Machine Learning on AWS

1. Amazon Rekognition

Find objects, people, text and scenes in images and videos using ML.

• Facial analysis and facial search to do user verification and people counting.
• Create a database of “familiar faces”, or compare against celebrities.

Use cases from the slide:

• Labeling
• Content Moderation
• Text Detection
• Face Detection and Analysis (gender, age range, emotions…)
• Face Search and Verification
• Celebrity Recognition
• Pathing (e.g. for sports game analysis)

Exam tip

If the input is an image or a video, the answer is almost always Rekognition. Content moderation and celebrity recognition are the two use cases that appear most often as distractors elsewhere.

2. Amazon Transcribe

Automatically convert speech to text.

• Uses a deep learning process called automatic speech recognition (ASR) to convert speech to text quickly and accurately.
• Automatically removes Personally Identifiable Information (PII) using Redaction.
• Supports Automatic Language Identification for multi-lingual audio.

Use cases:

• Transcribe customer service calls.
• Automate closed captioning and subtitling.
• Generate metadata for media assets to create a fully searchable archive.

Real-world example

A contact centre records thousands of calls a day. Transcribe converts them to text with PII redaction turned on, so analysts can search the transcripts without ever seeing a customer’s card number.

3. Amazon Polly

Turn text into lifelike speech using deep learning, allowing you to create applications that talk.

Exam tip

Polly and Transcribe are mirror images and are constantly swapped in questions: Polly = text to audio, Transcribe = audio to text.

4. Amazon Translate

• Natural and accurate language translation.
• Allows you to localize content — such as websites and applications — for international users, and to easily translate large volumes of text efficiently.

Real-world example

A product catalogue published in one language is run through Translate to produce the other market locales, rather than hand-translating every new item description.

5. Amazon Lex & Amazon Connect

Amazon Lex

Built on the same technology that powers Alexa:

• Automatic Speech Recognition (ASR) to convert speech to text.
• Natural Language Understanding to recognize the intent of text and of callers.
• Helps build chatbots and call center bots.

Amazon Connect

• Receive calls, create contact flows — a cloud-based virtual contact center.
• Can integrate with other CRM systems or AWS.
• No upfront payments, 80% cheaper than traditional contact center solutions.

The slide’s flow ties them together: a phone call arrives at Connect, which streams it to Lex; Lex recognizes the intent (“Schedule an Appointment”) and invokes a Lambda function, which schedules the appointment in the CRM.

Exam tip

Connect is the contact center (it takes the call); Lex is the bot brain (it understands what the caller wants). A question about building a chatbot is Lex; a question about replacing a call center is Connect.

6. Amazon Comprehend

For Natural Language Processing – NLP. It is a fully managed and serverless service that uses machine learning to find insights and relationships in text:

• The language of the text.
• Extracts key phrases, places, people, brands or events.
• Understands how positive or negative the text is.
• Analyzes text using tokenization and parts of speech.
• Automatically organizes a collection of text files by topic.

Sample use cases:

• Analyze customer interactions (emails) to find what leads to a positive or negative experience.
• Create and group articles by topics that Comprehend will uncover.

Real-world example

Feed a month of support emails to Comprehend: it tells you which are angry, pulls out the product names being complained about, and clusters the whole set by topic without anyone labelling the data first.

7. Amazon SageMaker AI

A fully managed service for developers / data scientists to build ML models. Typically it is difficult to do all the processes in one place, on top of having to provision servers — SageMaker AI removes both problems.

The deck illustrates the simplified machine learning process by predicting your exam score:

1. Historical data with labels — number of years of experience in IT, number of years of experience with AWS, time spent on the course… each row labelled with a score such as 670, 890 or 934.
2. Build an ML model, then train and tune it.
3. Apply the model to new data to get a prediction — “PASS WITH 906”.

Exam tip

SageMaker AI is the odd one out in this chapter. Every other service is a pre-trained, ready-to-call AI service; SageMaker AI is where you build, train and deploy your own model. If a question says “our data scientists need to build a custom model”, it is SageMaker AI.

8. Amazon Kendra

A fully managed document search service powered by Machine Learning.

