Machine Learning Classifier cannot use?

Help Document Understanding
1

I’m learning about Document Understanding, and I’m a bit confused. In the Classification step, I learned there are 3 classifiers. But when I got to the Machine Learning Classifier, I read this:

image
image2029×817 99.5 KB

I also found this in Docs:

image
image2117×966 137 KB

What does this mean? Is this classifier not available for use? Can someone explain why it “went away”?

How does one classify a document now using a ML Skill in Studio?
OR HAVE THEY TAKEN THAT ABILITY AWAY?

EDIT: I really need an answer to this, as I am writing a user manual to explain this stuff. I have scoured all the docs and cannot find any explanation.

2

@Terry_Marr

  1. There is a complete revamped document understanding modern version release…which uses a differenr set of activities and ml classifier is different in it and not this one
  2. Coming to out of aupport that is because of the version that is select 21.10 …just change version and it ahould be different as 21.10 already went out of support

Check this for more details and differences in packages etc - Document Understanding - About Document Understanding™

Cheers

3

So why does it say “you cannot use in modern projects”? Which activity do we use if not the Machine learning Classifier? Is there a different package? A different activity?

Do you have a link to a tutorial or docs on how to setup, train, and use the Machine Learning Classifier?

4

@Terry_Marr

To know differences check this…this is same link as above

modern works in different way

To use ml classifier…you can go to ai center create a classic project and then train a classification model and use the end point in ml classifier…this is how ypu would use it

In classic you have ml classifider and in modern you have classify document

Check these videos on how to create modern peojects…classic you have a template in studio directly can check that

Cheers

5

Thank you for that.

Where I got turned around is the use of the terms “classic” and “modern” project. I thought they meant classic v modern project in STUDIO. I didn’t realize they were using the exact same terms in AI Center as well. How confusing.

So, just to be clear, IF I WANT TO CREATE A CLASSIFICATION MODEL FROM SCRATCH, I need to create a classic project in AI Center. Correct? Then, the package I would use would be…what? Document Understanding? Or do I use Document Classifier as well?
And then I need to do the training on my own datasets… is this right? The video shows extraction, is it the same for classifying? Can I use the same datasets for both classification and extraction?

The above tutorials are both for modern projects. Is there one for classic projects?

Wait… I just saw your other answer on using the Uploader. Let me watch that and get back to you.

1 Like
6

@Terry_Marr

Classification can be done in both

If you need to use ml classifier activity then classic

If you want to use classify document then use modern

The videos are to show modern …classic you have tons of videos in youtube or is same as explained in the doc you are referring using classification trainer

Cheers

7

So I wanted to start from scratch. I created a new project in AI Center, but I don’t see any way to select “classic” or “modern” there. Am I missing something?

8

@Terry_Marr

Instead of ai center go to document understanding pge…then you would see the option once you click on create new project

Cheers

9

So I have to use Document Understanding now? Why can’t I do everything in AI Center? You can see that for beginners how confusing this all is. As a pro, I want to learn the most robust way to do it, without wizards doing it all for me, so that I understand how it all works.

According to your uploader video, you have to use the Uploader Activity in Studio to set up a classifying training set in AI Center, is that correct? There’s no way to just add a dataset directly in AI Center, which can be used for both classifying and extraction training and evaluation?

10

@Terry_Marr

See document understanding is a small part of ai center…ai center has much more capabilities…it is not limited to document understanding

I understand you want to learn and all of this is complex…yes it would take a little time

Coming to classification…using train classifier and uploading from studio is mandate…for extraction you have data labelling session in ai center…also if you want to indicate classification fields along with extraction then it is possible with data labelling as it gives options…but only classification it does not provide

Also its a broad concept…if you try to hurry you would get lost…my advise would be to take it show…and as in past few years due to advancements there are much more newers changes that have come in…which might be overwhelming…

Cheers