I want to use SHAP summary plot for multiclass classification problem using Deep Explainer. I have 3 classes and for shap_values I got a list of 3 arrays each having (1000,1,24) size. Each array representing a class, I am getting the summary plot for individual class
import shap
background = train_x[np.random.choice(train_x.shape[0], 1000, replace=False)]
explainer = shap.DeepExplainer(model, background)
back= test_x[np.random.choice(test_x.shape[0], 1000, replace=False)]
shap_values = explainer.shap_values(back)
shap.summary_plot(shap_values[0][:,0,:], plot_type = 'bar', feature_names = features)
but when i try to plot all three classes on a single summary plot by this code
shap.summary_plot(shap_values,back_x, plot_type="bar",feature_names = features)
it gives me following error
IndexError: index 12 is out of bounds for axis 0 with size 1
how to plot all the 3 classes on a single summary plot?
Related
Background
I'm learning about LSTMs and figured I'd try training a very basic LSTM model (big believer of learning through doing).
To start with something basic, I tried implement a LSTM that would sum up the last 10 inputs it has seen. I generated a dataset consisting of 1000 random numbers between 0 and 1, with 1000 labels representing the sum of the previous 10 numbers (label[i] = data[i-9:i+1].sum()), and tried to train the LSTM to recognize this pattern.
I know that simpler models can solve this (ie. linear regression), but I believe that LSTMs should also be able to solve this fairly basic problem.
My initial implementation does seem to work, but when I try to improve the implementation, I start getting constant output values after a few timestamps, looks like it's approximately the average of the training labels.
I'd appreciate any insights as to why the second and third iterations don't work, especially the third iteration, since it looks like it's the same implementation as what I read in "Deep Learning for Coders with fastai & PyTorch" book.
What I've tried so far
I've done 3 iterations so far:
Initially, I generated all sub-sequences of length 10 from the input data along with the corresponding label ([(data[i-9:i+1], label[i]) for i in range(9, len(data))] and fed this into the LSTM
This iteration worked very well, if I feed a sequence of 10 inputs I get an output from the LSTM that is very close to the sum. However, it is kinda cheating in that I'm basically telling the LSTM that the sequerce is length 10. I believe that LSTM should be able to infer the sequence length, so I tried to remove that bit of information.
In my second iteration, I feed the entire sequence into the LSTM at once: ([data[i] for i in range(len(data)), [label[i] for i in range(len(data))]). Basically, a single input with a sequence length of 1000, with a single output of 1000 labels.
However, after training, while running on validation data of 100, all except the first few labels are always a constant number, approximately the average of the training labels.
In my last iteration, I tried feeding inputs one at a time to the LSTM (1000 inputs with a sequence length of 1), with manually storing the hidden and cell states and passing it into the next run with the next input. This produces similar results to #2.
Network setup
For all runs, I used a single layer LSTM with 25 hidden since it's a fairly simple problem. I did try adding layers with dropout or increasing hidden size but didn't help.
Code samples
First iteration:
class LSTMModel(Module):
def __init__(self, seq_len, layers, input_size, hidden_size):
self.lstm1 = nn.LSTM(input_size=input_size, hidden_size=hidden_size, num_layers=layers, bidirectional=False, batch_first=True)
self.fc = nn.Linear(hidden_size, 1)
def forward(self,x):
x, _ = self.lstm1(x)
# [:,-1,:] to grab the last output for sequence length of 10
x = self.fc(x[:,-1,:])
return x[:,0]
Second iteration:
class LSTMModel(Module):
def __init__(self, layers, input_size, hidden_size):
self.lstm1 = nn.LSTM(input_size=input_size, hidden_size=hidden_size, num_layers=layers, bidirectional=False)
self.fc = nn.Linear(hidden_size, 1)
self.input_size = input_size
def forward(self,x):
x,h = self.lstm1(x)
x = self.fc(x)
return x
Final iteration:
class LSTMModel(Module):
def __init__(self, layers, input_size, hidden_size):
self.lstm1 = nn.LSTM(input_size=input_size, hidden_size=hidden_size, num_layers=layers, bidirectional=False, batch_first=True)
self.fc = nn.Linear(hidden_size, 1)
self.h = [torch.zeros(layers, 1, hidden_size).cuda() for _ in range(2)]
def forward(self,x):
x,h = self.lstm1(x, self.h)
self.h = [h_.detach() for h_ in h]
x = self.fc(x)
return x
def reset(self):
for h in self.h: h.zero_()
What is the most effective way to concatenate 4 corner, shown in this photo ?
