. It works fine when it is a non image NumPy array.
#https://e2eml.school/images_to_numbers.html
import numpy as np
from sklearn.utils import Bunch
from PIL import Image
monkey = [1]
dog = [2]
example_animals = Bunch(data = np.array([monkey,dog]),target = np.array(['monkey','dog']))
from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=2) #with KMeans you get to pre specify the number of Clusters
KModel = kmeans.fit(example_animals.data) #fit a model using the training data , in this case original example animal data passed through
import pandas as pd
crosstab = pd.crosstab(example_animals.target,KModel.labels_)
print(crosstab)
I have looked into how to make an image into a NumPy array at https://e2eml.school/images_to_numbers.html
The code below where I have converted images to NumPy array doesn't work.
When run it gets the following error
** 'setting an array element with a sequence'**
#https://e2eml.school/images_to_numbers.html
import numpy as np
from sklearn.utils import Bunch
from PIL import Image
monkey = np.asarray(Image.open("monkey.jpg"))
dog = np.asarray(Image.open("dog.jpeg"))
example_animals = Bunch(data = np.array([monkey,dog]),target = np.array(['monkey','dog']))
from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=2) #with KMeans you get to pre specify the number of Clusters
KModel = kmeans.fit(example_animals.data) #fit a model using the training data , in this case original example animal data passed through
import pandas as pd
crosstab = pd.crosstab(example_animals.target,KModel.labels_)
print(crosstab)
I would appreciate any insight how I fix the error 'setting an array element with a sequence' so that the images will be compatible with the sklearn processing.
You need to be sure that your images "monkey.jpg" and "dog.jpeg" have the same number of pixels. Otherwise, you will have to resize the images to have the same size. Moreover, the data of your Bunch object need to be of shape (n_samples, n_features) (you can check the documentation https://scikit-learn.org/stable/modules/generated/sklearn.cluster.KMeans.html#sklearn.cluster.KMeans.fit)
You need to be aware that you use an unserpervised learning model (Kmeans). So the output of the model is not directly "monkey" or "dog".
I found the solution to error setting an array element with a sequence
Kmeans requires the data arrays for comparison need to be the same size.
This means if importing pictures, the pictures need to be resized, converted into a numpy array (a format that is compatible with Kmeans) and finally made into a 1 dimensional array.
#https://e2eml.school/images_to_numbers.html
#https://machinelearningmastery.com/how-to-load-and-manipulate-images-for-deep-learning-in-python-with-pil-pillow/
import numpy as np
from matplotlib import pyplot as plt
from sklearn.utils import Bunch
from PIL import Image
from sklearn.cluster import KMeans
import pandas as pd
monkey = Image.open("monkey.jpg")
dog = Image.open("dog.jpeg")
#resize pictures
monkey1 = monkey.resize((180,220))
dog1 = dog.resize((180,220))
#make pictures into numpy array
monkey2 = np.asarray(monkey1)
dog2 = np.asarray(dog1)
#https://www.quora.com/How-do-I-convert-image-data-from-2D-array-to-1D-using-python
#make numpy array into 1 dimensional array
monkey3 = monkey2.reshape(-1)
dog3 = dog2.reshape(-1)
example_animals = Bunch(data = np.array([monkey3,dog3]),target = np.array(['monkey','dog']))
kmeans = KMeans(n_clusters=2) #with KMeans you get to pre specify the number of Clusters
KModel = kmeans.fit(example_animals.data) #fit a model using the training data , in this case original example food data passed through
crosstab = pd.crosstab(example_animals.target,KModel.labels_)
print(crosstab)
You may already know, that in matplotlib 1.2.0 there is a new experimental feature, that figures are pickable (they can be saved with pickle module).
However, it doesn't work when one uses logscale, eg.
import matplotlib.pyplot as plt
import numpy as np
import pickle
ax = plt.subplot(111)
x = np.linspace(0, 10)
y = np.exp(x)
plt.plot(x, y)
ax.set_yscale('log')
pickle.dump(ax, file('myplot.pickle', 'w'))
results in:
PicklingError: Can't pickle <class 'matplotlib.scale.Log10Transform'>: attribute lookup matplotlib.scale.Log10Transform failed
Anybody knows any solution/workaround to this?
I've opened this as a bug report on matplotlib's github issue tracker. Its a fairly easy fix to implement on the matplotlib repository side (simply don't nest the Log10Transform class inside the LogScale class), but that doesn't really help you in being able to use this with mpl 1.2.0...
There is a solution to getting this to work for you in 1.2.0, but I warn you - its not pretty!
Based on my answer to a pickling question it is possible to pickle nested classes (as Log10Transform is). All we need to do is to tell Log10Transform how to "reduce" itself:
import matplotlib.scale
class _NestedClassGetter(object):
"""
When called with the containing class as the first argument,
the name of the nested class as the second argument,
and the state of the object as the third argument,
returns an instance of the nested class.
"""
def __call__(self, containing_class, class_name, state):
nested_class = getattr(containing_class, class_name)
# return an instance of a nested_class. Some more intelligence could be
# applied for class construction if necessary.
c = nested_class.__new__(nested_class)
c.__setstate__(state)
return c
def _reduce(self):
# return a class which can return this class when called with the
# appropriate tuple of arguments
cls_name = matplotlib.scale.LogScale.Log10Transform.__name__
call_args = (matplotlib.scale.LogScale, cls_name, self.__getstate__())
return (_NestedClassGetter(), call_args)
matplotlib.scale.LogScale.Log10Transform.__reduce__ = _reduce
You might also decide to do this for other Log based transforms/classes, but for your example, you can now pickle (and successfully unpickle) your example figure:
import matplotlib.pyplot as plt
import numpy as np
import pickle
ax = plt.subplot(111)
x = np.linspace(0, 10)
y = np.exp(x)
plt.plot(x, y)
ax.set_yscale('log')
pickle.dump(ax, file('myplot.pickle', 'w'))
plt.savefig('pickle_log.pre.png')
plt.close()
pickle.load(file('myplot.pickle', 'r'))
plt.savefig('pickle_log.post.png')
I'm going to get on and fix this for mpl 1.3.x so that this nasty workaround isn't needed in the future :-) .
HTH,