AWS AI-ML VIRTUAL INTERNSHIP

An Internship Report submitted in

partial fulfillment of the requirements
for the award of the degree of

Bachelor of Technology
in
ARTIFICIAL INTELLEGENCE & DATA SCIENCE

by

KESAGANI ASHA JYOTHI

Reg. No: 21H71A5404

Offered by

APSCHE - EDUSKILLS-AWS ACADEMY

Department of Computer Science & Engineering

DVR & Dr. HS

MIC College of Technology
(Autonomous)
Kanchikacherla–521180, NTR Dist., Andhra Pradesh

December-2023

1
 CERTIFICATE

This is to certify that the Internship Report entitled “AWS Academy AI-ML

Virtual Internship” submitted by KESAGANI ASHA JYOTHI (Reg. No:

21H71A5404) to the DVR & Dr. HS MIC College of Technology in partial

fulfillment of the requirements for the award of the Degree of Bachelor of

Technology in ARTIFICIAL INTELLEGENCE AND DATA SCIENCE i s

a bonafide record of work.

Internship Coordinator Head of the Department

Examiner 1 Examiner 2

2
3
Certificate:

4
 ACKNOWLEDGEMENT

The satisfaction that accompanies the successful completion of any task would be incomplete
without the mention of the people who made it possible and whose constant guidance and
engagement crown all the efforts with success. I thank our college management and respected
Sri D. PANDURANGA RAO, CEO for providing us the necessary infrastructure to carry
out the Internship.

I express my sincere thanks to Dr. T. Vamsi Kiran, Director Academics and Dr. K.
Srinivas, Principal who has been a great source of inspiration and motivation for the
internship program.

I profoundly thank Mr .Dr. G. Sai chaitanya sir department of AI for permitting me to

carry out the internship.

I am thankful to the APSCHE and EduSkills for enabling me an opportunity to carry out the
internship in such a prestigious organization.

I am thankful to our Internship Coordinator Mr. Kalyan sir Associate Professor, Department
of AI for their internal support and professionalism who helped us in completing the
internship on time.

I take this opportunity to express our thanks to one and all who directly or indirectly helped
me in bringing this effort to present form.

Finally, my special thanks go to my family for their continuous support and help throughout
and for their continual support and encouragement for the completion of the Internship on
time.

5
 ABSTRACT
AWS ACADEMY AI-ML VIRTUAL INTERNSHIP helps bring innovation to life,
bridging the gap between business requirements and technology. This means automating
complex business processes, creating reports and dashboards, and training users on
Salesforce. For your team, it means helping them stay efficient and on top of things while
keeping an eye on their successes and customer relationships
This helps the organization to learn more about their customers and their requirements.
Providing better facilities and efficient approach to be in touch with customers and develop
their business simultaneously
Organization Information:
AWS ACADEMY is an Edu-Tech organization with a vision to bridge the gap between
academia & industry. Our outcome-based experiential learning programs on emerging
technologies (Internet of Things, Machine Learning, Data Science, Artificial Intelligence,
Robotics) are building skilled entry - level engineers, for the corporate world. AWS Academy
is in mission to build technology communities in academia to encourage students towards
innovation & entrepreneurship. Since inception, we have trained thousands of students,
faculty and working professionals on emerging technologies via technical bootcamps,
hackathons, Summer &Winter Internship Programs.

Following are AWS Academy program benefits

 Access to a ready-to-deliver AWS-developed curriculum.

 Regular course updates from AWS that reflect releases and best practices.
 Hands-on practical experience in the AWS Console.
 Complimentary AWS Professional learning for educators.
 Preparation for industry recognized AWS Certification to showcase knowledge for Educators
& Students.

