{"id":195,"date":"2024-09-12T19:50:19","date_gmt":"2024-09-12T19:50:19","guid":{"rendered":"https:\/\/blog.codeandtech.com\/blog\/?p=195"},"modified":"2024-09-13T13:46:17","modified_gmt":"2024-09-13T13:46:17","slug":"aws-certified-ai-practitioner-tips","status":"publish","type":"post","link":"https:\/\/blog.codeandtech.com\/blog\/2024\/09\/12\/aws-certified-ai-practitioner-tips\/","title":{"rendered":"AWS Certified AI Practitioner &#8211; Tips"},"content":{"rendered":"\n<p><\/p>\n\n\n\n<p><\/p>\n\n\n\n<p>Howdy! I was able to crack <a href=\"https:\/\/aws.amazon.com\/certification\/certified-ai-practitioner\/\">AWS Certified AI Practitioner Certification | AWS Certification | AWS (amazon.com)<\/a> recently.<\/p>\n\n\n\n<p><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Udemy course by Stephane Maarek &#8211; <a href=\"https:\/\/www.udemy.com\/course\/aws-ai-practitioner-certified\">https:\/\/www.udemy.com\/course\/aws-ai-practitioner-certified<\/a><\/li>\n\n\n\n<li>AWS Skill Builder course for AP Practitioner &#8211; <a href=\"https:\/\/explore.skillbuilder.aws\/learn\/course\/internal\/view\/elearning\/19554\/exam-prep-standard-course-aws-certified-ai-practitioner-aif-c01\">https:\/\/explore.skillbuilder.aws\/learn\/public\/learning_plan\/view\/2194\/enhanced-exam-prep-plan-aws-certified-ai-practitioner-aif-c01<\/a><\/li>\n\n\n\n<li>AWS Documentation\n<ul class=\"wp-block-list\">\n<li>Nothing comes closer to AWS Overview and FAQs<\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/sagemaker\/faqs\">https:\/\/aws.amazon.com\/sagemaker\/faqs<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/q\/faqs\/\">https:\/\/aws.amazon.com\/q\/faqs\/<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/polly\/faqs\/\">https:\/\/aws.amazon.com\/polly\/faqs\/<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/transcribe\/faqs\">https:\/\/aws.amazon.com\/transcribe\/faqs<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/textract\/faqs\/\">Amazon Textract FAQs | AWS<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/kendra\/faqs\/\">Amazon Kendra FAQs &#8211; Amazon Web Services<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/personalize\/faqs\/\">Amazon Personalize FAQs &#8211; Amazon Web Services<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/rekognition\/faqs\">https:\/\/aws.amazon.com\/rekognition\/faqs<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/lex\">https:\/\/aws.amazon.com\/lex<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/comprehend\/faqs\/\">Amazon Comprehend \u2013 FAQs<\/a><\/li>\n\n\n\n<li>&#8211; <a href=\"https:\/\/aws.amazon.com\/translate\/faqs\/\">AWS Translate FAQs \u2013 Amazon Web Services (AWS)<\/a><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p><\/p>\n\n\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\"><summary>AWS Services Notes<\/summary>\n<p><html><\/p>\n<p><head><\/p>\n<style>\n        \/* table td {<br \/>\n                border-top: thin solid;<br \/>\n                border-left: thin solid;<br \/>\n                border-right: thin solid;<br \/>\n                border-bottom: thin solid;<br \/>\n            }<\/p>\n<p>            table td:first-child {<br \/>\n                border-left: thin solid;<br \/>\n            }<\/p>\n<p>            table td:last-child {<br \/>\n                border-right: thin solid;<br \/>\n            } *\/<br \/>\n        table.stats4 {<br \/>\n            text-align: center;<br \/>\n            font-family: Verdana, Geneva, Arial, Helvetica, sans-serif;<br \/>\n            font-weight: normal;<br \/>\n            font-size: 12px;<br \/>\n            color: #fff;<br \/>\n            background-color: #666;<br \/>\n            border: 0px;<br \/>\n            border-collapse: collapse;<br \/>\n            border-spacing: 7px;<br \/>\n        }<\/p>\n<p>        table.stats4 td {<br \/>\n            background-color: #c9c4c4;<br \/>\n            color: #000;<br \/>\n            padding: 5px 7px 5px 7px;<br \/>\n            text-align: left;<br \/>\n            border: 1px #fff solid;<br \/>\n        }<\/p>\n<p>        table.stats4 tr {<br \/>\n            background-color: #f19a9a;<br \/>\n            color: #000;<br \/>\n            padding: 5px 7px 5px 7px;<br \/>\n            text-align: left;<br \/>\n            border: 1px #fff solid;<br \/>\n        }<\/p>\n<p>        table.stats4 td.hed {<br \/>\n            background-color: #666;<br \/>\n            color: #fff;<br \/>\n            text-align: left;<br \/>\n            border: 2px #fff solid;<br \/>\n            font-size: 12px;<br \/>\n            font-weight: bold;<br \/>\n        }<\/p>\n<p>        tr.trHeader {<br \/>\n            background-color: #9f0404;<br \/>\n            color: #fff;<br \/>\n            text-align: left;<br \/>\n            border: 2px #fff solid;<br \/>\n            font-size: 12px;<br \/>\n            font-weight: bold;<br \/>\n        }<br \/>\n    <\/style>\n<p><\/head><\/p>\n<p><body><\/p>\n<div>\n<table class=\"stats4\">\n<thead><\/thead>\n<tbody>\n<tr class=\"trHeader\">\n<td>1<\/td>\n<td>SNo<\/p>\n<\/td>\n<td>Service<\/p>\n<\/td>\n<td>Details<\/p>\n<\/td>\n<td>Comments<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>2<\/td>\n<td>1<\/p>\n<\/td>\n<td>SageMaker<\/p>\n<\/td>\n<td>prepare data and build, train, and deploy machine learning (ML) models<\/p>\n<\/td>\n<td>End to end Managed service<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>3<\/td>\n<td>\n<\/td>\n<td>SageMaker Studio<\/p>\n<\/td>\n<td>single, web-based visual interface to perform all ML development steps<\/p>\n<\/td>\n<td>prepare data and build, train, and deploy model, upload data, create new notebooks, train and<br \/>\n                        tune models, move back and forth between steps to adjust experiments, compare results, and<br \/>\n                        deploy models to production <\/p>\n<p> All ML development activities including notebooks,<br \/>\n                        experiment management, automatic model creation, debugging and profiling, and model drift<br \/>\n                        detection can be performed within the unified SageMaker Studio visual interface.<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>4<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>5<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>6<\/td>\n<td>\n<\/td>\n<td>SageMaker Data Wrangler<\/p>\n<\/td>\n<td>For data preparation, transformation and feature engineering <br \/> Prep tabular and image data for<br \/>\n                        ML <br \/> Single interface for data selection, cleansing, exploration, visualization and<br \/>\n                        processing <br \/> Sql support and Data Quality tool<\/p>\n<\/td>\n<td>Use case &#8211; music dataset, song ratings, listening duration<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>7<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>8<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>9<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>10<\/td>\n<td>\n<\/td>\n<td>SageMaker Canvas<\/p>\n<\/td>\n<td>No code interface <br \/> Build\/tune\/train model using a visual interface <br \/> Build your own custom<br \/>\n                        model using automl <br \/> Leverage data wrangler<\/p>\n<\/td>\n<td>visual drag-and-drop service that allows business analysts to build ML models and generate<br \/>\n                        accurate predictions without writing any code or requiring ML expertise. <\/p>\n<p> Use case<br \/>\n                        Sentiment analysis<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>11<\/td>\n<td>\n<\/td>\n<td>SageMaker Clarify<\/p>\n<\/td>\n<td>For data preparation. <br \/> Evaluate foundation models &#8211; compare Model A vs Model B <br \/> Evaluate<br \/>\n                        using human factors <br \/> Use built in datasets or bring your own dataset <br \/> Built inn metrics<br \/>\n                        and algorithms <\/p>\n<p> Model Explainability &#8211; debug predictions. To increase the trust and<br \/>\n                        understanding of the model<\/p>\n<\/td>\n<td>To identify potential bias <br \/> Bring your own employee or aws employee <\/p>\n<p> Detect Bias<br \/>\n                        (human) <br \/> Specify input features and bias will be automatically detected<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>12<\/td>\n<td>\n<\/td>\n<td>SageMaker Feature Store<\/p>\n<\/td>\n<td>Store, share and manage features of ML models<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>13<\/td>\n<td>\n<\/td>\n<td>SageMaker Ground Truth, <br \/> SageMaker Ground Truth Plus<\/p>\n<\/td>\n<td>Labeling <br \/> For RLHF &#8211; reinforcement learning from human feedback <br \/> Model review<br \/>\n                        customization and evaluation<\/p>\n<\/td>\n<td>identify raw data, such as images, text files, and videos, and add informative labels to create<br \/>\n                        high-quality training datasets for your ML model<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>14<\/td>\n<td>\n<\/td>\n<td>SageMaker Studio Notebooks<\/p>\n<\/td>\n<td>Jupyter notebooks in SageMaker for the complete ML development<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>15<\/td>\n<td>\n<\/td>\n<td>SageMaker Studio Lab<\/p>\n<\/td>\n<td>ML development environment<\/p>\n<\/td>\n<td>that provides the compute, storage (up to 15 GB), and security<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>16<\/td>\n<td>\n<\/td>\n<td>SageMaker HyperPod<\/p>\n<\/td>\n<td>Train models<\/p>\n<\/td>\n<td>purpose-built to accelerate foundation model (FM) training<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>17<\/td>\n<td>\n<\/td>\n<td>SageMaker Experiments<\/p>\n<\/td>\n<td>organize and track iterations to ML models<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>18<\/td>\n<td>\n<\/td>\n<td>SageMaker Debugger<\/p>\n<\/td>\n<td>captures real-time metrics during training<\/p>\n<\/td>\n<td>monitors CPUs, GPUs, network, and memory<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>19<\/td>\n<td>\n<\/td>\n<td>SageMaker Serverless Inference<\/p>\n<\/td>\n<td>deploy and scale ML models<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>20<\/td>\n<td>\n<\/td>\n<td>SageMaker Edge Manager<\/p>\n<\/td>\n<td>Optimize, secure, monitor, and maintain ML models on fleets of edge devices<\/p>\n<\/td>\n<td>smart cameras, robots, personal computers, and mobile devices<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>21<\/td>\n<td>\n<\/td>\n<td>SageMaker Neo<\/p>\n<\/td>\n<td>After training, use Neo to compile the model <\/p>\n<p> train once and run anywhere in the cloud<br \/>\n                        and at the edge<\/p>\n<\/td>\n<td>supports the most popular DL models &#8211; AlexNet, ResNet, VGG, Inception, MobileNet, SqueezeNet,<br \/>\n                        and DenseNet models trained in MXNet and TensorFlow, and classification and random cut forest<br \/>\n                        models trained in XGBoost<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>22<\/td>\n<td>\n<\/td>\n<td>SageMaker Model Monitor<\/p>\n<\/td>\n<td>monitors the quality of Amazon SageMaker machine learning models <\/p>\n<p> Monitors data\/model<br \/>\n                        quality, bias drift for models, feature attribution drift for models <\/p>\n<p> Get alert for<br \/>\n                        deviations. Either fix or retrain<\/p>\n<\/td>\n<td>Continuous &#8211; real-time endpoint <br \/> Continuous &#8211; batch transform job <br \/> Scheduled &#8211;<br \/>\n                        asynchronous batch transform jobs<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>23<\/td>\n<td>\n<\/td>\n<td>SageMaker Model Registry<\/p>\n<\/td>\n<td>Centralized repository allows to track\/manage and version models <br \/> Catalog models, manage<br \/>\n                        model versions, associate metadata with a model <br \/> Manage approval status of a model, automate<br \/>\n                        model deployment, share models<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>24<\/td>\n<td>\n<\/td>\n<td>SageMaker Pipelines<\/p>\n<\/td>\n<td>Process of building training and deploy <br \/> CICD for ML <br \/> Steps <br \/> Processing, training,<br \/>\n                        tuning, automl,model, clarifycheck, quality check<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>25<\/td>\n<td>\n<\/td>\n<td>SageMaker Feature Store<\/p>\n<\/td>\n<td>for sharing and managing variables (features) across multiple teams during model<br \/>\n                        development <\/p>\n<p> Ingests feature from variety of sources <br \/> Can publish directly from<br \/>\n                        sagemaker data wrangler into feature store<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>26<\/td>\n<td>\n<\/td>\n<td>SageMaker Model Cards<\/p>\n<\/td>\n<td>Provide model documentation, not feature management.