Questions tagged [deep-learning]
a new area of Machine Learning research concerned with the technologies used for learning hierarchical representations of data, mainly done with deep neural networks (i.e. networks with two or more hidden layers), but also with some sort of Probabilistic Graphical Models.
276
questions
40
votes
6
answers
10k
views
How to set the number of neurons and layers in neural networks
I am a beginner to neural networks and have had trouble grasping two concepts:
How does one decide the number of middle layers a given neural network have? 1 vs. 10 or whatever.
How does one decide ...
- 573
256
votes
10
answers
415k
views
How to set class weights for imbalanced classes in Keras?
I know that there is a possibility in Keras with the class_weights parameter dictionary at fitting, but I couldn't find any example. Would somebody so kind to ...
- 8,387
46
votes
5
answers
65k
views
Does gradient descent always converge to an optimum?
I am wondering whether there is any scenario in which gradient descent does not converge to a minimum.
I am aware that gradient descent is not always guaranteed to converge to a global optimum. I am ...
- 563
35
votes
6
answers
15k
views
Why do convolutional neural networks work?
I have often heard people saying that why convolutional neural networks are still poorly understood. Is it known why convolutional neural networks always end up learning increasingly sophisticated ...
- 451
191
votes
5
answers
144k
views
What is the "dying ReLU" problem in neural networks?
Referring to the Stanford course notes on Convolutional Neural Networks for Visual Recognition, a paragraph says:
"Unfortunately, ReLU units can be fragile during training and can
"die". For ...
- 3,965
68
votes
5
answers
49k
views
Adding Features To Time Series Model LSTM
have been reading up a bit on LSTM's and their use for time series and its been interesting but difficult at the same time. One thing I have had difficulties with understanding is the approach to ...
- 1,205
67
votes
4
answers
74k
views
Why mini batch size is better than one single "batch" with all training data?
I often read that in case of Deep Learning models the usual practice is to apply mini batches (generally a small one, 32/64) over several training epochs. I cannot really fathom the reason behind this....
- 8,387
43
votes
2
answers
64k
views
Should we apply normalization to test data as well?
I am doing a project on an author identification problem. I applied the tf-idf normalization to train data and then trained an SVM on that data.
Now when using the classifier, should I normalize test ...
- 655
67
votes
11
answers
96k
views
Why should the data be shuffled for machine learning tasks
In machine learning tasks it is common to shuffle data and normalize it. The purpose of normalization is clear (for having same range of feature values). But, after struggling a lot, I did not find ...
- 13.9k
80
votes
5
answers
44k
views
What is the difference between "equivariant to translation" and "invariant to translation"
I'm having trouble understanding the difference between equivariant to translation and invariant to translation.
In the book Deep Learning. MIT Press, 2016 (I. Goodfellow, A. Courville, and Y. Bengio)...
- 963
59
votes
5
answers
75k
views
Number of parameters in an LSTM model
How many parameters does a single stacked LSTM have? The number of parameters imposes a lower bound on the number of training examples required and also influences the training time. Hence knowing the ...
- 1,297
23
votes
2
answers
38k
views
Why ReLU is better than the other activation functions
Here the answer refers to vanishing and exploding gradients that has been in sigmoid-like activation functions but, I guess, Relu...
- 13.9k
175
votes
20
answers
230k
views
How do you visualize neural network architectures?
When writing a paper / making a presentation about a topic which is about neural networks, one usually visualizes the networks architecture.
What are good / simple ways to visualize common ...
- 18.7k
46
votes
4
answers
31k
views
Why is ReLU used as an activation function?
Activation functions are used to introduce non-linearities in the linear output of the type w * x + b in a neural network.
Which I am able to understand ...
- 583
42
votes
5
answers
53k
views
What is the relationship between the accuracy and the loss in deep learning?
I have created three different models using deep learning for multi-class classification and each model gave me a different accuracy and loss value. The results of the testing model as the following:
...
- 1,975
12
votes
1
answer
5k
views
Using a pre trained CNN classifier and apply it on a different image dataset
How would you optimize a pre-trained neural network to apply it to a separate problem? Would you just add more layers to the pre-trained model and test it on your ...