Programs and opportunities:

This helps company to establish a healthy relationship between the customers and company or
institution. As the AWS AI-ML Internship mainly focuses on three domains Cloud concepts,
Artificial Intelligence, Machine Learning

6
 INDEX

S.NO CONTENTS

1. Introduction

2. Cloud Foundations

2.1 Cloud Concepts Overview

2.2Cloud Economics and Billing

2.3AWS Global Infrastructure Overview

2.4 AWS cloud Adaption Framework (CAF)

2.5 AWS Cloud Security

2.6 Networking & Content Delivery

2.7 Compute

2.8 Storage

2.9 Databases

2.10 Cloud Architecture

2.11 Auto Scaling and Monitoring

3. AWS Academy Machine Learning

3.1 welcome to AWS Academy Machine Learning Foundations

3.2 Introducing Machine Learning

3.3 Implementing a Machine Learning Pipeline with Amazon Sagemaker

3.4 Introducing Forecasting

3.5 Introducing Computer Vision

3.6 Introducing Natural Language Processing

4. Conclusion

7
 1. INTRODUCTION

Amazon Web Services (AWS) is the world’s most comprehensive and broadly adopted cloud
platform, offering over 200 fully featured services from data centers globally. Millions of
customers—including the fastest-growing startups, largest enterprises, and leading
government agencies—are using AWS to lower costs, become more agile, and innovate
faster.

AWS provides extensive and detailed documentation for each AWS service. Guides and
application programming interface (API) references are organized by service category. There
are also many general resources and tutorials that can be accessed from the AWS
Documentation pages.

The AWS Certified Cloud Practitioner certification provides individuals in various cloud and
technology roles with a way to validate their AWS Cloud knowledge and enhance their
professional credibility. This exam covers four domains, including cloud concepts, security,
technology, and billing and pricing.

What is AWS CLOUD FOUNDATIONS?

AWS cloud foundations consist of cloud concepts, cloud economics and billing, AWS Global
Infrastructure, AWS Cloud Security, Networking and content delivery, compute, storage.
databases, cloud architecture and Automatic scaling and monitoring.

 General Required Knowledge

 IT Technical knowledge

 IT Domain knowledge

 Preferred Knowledge

 Familiarity with cloud computing concepts

 Working knowledge of distributed systems

 Familiarity with general networking concepts

8
 What is AWS AI-ML?
In this course we learn
 Recognize how machine learning and deep learning are part of artificial intelligence
 Describe artificial intelligence and machine learning terminology
 Identify how machine learning can be used to solve a business problem
 Describe the machine learning process
 List the tools available to data scientists
 Identify when to use machine learning instead of traditional software development
methods
 Formulate a problem from a business request
 Obtain and secure data for machine learning
 Use Amazon SageMaker to build a Jupyter notebook
 Outline the process for evaluating data
 Explain why data must be preprocessed
 Use open-source tools to examine and preprocess data
 Use Amazon SageMaker to train and host a machine learning model
Finally, we will learn how to use managed Amazon ML services to solve specific machine

learning problems in forecasting, computer vision, and natural language processing.

Machine Learning Job Role

 Data Scientist role
 Machine Learning Engineer
 Applied Science Researcher
 Machine Learning Developer Role

2.Cloud Foundations

2.1 Cloud Concepts Overview

•Cloud computing is the on-demand delivery of IT resources via the internet with pay
as you go pricing.

9
•Cloud computing enables you to think of (and use) your infrastructure as software.
•There are three cloud service models: IaaS, PaaS, and SaaS.

•There are three cloud deployment models: cloud, hybrid, and on premises or private
cloud.

10
Advantages

•Trade capital expense for variable expense

•Benefit from massive economies of scale
•Stop spending money on running and maintaining data centers
•AWS is a secure cloud platform that offers a broad set of global cloud-based
products called services that are designed to work together.
•There are many categories of AWS services, and each category has many services
to choose from.
•Choose a service based on your business goals and technology requirements.

11
 This course focuses on some of the more common services in the following
service categories: compute, storage, database, networking and content
delivery, security, identity, and compliance, management and governance, and
AWS cost management.
 There are 3 ways to interact with AWS

2.2 Cloud Economics and Billing

•There is no charge (with some exceptions) for:

 Inbound data transfer
 Data transfer between services within the same AWS Region
 Pay for what you use
 No long term contracts are required.

