Hexagon Geospatial
MENU

ERDAS IMAGINE

Discuss and share topics of interest using ERDAS IMAGINE the world’s leading geospatial data authoring system.
Showing results for 
Search instead for 
Do you mean 
Reply
New Contributor
Posts: 9
Registered: ‎10-17-2016
Accepted Solution

What is the difference between Machine Learning and Deep Learning as used in ERDAS IMAGINE

Hi 

 

Can someone please provide me with a clear distinction between Machine Leaning and Deep Learning as used in ERDAS IMAGINE.

 

I also saw a presentation in which it is says ERDAS IMAGINE integrated the latest deep learning API from Google for dynamic learning to better predict local data.

 

Does this imply that users need to be connected to the internet for the learning process? How is the knowledge created and stored for future use?

 

Any pointers would be greatly appreciated 

 

kind Regards

 

Nisbert

Staff
Posts: 123
Registered: ‎06-30-2016

Re: What is the difference between Machine Learning and Deep Learning as used in ERDAS IMAGINE

Hi Nisbert,

 

Some of the differences between machine learning (ML) and deep learning (DL) in IMAGINE:

 

  • Classification using DL is implemented for Raster data, while classification using ML is implemented for both raster and vector data.
  • When classifying a raster data, ML classification is done at the pixel level, meaning each pixel is assigned a class. DL classification is done at a grid level. An Image is divided into grids and each grid is assigned a class.
  • For ML, you have to select the training data and attributes of the training data. In DL, you only specify the training data
  • When you do image classification using ML, the output is also an image. When you do classification using DL, the output is a vector that defines the grid. The output vector will have an attribute that specifies the class assigned to each grid

 

You dont have to be connected to the internet for training or classification using DL. The knowledge that is created during the training stage is saved as a model (a trained machine intellect) on your local machine.

 

Best regards, Sam

 

New Contributor
Posts: 9
Registered: ‎10-17-2016

Re: What is the difference between Machine Learning and Deep Learning as used in ERDAS IMAGINE

Hi Sam

 

Thanks for the detailed response. 

 

With reference to the use Google API in the DL mechanism - what would that imply or mean to the end user, If we are to explain that component? In simpler terms API is "a set of functions and procedures that allow the creation of applications which access the features or data of an operating system, application, or other service". However, most people understand API to mean that "the application connects to the Internet and sends data to a server which retrieves that data, interprets it, performs the necessary actions and sends it back to the user". 

 

This then introduces concerns around security and privacy etc. - Your assistance in this regard will help clear a lot of uncertainities for all.

 

On Knowledge generation during training:

 

So if the knowledge that is created during the training stage and saved as a model (a trained machine intellect) on your local machine - can the use access this model? 

 

Also does the system make use of knowledge generated in other projects when the user runs a different project?

 

 

Kind Regards

 

Nizza

Technical Evangelist
Posts: 687
Registered: ‎10-01-2015

Re: What is the difference between Machine Learning and Deep Learning as used in ERDAS IMAGINE

Sam was using the term "API" in the traditional sense of the term. There is no internet communcation occuring when using the Inception deep learning operators. They were simply developed using Google's TensorFlow package.

 

The only operator I'm aware of that communicates with external servers via internet services is the Get AWS Landsat 8 Scenes operator.

 

Cheers

 

Ian Anderson
Chief Product Owner, Desktop Remote Sensing
Hexagon Geospatial
Do you need immediate support?
If you encounter a critical issue and need immediate assistance please submit a Service Request through our Support Portal.