(conducting in getitem())
left_img = Image.open('image.jpg')
...
output = right_img
This is how I would do it.
Firstly I would convert the image to a Tensor Image temporarily
from torchvision import transforms
tensor_image = transforms.ToTensor()(image)
Now assuming you have a 3 channel image (although similiar principles apply to any matrices of any number of channels including 1 channel gray scale images).
You can find the Red channel with tensor_image[0] the Green channel with tensor_image[1] and the the Blue channel with tensor_image[2]
You can make a for loop iterating through each channel like
for i in tensor_image.size(0):
curr_channel = tensor_image[i]
Now inside that for loop with each channel you can extract the
First corner pixel with float(curr_channel[0][0])
Last top corner pixel with float(curr_channel[0][-1])
Bottom first pixel with float(curr_channel[-1][0])
Bottom and last pixel with float(curr_channel[-1][-1])
Make sure to convert all the pixel values to float or double values before this next appending step
Now you have four values that correspond to the corner pixels of each channel
Then you can make a list called new_image = []
You can then append the above mentioned pixel values using
new_image.append([[curr_channel[0][0], curr_channel[0][-1]], [curr_channel[-1][0], curr_channel[-1][-1]]])
Now after iterating through every channel you should have a big list that contains three (or tensor_image.size(0)) number of lists of lists.
Next step is to convert this list of lists of lists to a torch.tensor by running
new_image = torch.tensor(new_image)
To make sure everything is right new_image.size() should return torch.Size([3, 2, 2])
If that is the case you now have your wanted image but it is tensor format.
The way to convert it back to PIL is to run
final_pil_image = transforms.ToPILImage()(new_image)
If everything went good, you should have a pil image that fulfills your task. The only code it uses is clever indexing and one for loop.
There is a possibility however if you look more than I can, then you can avoid using a for loop and perform operations on all the channels without the loop.
Sarthak Jain
I don't know how quick this is but here:
import numpy as np
img = np.array(Image.open('image.jpg'))
w, h = img.shape[0], image.shape[1]
# the window size:
r = 4
upper_left = img[:r, :r]
lower_left = img[h-r:, :r]
upper_right = img[:r, w-r:]
lower_right = img[h-r:, w-r:]
upper_half = np.concatenate((upper_left, upper_right), axis=1)
lower_half = np.concatenate((lower_left, lower_right), axis=1)
img = np.concatenate((upper_half, lower_half))
or short:
upper_half = np.concatenate((img[:r, :r], img[:r, w-r:]), axis=1)
lower_half = np.concatenate((img[h-r:, :r], img[h-r:, w-r:]), axis=1)
img = np.concatenate((upper_half, lower_half))
I am trying to cluster NTU-RGB+D 120 skeleton dataset using HDBSCAN. The numpy array of the skeleton data has 5 dimention
**dataset.shape=[40091, 3, 300, 25, 2]**
where No of data = 40091, Coordinates = 3 (x-y-z), No of frame = 300, No of joints = 25, No of body in the video = 2
When I am trying to cluster it using hdbscan but the fit raises an error message saying it only accepts 2D data. How can I do it with 5 dimension. I am completely new to work with skeleton data and clustering.
I have changed the shape of the data using view() so that now it contains
samples * (frames*joints) to reduce the dimension
I have a MNIST Sign Language dataset with pixel values as columns.
I get the error when I try to plot an image at one of the indexes as follows:
#Training dataset
dfr = pd.read_csv("sign_mnist_train.csv")
X_train_orig = dfr.iloc[:,1:]
Y_train_orig = dfr['label']
#Testing dataset
dfe = pd.read_csv("sign_mnist_test.csv")
X_test_orig = dfe.iloc[:,1:]
Y_test_orig = dfe['label']
#shapes of dataset
print(dfr.shape) #(27455, 785)
print(dfe.shape) #(7172, 785)
#Example of a picture
index = 1
plt.imshow(X_train_orig.iloc[index])
#TypeError: Invalid shape (784,) for image data
Looks like the image you are trying to plot is a flattened one corresponding to [B, N], where N is 1x28x28, and B is 27455 which is your image size (27455, 784). This is fine if you want to feed it to a Linear layer of 784 long vector. To plot this image you have to reshape it to correspond to [27455, 1, 28, 28]. You can try this out:
image = X_train_orig.iloc[index]
image = np.reshape(image.values, (28, 28))
plt.imshow(image)
I want to use a Linear, Fully-Connected Layer as one of the input layers in my network. The input has shape (batch_size, in_channels, num_samples). It is based on the Tacotron paper: https://arxiv.org/pdf/1703.10135.pdf, the Enocder prenet part.
