• ↑↓ pour naviguer
  • ↵ pour ouvrir
  • ⟶ pour sélectionner
  • ⌘ ⌥ ↵ pour ouvrir dans un panneau
  • esc pour rejeter
⌘ '
raccourcis clavier
Search

Convolutional Neural Network

See also: this one assignment on CNN how can we exploit sparsity and locality? think of sparse connectivity rather than full connectivity where we exploiting invariance, it might be useful in other parts of the image as well convolution accept volume of size W_1 \times H_1 \times D_1 with four hyper...

  • Étiquette

    • publié à

    • modifié à

    • durée

      1 min de lecture (140 words)

    • source

      llms.txt

    See also: this one assignment on CNN

    how can we exploit sparsity and locality?

    think of sparse connectivity rather than full connectivity

    where we exploiting invariance, it might be useful in other parts of the image as well

    convolution

    accept volume of size W1×H1×D1W_1 \times H_1 \times D_1 with four hyperparameters

    • filters KK
    • spatial extent FF
    • stride SS
    • amount of zero padding PP

    calculation

    produces a volume of size W2×H2×D2W_2 \times H_2 \times D_2 where:

    • W2=W1−F+2PS+1W_2 = \frac{W_1 - F + 2P}{S} + 1
    • H2=H1−F+2PS+1H_2 = \frac{H_1 - F + 2P}{S} + 1
    • D2=KD_2 = K

    1D convolution:

    y=(x∗w)y(i)=∑tx(t)w(i−t)\begin{aligned} y &= (x*w) \\ y(i) &= \sum_{t}x(t)w(i-t) \end{aligned}

    2D convolution:

    y=(x∗w)y(i,j)=∑t1∑t2x(t1,t2)w(i−t1,j−t2)\begin{aligned} y &= (x*w) \\ y(i,j) &= \sum_{t_1} \sum_{t_2} x(t_1, t_2) w(i-t_1,j-t_2) \end{aligned}

    max pooling

    idea to reduce number of parameters

    batchnorm

    xj=[x1j,…,xdj]x^{j} = [x_1^j,\ldots,x_d^j]

    Batch X=[(x1)T…(xb)T]TX = [(x^1)^T \ldots (x^b)^T]^T

    Vous pourriez aimer ce qui suit

  • HTTP
  • images
  • In memory representation
  • un ancien amour.
  • Cauchy-Schwarz
  • Laplace transform
  • Pavlovian scale
    • Machine learning
    and learning machine.
    Connectionist networks
    and an unstructured overview of patterns in machine learning.