Can we improve performance by using BatchNorm with PReLU? [closed] - machine-learning

Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 4 years ago.
Improve this question
I have a network as follows
BN-Scale-ReLU
I want to replace ReLU by PReLU. Then, it will be
BN-Scale-PReLU
Could I obtain any gain with the second setting? Why? As I search,The second setting is not so popular. In some paper, they replaced BN-Scale-ReLU=PReLU. Is it right?

There is a paper evaluating these choices, which can be found here: https://arxiv.org/pdf/1606.02228.pdf. They do get better accuracy by using PReLU, but that is very minor. I am unsure if the improvement offsets the higher workload you have to do by using PReLU instead of ReLU. The question is are you already evaluating for that last percentage point in accuracy? If not do not bother yet with choices that only have minor impact on the performance of the model.

Related

End-to-end machine learning project processes [closed]

Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 1 year ago.
Improve this question
I've read a book chapter that walks you through all the steps involved in an end-to-end machine learning project. After doing all the practical exercises I'm still not quite sure that my way of thinking about the whole process is right.
I've tried to depict it in the following flowchart:
Is this the right way of thinking about all the steps in an ML project? Is something missing?
Seems decent.
Just want to mention that the cross-validation and model-selection in your short-listing step could also include tuning the pipelines, because different types of transformations may be suitable to different models.
For example, when there are sparse or categorical features, the pipelines may matter a lot.

Is this possible to predict the lottery numbers (not the most accurate)? [closed]

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 3 years ago.
Improve this question
I am looking for the machine learning correct approach for predicting the lottery numbers, not the most accurate answer but at least we have some predicted output. I am implementing the regression based and neural network models for this. Is their any specific approach which follows this?
It is impossible. The lottery numbers are random - actually to be more specific, the system is chaotic. You would require the initial configuration (positions etc) to insane (possibly infinite) precision to be able to make any predictions. Basically, don't even try it.

Why are there no dropout layers in inception net? [closed]

Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 2 years ago.
Improve this question
I recently was implementing the InceptionNet and came across the scenario where the dropout layer was not implemented in the network at all in the early or mid stages. Any particular reason for this?
You can see this paper posted model:
It actually has a slight regularization effect which is similar to dropout.
Think like that we are choosing every node with a certain possibility for that
layer so we creating our NN architecture with a possibilities. Similar
situation is valid also in here but this time we apply the all possibilities.
Hence, inception network helps to prevent over fitting the parameters so that
learning is happening for more deeper understanding please check out the
original paper but that is just an observation not a prove.

Training and Testing deep CNN with same piece of data [closed]

Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 5 years ago.
Improve this question
I have created a deep CNN from a research paper (using tensorflow) and am now curious if I have done everything correctly. Eventually I want to train and test the CNN on many images, but at the moment I only have one image on hand. If I was to use this one image as training and testing data, should the CNN always have 100% accuracy
Yes, with only one image to match, you will have a trivial case of perfect accuracy (1-for-1). However, this is only a "breath of life" test for your model. All you'll know is that you are functionally capable of running one image through that model; this will tell you nothing (or very little) about your topology's effectiveness with that type of image.

How to prove the convergence of EM? [closed]

Closed. This question is off-topic. It is not currently accepting answers.
Want to improve this question? Update the question so it's on-topic for Stack Overflow.
Closed 10 years ago.
Improve this question
Can you anybody explain how to prove the convergence of the Expectation Maximization algorithm?
For example EM for coins problems: https://math.stackexchange.com/questions/25111/how-does-expectation-maximization-work
EM algorithm does maximum likelihood estimation. If you look at the log likelihood, it's not true that both E and M steps always maximize it. However, if you look at the negative free energy function, both of them always maximizes it, with respect to different things though (so kind of like coordinate descent). So yes, EM algorithm always converges, even though it might converge to bad local extrema, which is a different issue.
Take a look at the classical paper www.cs.toronto.edu/~radford/ftp/emk.pdf to learn more yourself.
EM algorithms do not always converge. So the specifics of your problem will be important.

Resources