12
 There are several free AWS services, including:
•Amazon VPC
•Elastic Beanstalk
•AWS Cloud Formation
•IAM
•Automatic scaling services
•Consolidated Billing
While the services themselves are free, the resources that they provision might not be free. In
most cases, there is no charge for inbound data transfer or for data transfer between other
AWS services within the same
The AWS Pricing Calculator helps you:
o Estimate monthly costs of AWS services
o Identify opportunities for cost reduction
o Model your solutions before building them
o Explore price points and calculations behind your estimate
o Find available instance types and contract terms that meet our need

2.3 Aws Global Infrastructure Overview

•The AWS Global Infrastructure consists of Regions and Availability Zones.

•Your choice of a Region is typically based on compliance requirements or to reduce latency.
•Each Availability Zone is physically separate from other Availability Zones and has
redundant power, networking, and connectivity.
13
•Edge locations, and regional edge caches improve performance by
Caching content closer to users

AWS Regions:
An AWS Region is a physical geographical location with one or more Availability Zones.
Availability Zones in turn consist of one or more data centers.
Availability Zones:
Each AWS Region has multiple, isolated locations that are known as Availability Zones.
Each Availability Zone provides the ability to operate applications and databases that are
more highly available, fault-tolerant, and scalable than would be possible with a single data
center. Each Availability Zone can include multiple data centers.

AWS data centers:

AWS data centers are designed for security. Data centers are where the data resides and data
processing occurs. Each data center has redundant power, networking, and connectivity, and
is housed in a separate facility. A data center typically has 50,000 to 80,000 physical servers.

2.4 AWS Cloud Adoption Framework (AWS CAF)

AWS CAF provides guidance and best practices to help organizations build a comprehensive
approach to cloud computing across the organization and
throughout the lifecycle to accelerate successful cloud adoption.
AWS CAF is organized into six perspectives.

14
 Perspectives consist of sets of capabilities.
Six core perspectives

Focus on business Focus on technical

Capabilities Capabilities

2.5. Aws Cloud Security

Security is the highest priority at Amazon Web Services (AWS). AWS delivers a scalable
cloud computing environment that is designed for high availability and dependability, while
providing the tools that enable you to run a wide range of applications.

o AWS is responsible for protecting the infrastructure that runs all the services that are offered
15
 in the AWS Cloud.
o The customer is responsible for the encryption of data at rest and data in transit. The customer
should also ensure that the network is configured for security and that security credentials and
logins are managed safely.

AWS IDENTITY AND ACCESS MANAGEMENT(IAM):

AWS Identity and Access Management (IAM) allows you to control access compute,
storage, database, and application services in the AWS Cloud. IAM can be used to handle
authentication, and to specify and enforce authorization policies so that you can specify which
users can access which services.
 IAM is a tool that centrally manages access to launching, configuring, managing, and
terminating resources in your AWS account
 With IAM, you can manage which resources can be accessed by who, and how these
resources can be accessed. You can grant different permissions to different people for
different resources. For example, you might allow some useful access to Amazon EC2,
Amazon S3, Amazon DynamoDB, Amazon Redshift, and other AWS services.

 An IAM user is a person or application that is defined in an AWS account, and that must
make API calls to AWS products. Each user must have a unique name (with no spaces in the
name) with in the AWS account, and a set of security credentials that is not shared with other
users. These credentials are different from the AWS account root user security credentials.
Each user is defined in one and only one AWS account.

16
 An IAM group is a collection of IAM users. You can use IAM groups to simplify specifying
and managing permissions for multiple users.

 An IAM policy is a document that defines permissions to determine what users can do in the
AWS account. A policy typically grants access to specific resources and specifies what the
user can do with those resources. Policies can also explicitly deny access.

 An IAM roles a tool for granting temporary access to specific AWS resources in an AWS
account.

