Skip to main content
2 votes

How translation invariance is achieved in CNNs?

Take a vector: $V_1 = [v_1, v_2, ..., v_n]$ Calculate the max: $m_1 = \max V_1 = v_i$ Shuffle the vector: $V_2 = mix(V_1)$ Calculate the max: $m_2 = \max V_2 = v_j$ The only possible outcome is that $...
Alberto's user avatar
  • 2,293
2 votes

Can a concept/feature be represented using more than one layer of a Neural Network?

You cannot reason in a mathematical way over features in my opinion, as they are not defined. However, you can think of deep neurons as a hierarchy of always more high level concepts, as observed in ...
Alberto's user avatar
  • 2,293
2 votes
Accepted

What is $\mathbf{S}$ (sample covariance matrix) in image compression based on PCA?

Good question! There's actually some ambiguity here: it's possible to consider the lower-dimensional projection with respect to the pixels within a single image or across a dataset of images. A ...
Alexander Wan's user avatar
2 votes
Accepted

Is geodesic distance between two similar photos less than the Euclidean distance between them? If so, why?

Imagine your images are embedded in a space that is a circle. Now, to simplify, we put an extra condition in which each embedding (i.e., an image represented somehow in 2 dimensions) must lie on the ...
Luca Anzalone's user avatar
2 votes

how to determine the number of units for dense layer for transfer learning?

Not only the units but also the number of layers... you can reason over something like "how complex is your task", but usually we resort to grid search over some educated guesses (like 2/3 ...
Alberto's user avatar
  • 2,293
2 votes

the best choice to reduce a features vector

Feature selection -- the case in which the features are highly correlated is the prototypical case in which you want to select a subset of independent features that allows for an equal performance. ...
Peblo's user avatar
  • 31
1 vote
Accepted

Why CNN filters (kernels) are randomly initialized?

It is usual to initialize parameters with "a good guess", when you have prior information, in order to help the model converging. However in deep learning most of the time you have no clue ...
Lelouch's user avatar
  • 216
1 vote

how to determine the number of units for dense layer for transfer learning?

In adition to @Alberto answer, Start by the same number of layer and units you removed from original model. If the problem is the same, this will be probably the best solution. After that, if the ...
Jonathan Roy's user avatar
1 vote

Feature vector representation of probability distribution

You will lose a lot of information mapping the 3-dimensional probability distributions onto scalars. The best you can do without loss of information is 2-dimensions per distribution. To get the 2-...
Chillston's user avatar
  • 1,748
1 vote
Accepted

Can Vision Transformers be used to extract features?

Yes, of course they can be used to extract features, just like convolutional networks, even in supervised settings. ViTs are not exclusive to classification, their intermediate layers can also be used ...
Dr. Snoopy's user avatar
  • 1,345
1 vote

How translation invariance is achieved in CNNs?

When you apply a convolutional layer to an image $x$, you obtain a certain list of values: $$h_1(x), h_2(x), h_3(x), ..., h_n(x) \tag 1$$ where each $h_i$ is just the function that applies the ...
Jack M's user avatar
  • 262
1 vote
Accepted

How to extract the high-level features of YOLOv5?

Set export = True at the yolo head. If the net is the AutoShape instance, you can achieve it ...
8ToThePowerOfMol's user avatar

Only top scored, non community-wiki answers of a minimum length are eligible