- 667
11
votes
1
answer
21k
views
Keras LSTM with 1D time series
I'm learning how to use Keras and I've had reasonable success with my labelled dataset using the examples on Chollet's Deep Learning for Python. The data set is ~1000 Time Series with length 3125 with ...
- 165
8
votes
1
answer
17k
views
Why should softmax be used in CNN
In the last layer of CNNs and MLPs it is common to use softmax layer or units with sigmoid activation functions for multi-class ...
- 13.9k
6
votes
2
answers
8k
views
What are useful evaluation metrics used in machine learning
I am using CNN in order to predict codes after analyzing text. As an example, I will write "I am crazy" .. the model will predict some code " X321".
All this based on CNN.
I want to evaluate my ...
- 215
87
votes
1
answer
86k
views
When to use (He or Glorot) normal initialization over uniform init? And what are its effects with Batch Normalization?
I knew that Residual Network (ResNet) made He normal initialization popular. In ResNet, He normal initialization is used , while the first layer uses He uniform initialization.
I've looked through ...
- 2,186
73
votes
6
answers
146k
views
Cross-entropy loss explanation
Suppose I build a neural network for classification. The last layer is a dense layer with Softmax activation. I have five different classes to classify. Suppose for a single training example, the <...
- 3,011
30
votes
7
answers
18k
views
Can machine learning learn a function like finding maximum from a list?
I have an input which is a list and the output is the maximum of the elements of the input-list.
Can machine learning learn such a function which always selects the maximum of the input-elements ...
- 309
29
votes
6
answers
10k
views
Deep learning basics
I am looking for a paper detailing the very basics of deep learning. Ideally like the Andrew Ng course for deep learning. Do you know where I can find this ?
- 433
27
votes
2
answers
12k
views
local minima vs saddle points in deep learning
I heard Andrew Ng (in a video I unfortunately can't find anymore) talk about how the understanding of local minima in deep learning problems has changed in the sense that they are now regarded as less ...
oW_♦
- 6,254
24
votes
2
answers
19k
views
What's the difference between the cell and hidden state in LSTM?
LSTM cells consist of two types of states, the cell state and hidden state.
How do cell and hidden states differ, in terms of their functionality? What information do they carry?
user105907
23
votes
3
answers
27k
views
Should I use GPU or CPU for inference?
I'm running a deep learning neural network that has been trained by a GPU. I now want to deploy this to multiple hosts for inference. The question is what are the conditions to decide whether I should ...
- 341
15
votes
1
answer
19k
views
What is a 1D Convolutional Layer in Deep Learning?
I have a good general understanding of the role and mechanism of convolutional layers in Deep Learning for image processing in case of 2D or 3D implementations - they "simply" try to catch 2D patterns ...
- 8,387
10
votes
5
answers
15k
views
In which epoch should i stop the training to avoid overfitting
I'm working on an age estimation project trying to classify a given face in a predefined age range. For that purpose I'm training a deep NN using the keras library. The accuracy for the training and ...
- 188
8
votes
1
answer
40k
views
How to download dynamic files created during work on Google Colab?
I have two different files and on the first, I tried to save data to file as:
np.save(open(Q1_TRAINING_DATA_FILE, 'wb'), q1_data)
On second file, i'm trying to ...
- 185
8
votes
2
answers
13k
views
How to plot cost versus number of iterations in scikit learn?
One of the recommendations in the Coursera Machine Learning course when working with gradient descent based algorithms is:
Debugging gradient descent. Make a plot with number of iterations on the x-...
- 215
4
votes
1
answer
228
views
Fake News Detection problem
I would like to work on a project for Fake News Detection especially for Indians news which are in different languages and different formats.
Fake news as image with no or very less text
Fake news on ...
- 235
2
votes
2
answers
2k
views
Explain Binary Classification with output 0.5 (True)
What is the interpretation of output 0.5 of a typical classifier?
I made a prediction and the probability of that data point being from the True class is 0.5.
- 339
2
votes
1
answer
501
views
Compute backpropagation
I have the question which is mentioned in the above picture. It is trying to find the derivative of f with respect to weight matrix ...
- 561
0
votes
1
answer
302
views
Usefulness of intercept in layman terms - ELI5
I am working on a binary classification problem with 1000 rows and 10 features.