• Extracts answers from within a document — text, PDF, HTML, PowerPoint, MS Word, FAQs…
• Natural language search capabilities.
• Learns from user interactions/feedback to promote preferred results — Incremental Learning.
• Ability to manually fine-tune search results (importance of data, freshness, custom…).

Data sources shown: Amazon S3, Amazon RDS, Google Drive, MS SharePoint, MS OneDrive, 3rd party, APNs, Custom. Kendra indexes them into a Knowledge Index powered by ML, so a user asking “Where is the IT support desk?” gets back “1st floor”.

Real-world example

An internal document search that answers a plain-English question directly out of an HR PDF, instead of returning a list of ten files for the employee to read.

9. Amazon Personalize

A fully managed ML service to build apps with real-time personalized recommendations.

• Examples: personalized product recommendations / re-ranking, and customized direct marketing.
• Example scenario: a user bought gardening tools, so provide recommendations on the next one to buy.
• Uses the same technology used by Amazon.com.
• Integrates into existing websites, applications, SMS, email marketing systems…
• Implement in days, not months — you don’t need to build, train and deploy ML solutions yourself.
• Use cases: retail stores, media and entertainment…

The flow on the slide: Amazon Personalize reads data from Amazon S3 and takes real-time data integration through the Amazon Personalize API, then exposes a customized personalized API to websites and apps, mobile apps, SMS and emails.

Exam tip

“Recommendations” is the keyword for Personalize. Do not confuse it with SageMaker AI — Personalize ships a working recommendation engine, SageMaker AI is where you would build one from scratch.

10. Amazon Textract

Automatically extracts text, handwriting and data from any scanned documents using AI and ML.

• Extract data from forms and tables.
• Read and process any type of document — PDFs, images…

Use cases:

• Financial Services — e.g. invoices, financial reports.
• Healthcare — e.g. medical records, insurance claims.
• Public Sector — e.g. tax forms, ID documents, passports.

The slide shows a scanned ID being analyzed into structured JSON with fields such as Document ID, Name, SEX and DOB.

Exam tip

Textract reads scanned documents and handwriting into structured data; Rekognition’s Text Detection finds text in images and videos; Kendra searches documents. The word “forms”, “tables” or “handwriting” points at Textract.

11. Section summary

The deck closes with a one-line identity for each service — this list is the fastest revision for the exam:

Service	One-line identity
Rekognition	Face detection, labeling, celebrity recognition
Transcribe	Audio to text (e.g. subtitles)
Polly	Text to audio
Translate	Translations
Lex	Build conversational bots – chatbots
Connect	Cloud contact center
Comprehend	Natural language processing
SageMaker	Machine learning for every developer and data scientist
Kendra	ML-powered search engine
Personalize	Real-time personalized recommendations
Textract	Detect text and data in documents

Quick recap

Concept	What to remember for the exam
Amazon Rekognition	Objects, people, text and scenes in images and videos; labeling, content moderation, text detection, face detection and analysis, face search and verification, celebrity recognition, pathing
Amazon Transcribe	Speech to text via ASR; PII Redaction; Automatic Language Identification; call transcription, closed captioning, searchable media metadata
Amazon Polly	Text to lifelike speech with deep learning — applications that talk
Amazon Translate	Natural, accurate translation; localize websites and applications, translate large volumes efficiently
Amazon Lex	Same technology as Alexa; ASR plus Natural Language Understanding to recognize intent; chatbots and call center bots
Amazon Connect	Cloud-based virtual contact center; receive calls, contact flows, CRM/AWS integration; no upfront payments, 80% cheaper than traditional solutions
Amazon Comprehend	Serverless NLP: language, key phrases, places, people, brands, events, sentiment, tokenization and parts of speech, topic organization
Amazon SageMaker AI	Build, train, tune and apply your own ML models — the only build-it-yourself service in the chapter
Amazon Kendra	ML document search with natural language queries and incremental learning; indexes S3, RDS, Google Drive, SharePoint, OneDrive and more
Amazon Personalize	Real-time personalized recommendations, same technology as Amazon.com; reads from S3, exposes a custom API; days not months
Amazon Textract	Text, handwriting and data from scanned documents, forms and tables; financial services, healthcare, public sector