IAM: Authorization
o Assign permissions by creating an IAM policy
o Permissions determine which resources and operations are allowed

IAM policies:
o An IAM policy is a document that defines permissions
o Two types of policies –identity-based and resource-based

Securing a new AWS account:

o Stop using the account root user as soon as possible
o Enable multi-factor authentication (MFA)
o Use AWS CloudTrail
o Enable a billing report, such as the AWS Cost and Usage Report

2.6 Networking and Content Delivery:

A computer network is two or more client machines that are connected together to share
resources. A network can be logically partitioned into subnets. Networking requires a
networking device (such as a router or switch) to connect all the clients together and enable
communication between them. Each client machine in a network has a unique Internet
Protocol (IP) address that identifies it.
IPv4 (32-bit) address: 192.0.2.0
IPv6 (128-bit) address: 2600:1f18:22ba:8c00:ba86:a05e:a5ba:00FF

17
 Amazon VPC
It enables us to provision a logically isolated section of the AWS Cloud where
you can launch AWS resources in a virtual network that we define.
Gives you control over your virtual networking resources, including:
•Selection of IP address range
•Creation of subnets
•Configuration of route tables and network gateways
•A VPC belongs to one Region and requires a CIDR block.
•A VPC is subdivided into subnets.
•A subnet belongs to one Availability Zone and requires a CIDR block.

Subnets:
Range of IP addresses that divide a VPC
•Each instance in your VPC has a default network interface that is assigned a private IPv4
address from the IPv4 address range of your VPC.
•A route table contains a set of rules (or routes) that you can configure to direct network
traffic from your subnet
•Each route specifies a destination and a target
•Route tables control traffic for a subnet
•Route tables have a built-in local route.

There are several VPC networking options, which include:

•Internet gateway
•NAT gateway
•VPC endpoint
•VPC peering
•VPC sharing

VPC security
You can build security into your VPC architecture in several ways so that you have complete
control over both incoming and outgoing traffic.
•Build security into your VPC architecture:
o Isolate subnets if possible
o Choose the appropriate gateway device or VPN connection for your needs.
18
o Use firewalls.
o Security groups and network ACLs are firewall options that you can use to secure your VPC

2.7Compute
 Amazon EC2 enables you to run Windows and Linux virtual machines in the cloud.
 You can choose from many You launch EC2instancesfrom an AMI template into a VPC in
your account.
 You can configure security groups to control access to instances (specify allowed ports and
source).

19
 You can use Amazon Cloud Watch to capture and review metrics on EC2 instances.
 Amazon EC2 pricing models include On-Demand Instances, Reserved Instances, Spot
Instances, Dedicated Instances, and Dedicated Hosts. Per second billing is available for On-
Demand Instances, Reserved Instances, and Spot Instances that use only Amazon Linux and
Ubuntu.
 Spot Instances can be interrupted with a 2-minute notification. However, they can offer
significant cost savings over On-Demand Instances.
 The four pillars of cost optimization are–
•Right size
•Increase elasticity
•Optimal pricing model
•Optimize storage choices
 Containers can hold everything that an application needs to run.
 Docker is a software platform that packages software into containers.
 A single application can span multiple containers.
 Amazon Elastic Container Service (Amazon ECS) orchestrates the running of Docker
containers.
 Kubernetes is open-source software for container orchestration.
 Amazon Elastic Kubernetes Service (Amazon EKS) enables you to run Kubernetes on AWS
 Amazon Elastic Container Registry (Amazon ECR) enables you to store, manage, and deploy
your Docker containers.

20
 Serverless computing enables you to build and run applications and services without
provisioning or managing servers.
 AWS Lambda is a serverless compute service that provides built-in fault tolerance and
automatic scaling.
 An event source is an AWS service or developer-created application that triggers a Lambda
function to run.
 The maximum memory allocation for a single Lambda function is 10,240 MB.
 The maximum run time for a Lambda function is 15 minutes
 AWS Elastic Beanstalk enhances developer productivity.
•Simplifies the process of deploying your application.
•Reduces management complexity.
 Elastic Beanstalk supports Java, .NET, PHP, Node.js, Python, Ruby, Go, and Docker.
 There is no charge for Elastic Beanstalk. Pay only for the AWS resources you use.

2.8 Storage
 Amazon EBS features:
•Persistent and customizable block storage for Amazon EC2
•HDD and SSD types
•Replicated in the same Availability Zone
•Easy and transparent encryption
•Elastic volumes
•Back up by using snapshots
 Amazon S3 is a fully managed cloud storage service.

