Scribd Logo
Upload

Welcome to Scribd!

  • 78 views

    CNN Course V1.3

    Uploaded by

    Rahul
    Convolutional neural networks (CNNs) are a type of deep learning algorithm that uses multi-layer neural networks and shared weights to process data with a grid-like topology such as images. CNNs apply successive layers of convolution and pooling operations to extract hierarchical features from images. CNN architectures typically include convolutional and nonlinear layers followed by pooling layers, with fully connected layers at the end for classification. CNNs achieved breakthrough results in visual recognition tasks by significantly outperforming other algorithms on the ImageNet challenge in 2012.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    CNN Course V1.3

    Uploaded by

    Rahul
    78 views19 pages
    Convolutional neural networks (CNNs) are a type of deep learning algorithm that uses multi-layer neural networks and shared weights to process data with a grid-like topology such as images. CNNs apply successive layers of convolution and pooling operations to extract hierarchical features from images. CNN architectures typically include convolutional and nonlinear layers followed by pooling layers, with fully connected layers at the end for classification. CNNs achieved breakthrough results in visual recognition tasks by significantly outperforming other algorithms on the ImageNet challenge in 2012.

    Original Title

    CNN course V1.3

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Convolutional neural networks (CNNs) are a type of deep learning algorithm that uses multi-layer neural networks and shared weights to process data with a grid-like topology such as images. CNNs apply successive layers of convolution and pooling operations to extract hierarchical features from images. CNN architectures typically include convolutional and nonlinear layers followed by pooling layers, with fully connected layers at the end for classification. CNNs achieved breakthrough results in visual recognition tasks by significantly outperforming other algorithms on the ImageNet challenge in 2012.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    78 views19 pages

    CNN Course V1.3

    Uploaded by

    Rahul
    Convolutional neural networks (CNNs) are a type of deep learning algorithm that uses multi-layer neural networks and shared weights to process data with a grid-like topology such as images. CNNs apply successive layers of convolution and pooling operations to extract hierarchical features from images. CNN architectures typically include convolutional and nonlinear layers followed by pooling layers, with fully connected layers at the end for classification. CNNs achieved breakthrough results in visual recognition tasks by significantly outperforming other algorithms on the ImageNet challenge in 2012.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    You are on page 1of 19

    Convolutional Neural Networks

    Training at Dell RMZ

    1
    Buzz…

    2
    Deep Learning – from Research to
    Technology

    Deep Learning - breakthrough in


    visual and speech recognition 3
    Deep Learning Basics
    Deep Learning – is a set of machine learning
    algorithms based on multi-layer networks
    CAT DOG

    OUTPUTS

    HIDDEN
    NODES

    INPUTS
    4
    Deep Learning Basics
    Deep Learning – is a set of machine learning
    algorithms based on multi-layer networks
    CAT DOG

    Training

    5
    Deep Learning Basics
    Deep Learning – is a set of machine learning
    algorithms based on multi-layer networks
    CAT DOG

    6
    Deep Learning Basics
    Deep Learning – is a set of machine learning
    algorithms based on multi-layer networks
    CAT DOG

    7
    Deep Learning Taxonomy

    Supervised:
    – Convolutional NN ( LeCun)
    – Recurrent Neural nets (Schmidhuber )

    Unsupervised
    – Deep Belief Nets / Stacked RBMs (Hinton)
    – Stacked denoising autoencoders (Bengio)
    – Sparse AutoEncoders ( LeCun, A. Ng, )

    8
    Convolutional Networks

    9
    Convolutional NN

    Convolutional Neural Networks is extension of


    traditional Multi-layer Perceptron, based on 3 ideas:
    1. Local receive fields
    2. Shared weights
    3. Spatial / temporal sub-sampling or Pooling
    See LeCun paper (1998) on text recognition:
    http://yann.lecun.com/exdb/publis/pdf/lecun-01a.pdf

    10
    What is Convolutional NN ?
    CNN - multi-layer NN architecture
    – Convolutional + Non-Linear Layer
    – Sub-sampling Layer or Pooling layer
    – Convolutional +Non-Linear Layer
    – Fully connected layers
    • Supervised

    Classi-
    Feature Extraction
    fication

    11
    Receptive field and padding

    12
    Convolution operation

    13
    Conv operation – filter size, stride and
    padding

    14
    What is Convolutional NN ?

    2x
    2

    Convolution + NL Sub-sampling Convolution + NL


    15
    CNN success story: ILSVRC 2012

    Imagenet data base: 14 mln labeled images, 20K categories

    16
    Imagenet 2013: top rankers

    http://www.image-net.org/challenges/LSVRC/2013/results.php

    N Error-5 Algorithm Team Authors


    1 0.117 Deep Convolutional Neural Clarifi Zeiler
    Network
    2 0.129 Deep Convolutional Neural Nat.Univ Min LIN
    Networks Singapore
    3 0.135 Deep Convolutional Neural NYU Zeiler
    Networks Fergus
    4 0.135 Deep Convolutional Neural Andrew Howard
    Networks
    5 0.137 Deep Convolutional Neural Overfeat Pierre Sermanet
    Networks NYU et al

    17
    Imagenet Classifications 2013

    18
    Conv Net Topology

    • 5 convolutional layers
    • 3 fully connected layers + soft-max
    • 650K neurons , 60 Mln weights

    19

    You might also like