AI: Custom Deep Neural Network Models for Object Recognition (18347) (Mar 2018)

Learning path description

This advanced level Artificial Intelligence learning path provides instruction on how to build a custom Deep Neural network for object detection and classification.


Difficulty Level : Advanced
Product : Microsoft Azure
Role : Technical
Country : Global
Essential Mathematics [48 h 0 m]

This course covers the key mathematical concepts required to understand and build machine learning models.

Launch course
Basic Statistical background [24 h 0 m]

After covering the basic mathematics, next step is to understand basic statistical concepts that are required to train a machine learning model and interpret the accuracy of the results.

Launch course
Basics of Machine Learning [24 h 0 m]

This course covers the theoretical underpinning of machine learning concepts and explains the key machine learning algorithms.

Launch course
Machine Learning - Programming [54 h 0 m]

This course covers the implementation of machine learning algorithms using data structures available in Python.

Launch course
Deep Learning Explained [48 h 0 m]

This course provides an introduction to Deep Neural Network and covers implementation of a DNN using Microsoft Cognitive Tool Kit.

Launch course
Understanding key aspects of Object Detection [0 h 20 m]

This talk outlines the key aspects of building an object detection mode using CNTK.

Launch course
Loading image data for experimenting and building a model [0 h 0 m]

This page provides details on loading the CIFAR-10 dataset for training a Deep Neural Network.

Launch course
Build an Object Detection and Classification model [0 h 0 m]

This tutorial provides the details to implement a Convolution Neural Network in CNTK to identify and classify the images of different objects.

Launch course
Using pre-trained models for Object Detection using Transfer Learning [0 h 0 m]

In the previous tutorial, you trained a CNN model from scratch. With Transfer Learning, you can take an existing trained model and adapt it to your own specialized domain.

Launch course
Faster Object Detection using R-CNN [0 h 0 m]

Deep Neural Networks generally take time to train especially if there are many training examples. Fast R-CNN is able to train faster than a conventional CNN thus reducing the training time.

Launch course

Saving please wait...