Namaskar I am Kushal from learnvern.
In the continuation to the tutorial on machine learning we will further see this tutorial today . So let’s get started.
Today we are going to see advanced trends in machine learning. Now just by seeing that you must already be having a sort of excitement.
In this I will tell you a specific thing, and based on that we will do the discussion and this topic and the topic we now have is Auto ML.
This word if you see auto ml, meaning automatic, which happens on its own, so in machine learning whatever discussions we did till now , and whatever programs we saw , all of them we had made our own but yes there was one thing in that the libraries which were already developed like sklearn or pandas which helped a lot in EDA and along with them we have other libraries too like keras and tensor flow.
So these libraries did help us and in pre-ease there were some extra libraries but if I want to choose a better model then I will have to run all codes manually and check their performance, then optimize them but to do all this If I have an automatic library or tool then, I believe that this is exciting.
So what we are discussing, this is what auto ml does, automated machine learning , where machine learning models are developed to solve real world problems but using automation.
So here we will have some libraries and tools that will help us to learn the first algorithm and then evaluate the second one and human intervention is the least.
So here as a budding data scientist or a machine learning engineer or data engineer , then you are going to get a very good help from this auto ml and here you can utilize your existing knowledge and skills , that is there, but also you can develop the models quickly and you can make better, means that you do not have to configure anything and there is no need to spend time doing that and you should configure it with auto ml in a better way and by saving you time you can do more, you can actually do more.
So here you have to set pipelines that algorithm 1 then 2 and then 3 so those pipelines you can set , if you want to do hyper parameter tuning then that also you can set , so you just set it and the execution is done by auto ml.
So let us see how it happens, so first of all identify the data set and prepare it and then choose the type of machine learning, choose the model of machine learning, then after that build the analytical model , the ML model , using the algorithm and run that model and the scores that are attained and the errors that are there must be identified , then again train the model using test data and do these revisions , so this complete process for ML, the automated ML toolset will follow and following it will also display and give you a suggestion also that this is the better model to run with, this is a better model with which you can work.
So let us see what are the benefits of this, the first benefit is efficiency, that as a human being we can work to a limited capacity, limited speed but if our repeated tasks that we do again and again are done by a machine then our efficiency will itself improve.
The other one is cost saving, because it is fast, efficient and fast cost saving is also done because in less time more work is done so automatically cost saving is being done.
Next part is that of accessibility, now accessibility means that if you are using auto ml then in this , many companies, many companies will want you to work for them because you will be able to deliver that work in a less period of time , in less time you can deliver so what is being achieved is that their time is being saved and their human resources are being saved so that is why this is again a benefit.
Then obviously performance is a benefit, because humanly if you optimize something and that optimization technique if you automate and the machine is optimizing that technique then there is a lot of difference between the two, the machine can work very quickly , fail quickly, quickly learn and quickly implement it also. So, performance is another benefit. Now let us see where we can use it, where its uses are , and the uses are similar to what we have understood till now .
Fraud detection is one use, in finance, in banking accounts then the other use is in research and development especially if we talk about health care data then from that we can draw a lot of insights.
Then Malware and spams, where messaging systems and emailing systems are there, we can use it there. In cybersecurity also for risk assessment, for monitoring and for risk testing also we can use. So this was about Auto ML and now let’s take it a step further to another trend that till now you are working in jupyter notebook or google colab , so ultimately where you will use these machine learning models , these you are going to ultimately use in the cloud.
So if you are going to use it in the cloud then let us introduce that ML, if we have to use it in the cloud then how will we use it.
So in the cloud what happens in that many applications, the major applications are connected to the cloud, they are connected with the cloud.
So in cloud vast data, a massive amount of data is there in the cloud and if we build models and start running them then on that vast , huge amount of data we can run machine learning algorithms.
Now cloud has its own benefits also, if you prepare infrastructure for a large amount of data then it will be so costly and if we assume that after that project you do not require that infrastructure then what will you do, either it will keep lying there or you will sell it at a low cost, so better option is that you use cloud, as cloud has pay per use model, pay as per the usage , so this is a benefit that cloud has given us and along with that you can make many models of machine learning very quickly and then delete them , make them and again delete them and if you want to do parallel processing with ML models then you can scale up the number of servers and when you want you can scale them down also , so that is all according to your requirement and so this is another benefit we see here.
The third benefit is that with the cloud you will observe that you will get a complete market capture and in market capture , you will observe that all applications being made today are cloud based and all those applications are smart applications,intelligent applications. So in a cloud you will already have all those built in capabilities that are intelligent capabilities, modules and libraries and which you can directly use in your applications.
So these are some benefits that we saw that if we move towards the cloud then what benefits we will have. And now what are the platforms that we have, Amazon web services popularly known as aws , microsoft azure, or azuure , Google cloud, then IBM cloud , so these are some of the applications with us. So now these are some cognitive clouds. OK some applications we will see , Cognitive clouds, cognitive cloud is derived from artificial intelligence and signal processing and any cloud using this is called cognitive cloud , so cognitive clous is one example where machine learning is involved.
The other application is chatbots and smart personal assistants , this earlier I was speaking that these are all pre-built and this has already been prepared and this is easily available with us , chatbots or smart personal assistants are not just alexa echo dot but you can take these chatbots as a service also whether they are textual chatbots or some chatbots which are being developed on basis of voice and some can talk on basis of video also.
You can see that Cortana in windows can work on the basis of voice , Google assistant can work on the basis of voice , so these are the kind of chatbots being developed. Then the next application is IOT cloud where you design some devices in which hardware is also involved, sensors are also involved , so that you can connect with the cloud and in their processing machine learning is involved , so machine learning model you can deploy in the cloud and here your hardware is connected to the network and the moment some data comes it will pass through this and got to the cloud and after processing, the results will be back here, so IOT cloud is another application , Then business intelligence , it is vastly used in business intelligence, business intelligence and business reporting , ML, EDA is vastly used, as ultimately whatever data is there to take out insights from it is the ultimate objective of running organizations in everyday operations , so business application is another application where this can be used. Now there some important trend that are there and I have mentioned them here and you can read about them like artificial intelligence for cyber security, Artificial intelligence and machine learning explainability, hyper automation technologies, democratization of AI with no code/ law code , low code, combination of internet of things and AI ML , and AI ops like devops there is AI ops , then AI enabled chatbots which are very popular , popularization of facial recognition , this algorithm is being widely used and being enhanced and developed , then edge AI means wherever edge devices are there , artificial intelligence may be used , then AI ethics because wherever technologies are being developed there should be ethics also and augmented analytics.
So in all these trends artificial intelligence and machine learning are being extensively used. So friends, today's session will end here and continue in the next session.
If you have any queries or comments, click the discussion button below the video and post there. This way, you will be able to connect to fellow learners and discuss the course. Also, Our Team will try to solve your query.
Share a personalized message with your friends.