How to use UiPath AI Center (former Ai Fabric) - Beginner Tutorial
HTML-код
- Опубликовано: 27 июл 2024
- Another one of my end-to-end UiPath examples of how AI Center (formerly called Ai Fabric) works and can be used, including model training and deployment, as well as use of ML-related activities in UiPath Studio. #UiPath #AiFabric #MachineLearning
Timecodes:
0:00 Intro
0:22 Machine Learning concepts
3:37 Activating AiFabric in the UiPath Platform
4:30 Creating a project
4:55 Adding a model/package
6:06 Adding a Training dataset
7:15 Creating a Pipeline for training
8:28 Adding an Evaluation dataset
9:07 Creating a Pipeline for evaluation
10:20 Deploying a Machine Learning Skill
11:15 Using a ML Skill inside UiPath Studio
13:23 Building a project using ML activities
15:40 Running the project, including Actions Center
Link to project:
drive.google.com/file/d/1hsqw... - Наука
Amazing work of yours always!!! Thankyou so much for making topic simple and clear..
Thank you :-)
amazing effort!
very clear and easy to understand
Thank you Eduardo, glad you like it. Much appreciated :-)
Awesome Content. Easy to understand. You rock always..
Wonderful intro, thank you for sharing.
Thanks Ken, I really appreciate it. Let me know if you have any ideas for future videos :-)
Hey Jeppe. The intro with the graphics made it awesome. My drawing needs to improve 😀 Kind regards, Anders
Thanks Anders, coming from you, that really means a lot. I tried freehand drawing at first, but decided to go with PowerPoint in the end :-D
Hey Jeppe, Can I get the same training and evaluation file which you used for this ?
excellent, you made it look so clear and simple ...amazing
Glad you liked it! :-)
@@UiPathwithJeppe will be watching all videos soon . great content and great presentation ...i hope the channel grows big
jeppe very well explained thank you :)
Thank you, that really means a lot :-)
Thank you so much!!
You're welcome :-)
Thanks
Is it possible for you to post your training and evaluation datasets? I don't understand how the two differ and if the syntax is different in the evaluation csv file. Thanks!
Great
Thanks :-)
Hey Great work @UiPath with Jeppe , Where I can get the Trainable data which you used in your project? Can you please share the CSV file?
Hi, I'm so sorry, I don't have that data handy anymore.
Great content. Thanks, Jeppe! BTW - Google Drive link is no longer working :(
Thank you for the kind words and the heads-up. I will see if I can find or reproduce the file for google
Where I can get the training data file
Thank you for your explanation. Is it possible to make a sentence context analysis with ai fabric instead of classification?
I'm not super familiar with all the packages in AI Center, but I don't think there is one that can do that. There's a sentiment analysis package for detecting general positivity/negativity in a sentence, but that's pretty much the closest one, I think. It is possible to create custom packages and publish to AI Center, but that is out of scope for my tiny brain, sorry :-)
@@UiPathwithJeppe thank you for your answer ☺️
Hi Jeppe,
Can we have video on how to make orchestration process?
Fair question, absolutely, and I'll probably use an Orchestration Process in Part 2. But also.... what I showed in this demo was - more or less - an orchestration process. An orchestration process differs only from a normal process in a couple of ways:
1. It has the UiPath.Persistance.Activities package installed by default. (I just did it manually).
2. It has the SupportsPersistence option set to True by default. (I just did it manually).
3. The main.xaml in an Orchestration Process is a Flowchart - in a standard Process it is a Sequence.
4. And finally, it Orchestration Process does not have UiAutomation activities installed by default (as well as Excel and Mail).
That's it :-)
Great, Jeppe!
In this examle the outcome has two scenarios, spam, and not-spam. How good is the model if it has to deal with for example 50 scenarios that you want to classify? Does it handle other languages than English well?
I don't have experience with that level of complexity. I imagine it would be possible, but the amount of training data would probably have to be insane.
In AI Center there are specific packages for classification in English, French and Japanese, but there is also a generic classification model that (supposedly) supports about a hundred languages. I'll have to look into that.
@@UiPathwithJeppe thank you for quick reply. This is relevant for handling customer service inquiries, which is a topic for interest for me.
Furthermore; handling 50 scenarios could possibly be done with 49 dummy-models with two scenarios. Of course the True scenario would be a very small percentage compared to the False.
Hi Jeppe..Where did you get the training_Data.csv? did you downloaded it from somewhere? or did you typed by yourself?
Hi, if you go to Kaggle.com, you’ll find all kinds of data science resources; there are tons of datasets there, usually both documented and downloadable. Check it out. :-)
Do you have advice on how to prep data you get from Kaggle? My pipeline failed and I believe it's because my data wasn't set up correctly.
Been there, done that. Can you find me on LinkedIn and connect?
@@UiPathwithJeppe What's your last name? Or can you share link to your profile?
Would you be willing to share your data?