A highlight from AI Today Podcast: AI Education Series: Data Preparation for AI

Automatic TRANSCRIPT

That we've been doing now for a while that actually goes all the way back to Twenty eighteen when we were delivering courses in person back when people did that sort of thing but now of course. Everybody's kind of getting back in person so we're still delivering education online. And the reason we do a lot is online. Education is because we have a worldwide audience. I mean we have folks who are listened to our to our podcast for sure but also participate in our courses from literally every continent Maybe even antarctica ought to go take a look at our notes on that one and Our education's really focused on on ai and machine learning and cognitive technology best practices. And what we're going to do on this particular. Podcast share some of the insights and so some of the actual education from of course we take a little snippet from that course and bring it in here and share what we assure with our with our folks who go through the courses and even get certifications and things like that so a little benefit for our today podcast listeners. Exactly because we understand that. You're listening to podcasts to learn something so hopefully you will get a little snippet of knowledge today from our data. Preparation course will be sharing some information from that today. you know. As we mentioned we do offer a wide variety of different courses. So today we'll just be snippet on our data prep course but we do encourage you to check out the other courses and educational offerings as well. In case you're interested in learning more. We offer a lot of role specific education. So it's really tailored you know at that executive level education. Where if you're working already and you're trying to gain further knowledge than this really is a great course for you so we go into a lot of detail breaking down artificial intelligence and related areas. And we do that in. Our podcast is what we've had a lot of great guests on here talking about how they're actually implementing. today in the real world so we wanted to spend a little bit of time in the podcast going over one of those clips On data preparation from our education so As mentioned you know we have dozens of courses and one of our course learning paths. we have also these learning paths. Where if you're trying to achieve a particular outcome with your education And you know you see all these courses don't know what to take. We're like okay. Well you should take these courses in this order. One of those passes are cpi certification which is a methodology cpi methodology for doing ai machine. Learning projects based on literally decades of experience of doing this. You might think wait a second decades. Ai like well. First of all a goes back to nineteen fifty. So let's not decades here. But it's the methodology is based on christie am which has been around literally since nineteen ninety nine two thousand Focus on data mining and has evolved in the form of cpi for a machine learning projects and as part of that methodology one the phases is data preparation and this clip shares. Some of the insights from the data preparation course specifically about the importance of the quality of data. Which you may have guessed is really important but you don't understand how how critical that is to assist him. They literally cannot function without good quality. Data and the next question which is just. How much data do you need to build an assist to make it make it usable and and then also what is that. Data looked like it kind of was the structure of that data. Look like or the lack of structure of that data so As mentioned in this this clip coming up here of we share about twelve and a half minutes or so of that education from our longer course on data preparation so hope you enjoy the core idea. Here what data. Preparation is that. Just like all machine based systems especially machine bae systems that have to learn from data garbage in is garbage out. it's got its own little acronym. Gigo g-go if you ever hear that just a nice way of saying that you can't really create models that learn from experience in any sort of machine learning approach that we've covered in our foundations of ai and applications and all the other courses if you don't have good data to train it on and i'm you need trait. You need good data and sort of two places you need good data when you train the models because the mall is going to use that data to create some representation of that learning in the machine learning model that will then use to pass you know unknown data to to make some predictions or classifications wherever the machinery. Molly's do but also we need good data even if we build a great model. Let's see we're using somebody else's model that's been trained on fantastic data. We built our own malls been been trained on amazing data with great characteristics tons of data. We can't throw garbage data into that either so it's not that we just need good data for training. We need good data across the board. And so we need to think about these these pipelines of bringing in and constantly cleansing and preparing data at both during the training phase. And the inference face so in general a good good data felt in fed into a model that is built on trash garbage will result in garbage results but also bad data felton to fed into a model built with good. Data will also result in bad results.

Coming up next