 You can store a virtually unlimited number of objects.

21
 You pay for only what you use.
 You can access Amazon S3 at anytime from anywhere through a URL.
 Amazon S3 offers rich security controls.
 Amazon EFS provides file storage over a network.
 Perfect for big data and analytics, media processing workflows, content management, web
serving, and home directories.
 Fully managed service that eliminates storage administration tasks.
 Accessible from the console, an API, or the CLI.
 Scales up or down as files are added or removed and you pay for what you use.
 Amazon S3 Glacier is a data archiving service that is designed for security, durability, and an
extremely low cost.
 Amazon S3 Glacier pricing is based on region.
 It is extremely low-cost design works well for long-term archiving.
 The service is designed to provide 11 9s of durability for objects.

2.9 Databases

 With Amazon RDS, you can set up, operate, and scale relational databases in the cloud.

 Features –

•Managed service

•Accessible via the console, AWS Command Line Interface (AWS CLI), or
application programming interface (API) calls

•Scalable (compute and storage)

•Automated redundancy and backup are available

•Supported database engines:

•Amazon Aurora, PostgreSQL, MySQL, MariaDB, Oracle, Microsoft SQL

Server.

 Amazon DynamoDB:

 Runs exclusively on SSDs.

 Supports document and key-value store models.

 Replicates your tables automatically across your choice of AWS Regions.

22
  Works well for mobile, web, gaming, adtech, and Internet of Things (IoT) applications.

 Is accessible via the console, the AWS CLI, and API calls.

 Provides consistent, single-digit millisecond latency at any scale.

 Has no limits on table size or throughput.

 Amazon Redshift features:

o Fast, fully managed data warehouse service

o Easily scale with no downtime

o Columnar storage and parallel processing architectures

o Automatically and continuously monitors cluster

 Amazon Aurora features:

o High performance and scalability

o High availability and durability

o Multiple levels of security

o Compatible with MySQL and PostgreSQL

2.10 Cloud Architecture

 The AWS Well-Architected Framework provides a consistent approach to evaluate cloud

architectures and guidance to help implement designs.

 The AWS Well-Architected Framework documents a set of design principles and best
practices that enable you to understand if a specific architecture aligns well with cloud best
practices.

 The AWS Well-Architected Framework is organized into six pillars.

 Each pillar includes its own set of design principles and best practices.

 Reliability is a measure of your system’s ability to provide functionality when desired by the
user, and it can be measured in terms of MTBF.

23
 Availability is the percentage of time that a system is operating normally or correctly
performing the operations expected of it (or normal operation time over total time).

 Three factors that influence the availability of your applications are fault tolerance,
scalability, and recoverability.

 You can design your workloads and applications to be highly available, but there is a cost
tradeoff to consider.

 AWS Trusted Advisor is an online tool that provides real-time guidance to help you provision
your resources by following AWS best practices.

 AWS Trusted Advisor looks at your entire AWS environment and gives you real-time
recommendations in five categories.

 You can use AWS Trusted Advisor to help you optimize your AWS environment as soon as
you start implementing your architecture designs.

2.11 Auto Scaling and Monitoring

 Elastic Load Balancing distributes incoming application or network traffic across multiple
targets (such as Amazon EC2 instances, containers, IP addresses, and Lambda functions) in
one or more Availability Zones.

 Elastic Load Balancing supports three types of load balancers:

•Application Load Balancer

•Network Load Balancer

•Classic Load Balancer

 Elastic Load Balancing offers several monitoring tools for continuous monitoring and logging
for auditing and analytics.

 Amazon CloudWatch helps you monitor your AWS resources—and the applications that you
run on AWS—in real time.

 CloudWatch enables you to –

•Collect and track standard and custom metrics.

•Set alarms to automatically send notifications to SNS topics or perform Amazon EC2 Auto
24
 Scaling or Amazon EC2 actions based on the value of the metric or expression relative to a
threshold over a number of time periods.

•Define rules that match changes in your AWS environment and route these events to targets
for processing.

