Receptive Field Calculator

We built a tool for calculating the receptive field of convolutional filters:  #deeplearning

  • A convolutional layer operates over a local region of the input to that layer with the size of this local region usually specified directly.
  • You can also compute the effective receptive field of a convolutional layer which is the size of the input region to the network that contributes to a layers’ activations.
  • For example, if the first convolutional layer has a receptive field of 3×3 then it’s effective receptive field is also 3×3 since it operates directly on the input.
  • However if the second layer of a convolutional network also has a 3×3 filter, then it’s (local) receptive field is 3×3, but it’s effective receptive field is 5×5.

A convolutional layer operates over a local region of the input to that layer with the size of this local region usually specified directly. You can also compute the effective receptive field of a convolutional layer which is the size of the input region to the network that contributes to a layers’ activations. For example, if the first convolutional layer has a receptive field of 3×3 then it’s effective receptive field is also 3×3 since it operates directly on the input.
Continue reading “Receptive Field Calculator”

GitHub

Visualizing output of activation functions of CNNs: Comments  #DeepLearning #ML#AI

  • From Andrej Karpathy’s course cs231n:CNNs for Visual Recognition

    All the plots were generated with one full forward pass across all the layers of the network with the same activation function

    There are layers, each layer having units.

  • Random data points of training examples are generated from a univariate “normal” (Gaussian) distribution of mean and variance .
  • Weights for each layer were generated from the same distribution as that of but later on varied to obtain different plots.

DeepNets – How weight initialization affects forward and backward passes of a deep neural network
Continue reading “GitHub”

Demystifying artificial intelligence

Demystifying artificial intelligence  #AI #analytics #data

  • you could say that the computer isn’t programmed to print the number 168 in the sense that the code did not say
  • But the computer was programmed to print the number 168.
  • Any talk of computers doing things they weren’t programmed to do is only a way of speaking.
  • Artificial intelligence is more interesting because it increases the degree of indirection, but it’s still software instructing a computer to take in data and apply algorithms.
  • I was at a presentation once where software vendors were claiming that their software “discovered” the equation of motion for a pendulum.

Read the full article, click here.


@EasyAnalytics1: “Demystifying artificial intelligence #AI #analytics #data”


Computers do what they’re programmed to do. They can be most useful when we program them indirectly.


Demystifying artificial intelligence