Fadi Bakoura
  • Member for 4 years
  • Last seen more than a month ago
What is the role of the hidden vectors in restricted Boltzmann machines?
Accepted answer
4 votes

Generative models The hidden units are just structural support and we don't care about what those hidden vectors really are. Generative modeling is concerned about $P(X)$, to be able to compute it,...

View answer
Can neural networks efficiently solve the traveling salesmen problem?
2 votes

There is some recent work addressing this issue, to learn the conditional probability of an output sequence with elements that are discrete tokens corresponding to positions in an input sequence. See ...

View answer
What are good parameters of an encoder?
Accepted answer
2 votes

I can not seem to grasp which specs make an encoder better than another one In general, in unsupervised settings, we want to learn the probability distribution of the data p(x) by some latent ...

View answer
Over- and underestimations of the lowest and highest values in LSTM network
1 votes

RNN is a deeply non-linear function over time, how the black linear line is deduced? Assuming you are doing just linear regression if the least square error was used as the loss function, it will ...

View answer
Is this overfitting avoidable?
Accepted answer
1 votes

an epoch takes 3.5 days First of all, use colab to iterate quickly, it offers unlimited 12 hours of free GPU. to retrain it on Imagenet. That said, we could use complex models without ...

View answer
Can we code rules for an agent in python language other than predicate calculus?
Accepted answer
1 votes

You could formulate the problem of predicting disease severity as a classification one, you give the algorithm those 900 attributes and their corresponding labels (severe/not severe) after training, ...

View answer
If we want to classify something as either a cat/dog or neither, do we need 2 or 3 classes?
1 votes

As far as generalization error is concerned, you are better off by learning the data distribution of (A and B) classes using unsupervised criterion. If you capture the underlying factors that ...

View answer
How can I deal with images of variable dimensions when doing image segmentation?
1 votes

Assuming you have a large dataset, and it's labeled pixel-wise, one hacky way to solve the issue is to preprocess the images to have same dimensions by inserting horizontal and vertical margins ...

View answer
Are there any references of NLP/text mining techniques for identifying the theme of news headlines?
Accepted answer
0 votes

You could formulate the problem as a topic classification task, hence you need labeled data. From an unsupervised point of view, you could represent sentences with some fixed feature vector (latent ...

View answer
Unable to overfit using MLP
0 votes

I guess you are using linear activation functions, maybe you are not initializing your weights, or you are regularizing your model enough. Initialize weights with glorot, insert dropout layers in ...

View answer
How to use a VAE to reconstruct an image starting from an initial image instead of starting from a random vector?
0 votes

You could use VAE as previously answered though it will not work well in practice. I think denoising auto-encoder is suitable for your problem because during training, the input is corrupted ...

View answer