Pycon nove

firenze

19-22 aprile 2018

Deep Learning, the Keras way

Goal of this Tutorial

  • Introduce main features of Keras

    • Plus some introductory overview of Tensorflow
  • Learn how simple and Pythonic is doing Deep Learning with Keras

  • Understand how easy is to do basic and advanced Deep Learning models in Keras;

    • Examples and Hand-on Excerises along the way.

Attention: Spoilers Warning!

  • Setup (10 mins)

  • Part I: Introduction (~65 mins)

    • Intro to ANN (~20 mins)

      • naive pure-Python implementation
      • fast forward, sgd, backprop
    • Intro to Tensorflow (15 mins)

      • Model + SGD with Tensorflow
    • Introduction to Keras (30 mins)

      • Overview and main features
        • Tensorflow backend
        • Theano backend
      • Multi-Layer Perceptron and Fully Connected
        • Examples with keras.models.Sequential and Dense
        • HandsOn: MLP with keras
  • Coffe Break (30 mins)

  • Part II: Supervised Learning and Convolutional Neural Nets (~45 mins)

    • Intro: Focus on Image Classification (5 mins)

    • Intro to ConvNets (25 mins)

      • meaning of convolutional filters
        • examples from ImageNet
      • Meaning of dimensions of Conv filters (through an exmple of ConvNet)
      • Visualising ConvNets
      • HandsOn: ConvNet with keras
    • Advanced CNN (10 mins)

      • Dropout
      • MaxPooling
      • Batch Normalisation
    • Famous Models in Keras (likely moved somewhere else) (10 mins) (ref: https://github.com/fchollet/deep-learning-models) - VGG16 - VGG19 - ResNet50 - Inception v3

      • HandsOn: Fine tuning a network on new dataset
  • Part III: Unsupervised Learning (10 mins)

    • AutoEncoders (5 mins)
    • word2vec & doc2vec (gensim) & keras.datasets (5 mins)
      • Embedding
      • word2vec and CNN
    • Exercises
  • Part IV: Advanced Materials (20 mins)

    • RNN and LSTM (10 mins)
      • RNN, LSTM, GRU
    • Example of RNN and LSTM with Text (~10 mins) – Tentative
    • HandsOn: IMDB
  • Wrap up and Conclusions (5 mins)


Requirements

This tutorial requires the following packages:

  • Python version 3.5.x

    • Python 3.4 should be fine as well
    • likely Python 2.7 would be also fine, but who knows? :P
  • numpy version 1.10 or later: http://www.numpy.org/

  • scipy version 0.16 or later: http://www.scipy.org/
  • matplotlib version 1.4 or later: http://matplotlib.org/
  • pandas version 0.16 or later: http://pandas.pydata.org
  • scikit-learn version 0.15 or later: http://scikit-learn.org
  • keras version 1.0 or later: http://keras.io
  • tensorflow version 0.9 or later: https://www.tensorflow.org
  • ipython/jupyter version 4.0 or later, with notebook support

(Optional but recommended):

The easiest way to get (most) these is to use an all-in-one installer such as Anaconda from Continuum. These are available for multiple architectures.

in on sabato 8 aprile at 14:30 See schedule

Comments

  1. Gravatar
    I don't want to brag about my detective skills, but the comments above might not be from 100% authentic conference attendees.
    — Matteo Barbieri,
  2. Gravatar
    I always spent my half an hour to read this website's posts everyday along with a mug of coffee.
    — broderie,
  3. Gravatar
    Excellent site you have here.. It's difficult to find excellent writing like yours these days.

    I seriously appreciate people like you! Take care!!
    — tonights girlfriend,

Nuovo commento