Artificial Intelligence (AI) on Amazon Web Services (AWS)

How AI/ML resembles to human brain

Human beings have brain and sensory organs using which they analyze and predict everything such as car, mask, animals, etc. So for human, brain is there for their natural intelligence. But in computers, CPU+RAM actually behaves as brain. But having only brain(RAM/CPU) is not important, for analyzing, computer should have intelligence like humans.

So here what we need to do, we anyhow try to provide intelligence to our computer by training our computer. But this intelligence is not real as we humans are the one who are trying to provide intelligence to computer, that’s why this intelligence is called as Artificial Intelligence (AI).

So for learning we need to create an model and then need to train that model by providing historical datasets to give intelligence to our machine (computer), therefore this is called as machine learning(ML).

Now our computer has brain with intelligence. Our human requires sensory organs for sending input to brain, likewise our computer also need some device who send the input. So here devices such as webcam(eyes), microphone (ears), speaker (mouth), etc act as sensory organs who sends the input to computers brain. So human brain and AI/ML resembles each other.

Why we need cloud based services for machine-learning based projects

Cloud service providers has pre-trained model which is more than enough to analyse anything and also they have lots of data with them which is required for ML based project.

Amazon Rekognition

Amazon Rekognition is the service from AWS for ML/AI related use-cases such as image moderation, proctoring service, object detection, image comparison, etc using their pre-trained model.

There are 3 ways to use Amazon Rekognition :- CLI, SDK, GUI.

Confidence Score

How much confident our model is to detect any object is called as confidence score.

Boto3

boto3 is the library of python which helps in contacting AWS from python.

How BoundingBox works in Rek

Bounding box is the area of image in which our model has detected an object.

How labels of REK helps in AI

Labels of REK helps in AI such as in proctoring, for business product, etc.

Function of Amazon-polly service

The function of Amazon-polly service is it is used for “text-to-speech”.

Amazon Lex

Amazon Lex is the service provided by AWS which allows us to create our own chatbots (both Text & Voice) for our application.

NLP (Natural Language Processing)

We human while talking to someone, we use grammer, expressions in our sentences but this capabilities are not available in computer (machine). So if you want that your computer also behave or give response like humans do then here we need to use NLP(Natural Language Processing). So NLP is the way use by computers so that they can text, speech, analyse, manipulate same like humans do.

Intent

Intent means specific function of your chatbot which is built for giving particular responses to the query of your customers or users of application.

Amazon Kendra

Amazon Kendra is the service provided by AWS which is used for creating our own search engine for our organization.

Benefits of Amazon Kendra

• Using Amazon Kendra, you can create your own customized search engine.
• You can’t able to put your organization docs and files in public sites just for searching a specific file, customer information, past deals, budget info, etc. So for such usecase you can create private search engine for your organization so that only employees of your organization can able to search for organization’s internal docs and files.

Amazon Code Guru

Using Amazon CodeGuru service, you can do profiling and review of your code for the optimization of code. This service automatically go deep dive into your code for profiling of code and provide you with details such as CPU leak, which specific line of code is taking more resources, etc. Also it gives you recommendations for doing some changes in the code so that you can optimize your code.

A python library needs to be installed to access the profiler of Code Guru : codeguru-profiler-agent

Amazon Comprehend

Amazon Comprehend is the service use for retrieving information from the text. From provided text, they will automatically find this particular word is name, this no. is mobile no., etc.

Amazon Textract

Amazon Texttract is the service which extract information from the scanned docs. Useful for automating manual things.

Amazon Translate

Amazon translate is the service for providing translation facility to your application.

Amazon Personalize do

Amazon Personalize is the service use for creating recommendation engine for your application.

Amazon Fraud Detector

Amazon Fraud Detector is the service use for detecting online frauds in our application.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —