Maxim
  • Member for 4 years, 4 months
  • Last seen more than a month ago
Do scientists know what is happening inside artificial neural networks?
Accepted answer
67 votes

There are many approaches that aim to make a trained neural network more interpretable and less like a "black box", specifically convolutional neural networks that you've mentioned. Visualizing the ...

View answer
How do I handle large images when training a CNN?
Accepted answer
21 votes

How do I handle such large image sizes without downsampling? I assume that by downsampling you mean scaling down the input before passing it into CNN. Convolutional layer allows to downsample the ...

View answer
How is it possible that deep neural networks are so easily fooled?
16 votes

All answers here are great, but, for some reason, nothing has been said so far on why this effect should not surprise you. I'll fill the blank. Let me start with one requirement that is absolutely ...

View answer
Why has the cross-entropy become the classification standard loss function and not Kullback-Leibler divergence?
Accepted answer
13 votes

When it comes to a classification problem in machine learning, the cross-entropy and the KL divergence are equal. As already stated in the question, the general formula is this: $$H(p, q) = H(p) + D_{...

View answer
Is it possible to train a neural network as new classes are given?
Accepted answer
11 votes

I'd like to add to what's been said already that your question touches upon an important notion in machine learning called transfer learning. In practice, very few people train an entire convolutional ...

View answer
What is the difference between a convolutional neural network and a regular neural network?
6 votes

A convolutional neural network is one that has convolutional layers. If a general neural network is, loosely speaking, inspired by a human brain (which isn't very much accurate), the convolutional ...

View answer
How is creativity generated in a currently rule based neural network?
Accepted answer
2 votes

Your question is a bit broad. Some examples of creativity and how it's achieved (in all of them there is an ordinary convolutional network trained via supervised learning, but applied differently): ...

View answer
How to detect the empty parking spots?
2 votes

Object detection is a regression of the bounding box (rectangle) around the object. In this way, the two ways you suggest are equivalent. What I suggest you to look at is lane detection for self-...

View answer
How generate variation in datasets
2 votes

I have many symptoms and duration of each drug. I create X and y data but, for example, LSD have an effect duration of 180 - 720 minutes. I really need make 540 arrays? You can (in this ...

View answer
Is there any free dataset of source code along with natural language description?
2 votes

StackOverflow answers with code snippets. This data needs some processing, because the description can be in the question (along with other notes) and along with the answer. But this dataset is very ...

View answer
Is overfitting always a bad thing?
2 votes

Overfitting is almost always bad and hurts generalization. You say what we want from the NN is to say "take a slight left turn" or "turn right hard" if another ship comes slightly close on the ...

View answer
What is $I$ in the noise described in the paper "Parameter Space Noise for Exploration"?
Accepted answer
1 votes

Yes, since $\tilde{\theta}$ is a vector, to define its distribution one needs a covariance matrix. Here $I$ is the identity matrix, which means that the noise has a zero-mean normal distribution with ...

View answer
Why is the change in cost wrt bias in neural network equal to error in the neuron?
1 votes

This is just an application of the chain rule. The same chapter has "Proof of the four fundamental equations" section, which proves BP1-2, while PB3-4 are left as exercise to the reader. I agree that ...

View answer
Why do we need weights when training a perceptron as an OR gate?
1 votes

This would successfully simulate an OR gate. Of course. In fact, hardware implementation of an OR gate needs just a few transistors. It may sound surprising, but the best python implementation of OR ...

View answer
What are the counterparts of non-linearities and dropout in fully convolutional networks?
1 votes

All-convolutional neural network is a more general concept which can be (and is often) used without deconvolutional and unpolling layers, e.g. for an ordinary classification task. The idea is to ...

View answer
How do I know if my backpropagation is implemented correctly?
1 votes

The given so far answers focus on numerical methods to check your gradients. It is really useful, especially if one doesn't have much experience in backprop. But I'd like to add here a pure practical ...

View answer
Understanding a paragraph about object detection with two objects
1 votes

If I understood that post right (I just skimmed through, so it's possible I missed some details), they are using several predictors on the inputs with several rectangles. This basically means separate,...

View answer
In a neural network given partial inputs and complete outputs, is it possible to predict remainig inputs
1 votes

It's important to understand that though neural networks generalize to the whole input space, usually the meaningful input space, from which the training data is taken, is a manifold inside that space....

View answer
ReLu, Sum and Convolution Layers to Count Pixels of Certain Color
Accepted answer
0 votes

I can judge from the excerpt only, but I don't see any mention of a convolutional layer there. This doesn't mean that convolution isn't suitable for this task (in fact, it's the best method for image ...

View answer
A good way to understand the mathematical details of variational autoencoders through implementation?
Accepted answer
0 votes

I even don't know what exactly back-propagation is used for, if this would help. Programming backprop manually is one of the best exercises in machine learning. In fact, to do this you don't need a ...

View answer