Exercise1.ipynb

{
  "cells": [
    {
      "metadata": {
        "collapsed": true
      },
      "cell_type": "markdown",
      "source": "# Execrise 1. Introduction \n\nDog breed classification using Pytorch Estimators on Azure Machine Learning service\n\nHave you ever seen a dog and not been able to tell the breed? Some dogs look so similar, that it can be nearly impossible to tell. For instance these are a few breeds that are difficult to tell apart:\n\n#### Alaskan Malamutes vs Siberian Huskies\n![Image of Alaskan Malamute vs Siberian Husky](http://cdn.akc.org/content/article-body-image/malamutehusky.jpg)\n\n#### Whippet vs Italian Greyhound \n![Image of Whippet vs Italian Greyhound](http://cdn.akc.org/content/article-body-image/whippetitalian.jpg)\n\nThere are sites like http://what-dog.net, which use Microsoft Cognitive Services to be able to make this easier. \n\nIn this tutorial, you will learn how to train a Pytorch image classification model using transfer learning with the Azure Machine Learning service. The Azure Machine Learning python SDK's [PyTorch estimator](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-train-pytorch) enables you to easily submit PyTorch training jobs for both single-node and distributed runs on Azure compute. The model is trained to classify dog breeds using the [Stanford Dog dataset](http://vision.stanford.edu/aditya86/ImageNetDogs/) and it is based on a pretrained ResNet18 model. This ResNet18 model has been built using images and annotation from ImageNet. The Stanford Dog dataset contains 120 classes (i.e. dog breeds), to save time however, for most of the tutorial, we will only use a subset of this dataset which includes only 10 dog breeds."
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "## What is Azure Machine Learning service?\nAzure Machine Learning service is a cloud service that you can use to develop and deploy machine learning models. Using Azure Machine Learning service, you can track your models as you build, train, deploy, and manage them, all at the broad scale that the cloud provides.\n![](aml-overview.png)\n\n\n## How can we use it for training image classification models?\nTraining machine learning models, particularly deep neural networks, is often a time- and compute-intensive task. Once you've finished writing your training script and running on a small subset of data on your local machine, you will likely want to scale up your workload.\n\nTo facilitate training, the Azure Machine Learning Python SDK provides a high-level abstraction, the estimator class, which allows users to easily train their models in the Azure ecosystem. You can create and use an Estimator object to submit any training code you want to run on remote compute, whether it's a single-node run or distributed training across a GPU cluster. For PyTorch and TensorFlow jobs, Azure Machine Learning also provides respective PyTorch and TensorFlow estimators to simplify using these frameworks.\n\n### Steps to train with a Pytorch Estimator:\nIn this tutorial, we will:\n- Connect to an Azure Machine Learning service Workspace \n- Create a remote compute target\n- Upload your training data (Optional)\n- Create your training script\n- Create an Estimator object\n- Submit your training job"
    },
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "Understanding Azure Machine Learning [Workspace](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture#workspace) is while it is being created.\n![](aml-workspace.png)\n\n\n## Create a remote compute target\nFor this tutorial, we will create an AML Compute cluster with a NC6 GPU machines, created to use as the [compute target](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture#compute-target) to execute your training script on. \n"
    }
  ],
  "metadata": {
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3",
      "language": "python"
    },
    "language_info": {
      "mimetype": "text/x-python",
      "nbconvert_exporter": "python",
      "name": "python",
      "pygments_lexer": "ipython3",
      "version": "3.5.4",
      "file_extension": ".py",
      "codemirror_mode": {
        "version": 3,
        "name": "ipython"
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 2
}