Hello guys, welcome to the 20% of api.ai tutorials for beginner. In this session we will see what is the machine learning and how to enable or disable or train our agent for that. Machine learning is a method of teaching computer to make improve prediction and the behavior based on the data using the machine learning agent can under show the user's interaction. in APA da da terminology, your agent uses the machine learning algorithm to match the user request to the specific intent and use the entity to extract the relevant data from them. for machine learning APA da da created a an algorithm that will learn from the user section and the language model that created by the epa.ai visa for this algorithm will be different for Each agent, these are all in them will be dynamically updated as any changes made in the agent. API dot A by default enable DML for your region.

However, you can disable it for the specific intent. Let's switch to the browser and click on the gear icon to open the settings. From here, click on the MLS setting. First is the match mode. There are two options for that hybrid which include the rule base and the emission running and another is the machine learning only. Hybrid is suggested when you have the view intent and ml is suggested when you have a large number of the intent.

Next is the classification of threshold which will be between zero to one I'm going to open the intent and they click on the any intent to disable the machine learning for any intent we need to click on the three dots and select here disable ml and to make it enable we need to again click on the Enable ama, but I suggest to remain the error enabled. Now I am going to send the request some of then and 20 and here we got the answer 30. If you notice that here is a spelling mistake, instead of oaf I am right the OD, still we got the result. This was the machine learning algorithm that was used by the pa da da. Let open this intent You can see that here is only one item in the user section and in which you I am specified Oh f not o g, still we got the response.

And now I am going to write the S om sum of 10 and 20. Again we got the response. Here you may or may not get the answer, because it was a depend on the algorithm. In my case, the algorithm learn it. In your case, it may not learn. However, we can also train the region manually.

For this we need to click on the training and now I'm going to send a request for the ad 10 and 20. Here it was the executed custom person demo intent. Let's open it. However here we are returning one entity or type custom entity. And in this query we specify the two numeric values stated was a call to this. Now I'm going to train the am eyes and say that it will execute the Kelsey demo for this query instead or the custom person demo.

For that, click on the training from the navigation. You can see that it includes the all of the requests that we send it and click on the last one and scroll down to the bottom. Here is the request of the ad 10 and 24 that it was provide the question personally and it should have this I want to use the Kelsey demo, so select it from the drop down. And I don't want to execute the Kelsey demo intend indicates or the SRM some for that time, click on the intent name and click the Remove parameter and click on the approve button. You can see that the our agent training was started and it was completed to now I am going to write add 10 and 20. You can see that now it was go to the Kelsey demo how it has the only one parameter value let open to its intent.

You can see that it was created the new parameter instead of using the our existing parameter for select the value and select the parameter that was executed I am using the enum to for that and click on the sale. Now send the same request for the 10 and 20. This time we got the values as well as the response. Here we got the response, it was not messy. Because we are added the value in the user says section. Let's see what we are learning in this session.

We see the machine learning improve the prediction and the behavior of the robot. Every agent use the algorithm and each has the a different result. agent will automatically train if any changes made in the agent API to enable the machine learning by default. From agent setting, we can configure the MLS. We can specify the match will be performed based on the rule And DML or just ml only, and specify the threshold for the classification. We can also use the training page to manually add the item into the our user section.

Thanks for watching, have a nice day. If you have any doubt feel free to contact me