While I did use random forest for classification, I also used LIME to explain the predictions of the random forest.
...
- 2,535
194
votes
6
answers
341k
views
How to draw Deep learning network architecture diagrams?
I have built my model. Now I want to draw the network architecture diagram for my research paper. Example is shown below:
- 2,447
86
votes
8
answers
64k
views
Time series prediction using ARIMA vs LSTM
The problem that I am dealing with is predicting time series values. I am looking at one time series at a time and based on for example 15% of the input data, I would like to predict its future values....
- 1,075
63
votes
4
answers
77k
views
Does batch_size in Keras have any effects in results' quality?
I am about to train a big LSTM network with 2-3 million articles and am struggling with Memory Errors (I use AWS EC2 g2x2large).
I found out that one solution is to reduce the ...
- 1,160
46
votes
2
answers
73k
views
Merging two different models in Keras
I am trying to merge two Keras models into a single model and I am unable to accomplish this.
For example in the attached Figure, I would like to fetch the middle layer $A2$ of dimension 8, and use ...
- 1,033
39
votes
1
answer
56k
views
How does Keras calculate accuracy?
How does Keras calculate accuracy from the classwise probabilities? Say, for example we have 100 samples in the test set which can belong to one of two classes. We also have a list of the classwise ...
- 1,032
28
votes
2
answers
18k
views
Is there away to change the metric used by the Early Stopping callback in Keras?
When using the early stopping callback in Keras, training stops when some metric (usually validation loss) is not increasing. Is there a way to use another metric (like precision, recall, or f-measure)...
- 393
28
votes
3
answers
2k
views
Why are NLP and Machine Learning communities interested in deep learning?
I hope you can help me, as I have some questions on this topic. I'm new in the field of deep learning, and while I did some tutorials, I can't relate or distinguish concepts from one another.
- 449
18
votes
4
answers
49k
views
Question about bias in Convolutional Networks
I am trying to figure out how many weights and biases are needed for CNN.
Say I have a (3, 32, 32)-image and want to apply a (32, 5, 5)-filter.
For each feature map I have 5x5 weights, so I should ...
- 1,971
15
votes
5
answers
8k
views
Why does adding a dropout layer improve deep/machine learning performance, given that dropout suppresses some neurons from the model?
If removing some neurons results in a better performing model, why not use a simpler neural network with fewer layers and fewer neurons in the first place? Why build a bigger, more complicated model ...
- 1,305
13
votes
2
answers
25k
views
Activation function between LSTM layers
I'm aware the LSTM cell uses both sigmoid and tanh activation functions internally, however when creating a stacked LSTM architecture does it make sense to pass their outputs through an activation ...
- 242
13
votes
5
answers
32k
views
How does Sigmoid activation work in multi-class classification problems
I know that for a problem with multiple classes we usually use softmax, but can we also use sigmoid? I have tried to implement digit classification with sigmoid at the output layer, it works. What I ...
- 131
13
votes
1
answer
8k
views
Reason for square images in deep learning
Most of the advanced deep learning models like VGG, ResNet, etc. require square images as input, usually with a pixel size of $224x224$.
Is there a reason why the input has to be of equal shape, or ...
- 583
12
votes
3
answers
17k
views
What are the consequences of not freezing layers in transfer learning?
I am trying to fine tune some code from a Kaggle kernel. The model uses pretrained VGG16 weights (via 'imagenet') for transfer learning. However, I notice there is no layer freezing of layers as is ...
- 337
11
votes
2
answers
1k
views
When do we say that the dataset is not classifiable?
I have many times analysed a dataset on which I could not really do any sort of classification. To see whether I can get a classifier I have usually used the following steps:
Generate box plots of ...
- 188
10
votes
2
answers
25k
views
weighted cross entropy for imbalanced dataset - multiclass classification
I am trying to classify images to more then a 100 classes, of different sizes ranged from 300 to 4000 (mean size 1500 with std 600). I am using a pretty standard CNN where the last layer outputs a ...
- 139
10
votes
2
answers
18k
views
Using Cross Validation technique for a CNN model
I am working on a CNN model. As always, I used batches with epochs to train my model. When it completed training and validation, finally I used a test set to measure the model performance and generate ...
- 1,127