It feels to me as if Chainer and PyTorch have different implementations of the Linear layer - are they really performing the same operations or am I misunderstanding something?
In PyTorch, behavior of the Linear layer follows the documentations: https://pytorch.org/docs/0.3.1/nn.html#torch.nn.Linear
according to which, the shape of the input and output data are as follows:
Input: (N,∗,in_features) where * means any number of additional dimensions
Output: (N,∗,out_features) where all but the last dimension are the same shape as the input.
Now, let's try creating a linear layer in pytorch and performing the operation. I want an output with 8 channels, and the input data will have 3 channels.
import numpy as np
import torch
from torch import nn
linear_layer_pytorch = nn.Linear(3, 8)
Let's create some dummy input data of shape (1, 4, 3) - (batch_size, num_samples, in_channels:
data = np.array([1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4], dtype=np.float32).reshape(1, 4, 3)
data_pytorch = torch.from_numpy(data)
and finally, perform the operation:
results_pytorch = linear_layer_pytorch(data_pytorch)
results_pytorch.shape
the shape of the output is as follows: Out[27]: torch.Size([1, 4, 8])
Taking a look at the source of the PyTorch implementation:
def linear(input, weight, bias=None):
# type: (Tensor, Tensor, Optional[Tensor]) -> Tensor
r"""
Applies a linear transformation to the incoming data: :math:`y = xA^T + b`.
Shape:
- Input: :math:`(N, *, in\_features)` where `*` means any number of
additional dimensions
- Weight: :math:`(out\_features, in\_features)`
- Bias: :math:`(out\_features)`
- Output: :math:`(N, *, out\_features)`
"""
if input.dim() == 2 and bias is not None:
# fused op is marginally faster
ret = torch.addmm(bias, input, weight.t())
else:
output = input.matmul(weight.t())
if bias is not None:
output += bias
ret = output
return ret
It transposes the weight matrix that is passed to it, broadcasts it along the batch_size axis and performs a matrix multiplications. Having in mind how a linear layer works, I imagine it as 8 nodes, connected through a synapse, holding a weight, with every channel in an input sample, thus in my case it has 3*8 weights. And that is exactly the shape I see in debugger (8, 3).
Now, let's jump to Chainer. The Chainer's linear layer documentation is available here: https://docs.chainer.org/en/stable/reference/generated/chainer.links.Linear.html#chainer.links.Linear. According to this documentation, the Linear layer wraps the function linear, which according to the docs, flattens the input along the non-batch dimensions and the shape of it's weight matrix is (output_size, flattend_input_size)
import chainer
linear_layer_chainer = chainer.links.Linear(8)
results_chainer = linear_layer_chainer(data)
results_chainer.shape
Out[21]: (1, 8)
Creating the layer as linear_layer_chainer = chainer.links.Linear(3, 8) and calling it causes a size mismatch. So in case of chainer, I have gotten a totally different results, because this time around I have a weight matrix that is of shape (8, 12) and my results have a shape of (1, 8). So now, here is my question : since the results are clearly different,both the weight matrices and the outputs have different shapes, how can I make them equivalent and what should be the desired output? In the PyTorch implementation of Tacotron it seems that the PyTorch approach is used as is (https://github.com/mozilla/TTS/blob/master/layers/tacotron.py) - Prenet. If that is the case, how can I make the Chainer produce the same results (I have to implement this in Chainer). I will be grateful for any inshight, sorry that the post has gotten this long.
Chainer Linear layer (a bit frustratingly) does not apply the transformation to the last axis. Chainer flattens the rest of the axes. Instead you need to provide how many batch axes there are, documentation which is 2 in your case:
# data.shape == (1, 4, 3)
results_chainer = linear_layer_chainer(data, n_batch_axes=2)
# 2 batch axes (1,4) means you apply linear to (..., 3)
# results_chainer.shape == (1, 4, 8)
You can also use l(data, n_batch_axes=len(data.shape)-1) to always apply to the last dimension which is the default behaviour in PyTorch, Keras etc.