 Scaling enables you to respond quickly to changes in resource needs.

 Amazon EC2 Auto Scaling helps you maintain application availability, and enables you to
automatically add or remove EC2 instances according to your workloads.

AWS Academy Machine Learning

3.1 Welcome to AWS Academy Machine Learning.

Data scientist role:

 Applying knowledge of statistics and analytical skills to interpret data

 Data scientists often have degrees in statistics, computer science, or economics

Machine learning engineer:

 Emphasis on programming and system design skills
 Often have background as a developer or software architect.
Applied science researcher:
 Applies machine learning technology to a specific domain
 Requires knowledge of both the domain and machine learning
Machine learning developer role:
 Integrating machine learning with software applications
 Requires strong application development skills and machine learning knowledge

3.2 INTRODUCING MACHINE LEARNING

 Artificial intelligence is the broad field of building machines to perform human tasks.

 Machine learning is a subset of AI. It focuses on using data to train ML models so the
models can make predictions.

 Deep learning is a technique that was inspired from human biology. It uses layers of
neurons to build networks that solve problems.

25
 Advancements in technology, cloud computing, and algorithm development have led to a rise
in machine learning capabilities and applications.

 Machine learning applications are already part of your everyday life.

 Machine learning problems can be grouped into –

Supervised learning: You have training data for which you know the answer.

Unsupervised learning: You have data, but you are looking for insights within the data.

Reinforcement learning: The model learns in a way that is based on experience and
feedback

 The machine learning pipeline process can guide you through the process of training and
evaluating a model.

 The iterative process can be broken into three broad steps

Data processing
Model Training
Model evaluation

Mode Predi
Data
l ction
 Python is the most popular language for performing machine learning tasks.
 Jupyter Notebooks provide you with a web browser hosted development environment for
machine learning.
 A large number of open-source tools exist, and only a few key ones are introduced in this
course.
 Depending on your requirements, you can start with the low-level frameworks and use your
own solution. You might use tools like Amazon SageMaker to help with larger tasks, or adapt
one of the managed ML services for your specific problem domain.
 You will face many machine learning challenges. The biggest problems that you directly
influence are related to data, but you will also deal with people, business and technology
challenges.
 Consider managed services to help you solve machine learning problems.

26
 The machine learning pipeline describes the different stages for developing a machine
learning application.
 The machine learning stack has three key layers.

 API services, machine learning services, and machine learning frameworks

 Machine learning development is different from traditional development.

 Training algorithms are applied to data to create a model for making predictions.

 Machine learning is a subset of artificial intelligence.

 Machine learning applies learning algorithms to develop models from large datasets.

3.3 Implementing a Machine Learning Pipeline with Amazon

SageMaker.
 Business problems must be converted into an ML problem.
 The first step in solving machine learning problems is to obtain data that is required to train
your machine learning model.
 ETL can be used to obtain data from multiple sources.
 Services such as AWS Glue can make it easy to obtain data from multiple data stores.
 Security requirements should be understood, and they should be based on both business need
and any regulatory requirements. You want to ensure that your data is secure, only authorized
users can access it, and it is encrypted where possible.
 The first step is to get your data into a format that can be used easily.
 Pandas is a popular and useful Python library for working with data.
 Descriptive statistics help you gain insights into the data.
 Use visualizations to examine the dataset in more detail.
 Feature engineering involves selecting or extracting the best features for machine learning.
 Preprocessing gives you better data to work with. Better data typically provides better results.
 Preprocessing has two categories:
•Converting data (to numerical values)
•Cleaning up dirty data (removing missing data, cleaning outliers)
 Develop a strategy for dirty data. Replace or delete rows with missing data.
 Delete, transform, or impute new values for outliers.
 Split data into training, testing and validation sets to help you validate the model accuracy
 Can use K-fold cross validation can help with smaller datasets
 Can use two key algorithms for supervised learning—XGBoost and linear learner
 Use k-means for unsupervised learning

27
 Use Amazon SageMaker training jobs to train models
 You can deploy your trained model by using Amazon SageMaker to handle API calls from
applications, or to perform predictions by using a batch transformation.
 The goal of your model is to generate predictions to answer the business problem. Be sure that
your model can generate good results before you deploy it to production.
 Use Single-model endpoints for simple use cases and use multi-model endpoint support to
save resources when you have multiple models to deploy.
 To evaluate the model, you must have data that the model hasn’t seen, through either a hold-
out set or by using k-fold cross validation.
 Classification can use confusion matrix, and the AUC-ROC that can be generated from it.
 Regression can use mean squared.

