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.
Related
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 2 years ago.
Improve this question
I only know linear regression and nothing else. Any shorthand explanations or tricks would do.
seed is just a random number to add randomness to the algorithm. Just put your lucky number there.
n_estimators is a hyperparameter that determines how many trees/estimators are built within the ensemble model. The more you use, the more accurate it's because of the nature of the Gradient Boosting algorithm. The downside is that the larger n_estimators size it's the longer it takes to train and also can, potentially, overfit to your train data, but again, considering the nature of the algorithm it may not.
Another thing regarding n_estimators to consider is that you can achieve a good score with not much estimators (i.e. 300 or 500) and after that point, larger estimators (i.e. 2000) doesn't add nothing more than a potential overfit.
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.
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.
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
I am working on a project for recommending contents to the users. I want to create a profile from each user so that I can cluster them and offer common recommendations, but before I have to be able to measure similarity between these users. I have thought in a questionnaire which can be filled using fuzzy labels.
My question is How I can measure the correlation (similarity) for two users U1 and U2 who answer questions of this kind?
Q1. I think that Tokyo is a nice city. U1: Fully agree U2: Partially
agree
Q2. I have read Don Quixote. U1: Do not agree at all U2: Fully agree
...
Qn. I think that Tarantino is a good film director. U1:
Partially agree U2: Partially agree
I have thought to convert answers into numeric values and then try to compute Pearson correlation coefficient. But I wonder if there are more elegant ways to do that.
It would be useful to convert the numeric values (from provided answers) into one single vector and then apply cosine similarity function. The cosine similarity has proven to be more reliable (and faster) than Pearson correlation coefficient.
Nevertheless, this is not a minor issue and the implementation could be very challenging.
Closed. This question needs details or clarity. It is not currently accepting answers.
Want to improve this question? Add details and clarify the problem by editing this post.
Closed 2 years ago.
Improve this question
I want to do segmentation of video in real time (30 fps, 640х480).
Is it possible on a Core 2 Duo E8500?
Could you point to some library or algorithm that can do this?
Take a look at S. Paris, Edge-preserving smoothing and mean-shift segmentation of video streams, ECCV 2008.
In general, yes, there are some segmentation algorithms that are computationally efficient and are likely to operate at frame rate. Often you can use some optimization tricks to improve the efficiency of algorithms, so if you want to know about a specific task or algorithm you need to provide more details so we can help. I disagree with Abid Rahman K, OpenCV is not necessarily the best option.