[Pycon] [new paper] "Prakhar Srivastava" - How to learn the world in Keras

[info a pycon.it]

Title: How to learn the world in Keras
Duration: 240 (includes Q&A)
Q&A Session: 0
Language: en
Type: Talk

Abstract: _**Deep Learning**_, a not so new yet still an esoteric field of science. Deep learning has shaped the way of working certain industries and proved to be a disruptive technology. It's the data revolution and everyone wants to jump on the bandwagon! This training will help you give a deep dive into Deep learning. The focus will be on absolute beginners and will introduce them this new field through the help of Keras, a high-level framework built on top of modern deep learning libraries.

This training will be divided into 3 segments:

1. __Beginners__

This would cover all the necessary functions and intuitions, right from cross entropy loss to sigmoids and then finally multi-layered perceptrons or Neural Networks

2. __Image__

This part would focus on the developments in the field of **Computer Vision** brought by deep learning architecture. An implementation of a fully connected - Convolution network and later on, an implementation of ResNet.

3. __Text__

This part would focus on the developments in **Natural language Processing** brought by deep learning, heavy focus on through the time learning models like RNNs and LSTMs. A language model by the help of LSTM will be implemented.

Then finally, implementation of Image captioning would be done to conclude and link everything together, Image and text, ending the training with a nice introduction to sequence to sequence learning

Pre-req: Basic algebra and previous experience with python

Tags: [u'ComputerVision', u'neural network', u'mathematics', u'nlp', u'GPUComputing', u'image-processing', u'statistics', u'mathematical-modelling', u'machine-learning', u'bigdata', u'opencv', u'Algorithms', u'Artificial Intelligence']