 Model tuning is important to find the best solution to your business problem.

 Hyperparameters can be tuned for the model, optimizer, and data.

 Amazon SageMaker can perform automatic hyperparameter tuning.

 Overall model development can be accelerated by using Autopilot.

3.4 Introduction to Forecasting.

 Timeseries data is sequenced data that includes a time element, which makes it different from
regular datasets

 Some of the time challenges include–

 Handling different time formats

28
 Handling missing data through down sampling, up sampling and smoothing

 Handling seasonality, such as weekdays and yearly cycles

 Avoiding bad correlations

 The pandas library offers support for time series data through functions that deal with time

 With Amazon Forecast, you can choose between five algorithms–

ARIMA

DeepAR+

ETS

NPTS

 You can use Amazon Forecast to train and use a model for time series data

 There are specific schemas defined for domains such as retail and EC2 capacity planning, or
you can use a custom schema
 You need to supply at least the time series data, but can also provide metadata and related
data to add move information to the model

 As with most supervised machine learning problems, your data is split into training and
testing data, but this split takes into account the time element

 Use RMSE and wQuantileLoss metrics to evaluate the efficiency of the model

3.5 Introduction to computer vision.

 Computer vision is the automated extraction of information from images

29
 From a practical point of view, you can divide computer vision into two distinct areas—image
analysis and video analysis

 Image analysis includes object classification, detection, and segmentation

 Video analysis includes instance tracking, action recognition, and motion estimation.

 Amazon Recognition is a computer vision service that is based on deep learning. You can
easily add image and video analysis to your applications,

 Amazon Recognition provides image and video detection of faces, sentiment, text, unsafe
content, and library search.

 Amazon Recognition is integrated into other AWS services.

 Models must be trained for the specific domain that you want to analyze

 You can set custom labeling for the specific business case

 Custom labeling workflow

 You must label images and create bounding boxes for objects

 You can use Amazon SageMaker Ground Truth to build training datasets for your models

3.6 Introducing Natural Language Processing

 As a domain, NLP predates machine learning
 NLP development maps directly to the ML development process.
 Some of the main use cases for NLP are search query analysis, human-machine interaction,
and market or social research.
 NLP is difficult because of the imprecise nature of human language.
 Amazon Transcribe can automatically convert spoken language to text

30
 Amazon Polly can convert written text to spoken language

 Amazon Translate can create real-time translation between languages

 Amazon Comprehend automates many of the NLP use cases that are reviewed in this module

 Amazon Lex can create a human-like interface to your applications.

31
4.Conclusion
AWS AI ML described how model explainability relates to AI/ML solutions,
giving customers insight to explainability requirements when initiating AI/ML use cases. To
help convey how these model explainability options relate to real-world scenarios, examples
from a range of industries were demonstrated.
o It is recommended that AI/ML owners or business leaders follow these steps when initiating a
new AI/ML solution:
o Collect business requirements to identify the level of explainability required for your business
to accept the solution.
o Based on business requirements, implement an assessment for model explainability.
o Work with an AI/ML technician to communicate model explainability assessment and find
the optimal AI/ML solution to meet your business objectives.
o After the solution is completed, revisit the model explainability assessment to evaluate that
business requirements are continuously met.
o By taking these steps, you will mitigate regulation risks and ensure trust in your model.
 Machine Learning on AWS. Get deeper insights from your data while lowering costs with
AWS machine learning (ML). AWS helps you at every stage of your ML adoption journey
with the most comprehensive set of artificial intelligence (AI) and ML services, infrastructure,
and implementation resources.

32