<\/p>\n<\/td>\n<td>Use case &#8211; intended uses, risk ratings and training details<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>27<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>28<\/td>\n<td>\n<\/td>\n<td>SageMaker JumpStart<\/p>\n<\/td>\n<td>ML hub to find pretrained foundation model, computer vision models or nlp models <br \/> for quickly<br \/>\n                        deploying and consuming a foundation model (FM) within a team&#8217;s VPC. <br \/> Models can be fully<br \/>\n                        customized or access prebuilt solutions and deployed<\/p>\n<\/td>\n<td>Provides access to a wide range of pre-trained models and solutions that can be easily deployed<br \/>\n                        and consumed within a VPC.  <br \/> Designed to simplify and accelerate the deployment of machine<br \/>\n                        learning models, including foundation models.<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>29<\/td>\n<td>\n<\/td>\n<td>SageMaker Role Manager<\/p>\n<\/td>\n<td>Define role for personas<\/p>\n<\/td>\n<td>Ex: data scientist, analyst<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>30<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>31<\/td>\n<td>2<\/p>\n<\/td>\n<td>Amazon Bedrock<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>32<\/td>\n<td>3<\/p>\n<\/td>\n<td>Q<\/p>\n<\/td>\n<td> generative AI\u2013powered assistant for accelerating software development and leveraging companies&#8217;<br \/>\n                        internal data<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>33<\/td>\n<td>\n<\/td>\n<td>Q Developer<\/p>\n<\/td>\n<td>Coding, testing, and upgrading applications, to diagnosing errors, performing security scanning<br \/>\n                        and fixes, and optimizing AWS resources<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>34<\/td>\n<td>\n<\/td>\n<td>Q business<\/p>\n<\/td>\n<td>generative AI\u2013powered assistant that can answer questions, provide summaries, generate content,<br \/>\n                        and securely complete tasks based on data and information in your enterprise systems<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>35<\/td>\n<td>\n<\/td>\n<td>Q for QuickSight<\/p>\n<\/td>\n<td>unified business intelligence (BI) <\/p>\n<p> multi-visual Q&amp;A responses, get AI-driven<br \/>\n                        executive summaries of dashboards, and create detailed and customizable data stories<br \/>\n                        highlighting key insights, trends, and drivers<\/p>\n<\/td>\n<td>customers get a Generative BI assistant that allows business analysts to use natural language to<br \/>\n                        build BI dashboards in minutes and easily build visualizations and complex calculations<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>36<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>37<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>38<\/td>\n<td>\n<\/td>\n<td>Q for Connect<\/p>\n<\/td>\n<td>real-time conversation with the customer along with relevant company content to automatically<br \/>\n                        recommend what to say or what actions an agent should take to better assist customers.<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>39<\/td>\n<td>\n<\/td>\n<td>Q for Supply Chain<\/p>\n<\/td>\n<td>inventory managers, supply and demand planners, and others will be able to ask and get<br \/>\n                        intelligent answers about what is happening in their supply chain, why it is happening, and what<br \/>\n                        actions to take. They will also be able to explore what-if scenarios to understand the<br \/>\n                        trade-offs between different supply chain choices<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>40<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>41<\/td>\n<td>3<\/p>\n<\/td>\n<td>Amazon Comprehend<\/p>\n<\/td>\n<td>For NLP <br \/> Language, extracts key phrases, <br \/> Custom classifier &#8212; organize documents into<br \/>\n                        categories <\/p>\n<p> Analyzes text using tokenization <br \/> Supports text\/pdf\/word\/images etc.,<\/p>\n<\/td>\n<td>Text and Documents <\/p>\n<p> Ex: analyze email, create group articles that comprehend will<br \/>\n                        uncover <br \/> Use case Custom entities &#8211; analyze text for specific terms, list of entities <\/p>\n<p>                        Sentiment analysis<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>42<\/td>\n<td>\n<\/td>\n<td>Amazon Translate<\/p>\n<\/td>\n<td>Natural and accurate translate languages <\/p>\n<p> Custom terminology &#8211; csv\/tsv\/tmx<\/p>\n<\/td>\n<td>Text and Documents <\/p>\n<p> Use cases &#8211; websites and applications, for international users <br \/>\n                        Html\/text documents from S3<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>43<\/td>\n<td>\n<\/td>\n<td>Amazon Textract<\/p>\n<\/td>\n<td>Extract text. Handwriting and data from any scanned documents using AI\/ML <\/p>\n<p> Ex: scan a<br \/>\n                        image and read the text<\/p>\n<\/td>\n<td>Text and Documents <\/p>\n<p> Use cases &#8211; financial services. Health care, public sector (health<br \/>\n                        forms etc.,_<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>44<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>45<\/td>\n<td>4<\/p>\n<\/td>\n<td>Amazon Rekognition<\/p>\n<\/td>\n<td>Find objects, people, text, scenes in images and videos <\/p>\n<p> Custom labels &#8211; identify\/find<br \/>\n                        your own pics\/logos. Ex: NFL <\/p>\n<p> Content moderation &#8211; detect inappropriate, unwanted,<br \/>\n                        offensive content <\/p>\n<p> Custom Moderation Adapters &#8211; extend rek capabilities by providing your<br \/>\n                        own labeled set of images<\/p>\n<\/td>\n<td>Vision <\/p>\n<p> Use cases &#8211; labeling, content moderation, text detection, face detection and<br \/>\n                        analysis (gender) <br \/> Celebrity recognition <\/p>\n<p> Filter out harmful images<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>46<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>47<\/td>\n<td>5<\/p>\n<\/td>\n<td>Amazon Kendra<\/p>\n<\/td>\n<td>Document search service <br \/> Extract answers from docs &#8211; text\/pdf\/html\/ppt\/word etc., <br \/> Natural<br \/>\n                        language search capabilities <br \/> Creates knowledge index\/powered by ML internally<\/p>\n<\/td>\n<td>Search<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>48<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>49<\/td>\n<td>6<\/p>\n<\/td>\n<td>Amazon Lex<\/p>\n<\/td>\n<td>Using voice and text <br \/> Conversational ai with multiple languages <br \/> Integrates with lambda,<br \/>\n                        Connect, comprehend, kendra<\/p>\n<\/td>\n<td>Chatbots<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>50<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>51<\/td>\n<td>7<\/p>\n<\/td>\n<td>Amazon Polly<\/p>\n<\/td>\n<td>Convert text to speech <\/p>\n<p> Lexicons &#8211; <br \/> &#8211; define how to read certain pieces of text <br \/>\n                        ex: AWS =&gt; Amazon Web Services <\/p>\n<p> SSML <br \/> &#8211; Speech synthesis markup language <br \/> &#8211;<br \/>\n                        markup how the text should be pronounced <\/p>\n<p> Voice engine <br \/> &#8211; generative, neural<br \/>\n                        standard <\/p>\n<p> Speech mark <br \/> &#8211; ex: lip syncing or highlight word as they are spoken <br \/>\n                        encode where a sentence\/word starts or ends in an audio<\/p>\n<\/td>\n<td>Speech <\/p>\n<p> Generative <br \/> Long form <br \/> Neural <br \/> standard<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>52<\/td>\n<td>\n<\/td>\n<td>Amazon Transcribe<\/p>\n<\/td>\n<td>Convert speech to text <br \/> Deep learning process called automatic speech recognition <br \/> Removes<br \/>\n                        PII using redaction <br \/> Supports automatic language identification for multi lingual<br \/>\n                        audio <\/p>\n<p> Custom Vocabularies &#8211; Can capture domain specific\/non-standard terms <br \/> Provide<br \/>\n                        hints to increase recognition <\/p>\n<p> Custom language models (for context) &#8211; for domain specific<\/p>\n<\/td>\n<td>Speech <\/p>\n<p> Use cases &#8211; <br \/> customer service calls, automate closed captioning\/subtitling,<br \/>\n                        generate meta data for media assets to create a fully searchable archive <\/p>\n<p> Can transcribe<br \/>\n                        multiple languages at the same time<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>53<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>54<\/td>\n<td>8<\/p>\n<\/td>\n<td>Amazon Personalize<\/p>\n<\/td>\n<td>Ex: retail stores, media and entertainment<\/p>\n<\/td>\n<td>Recommendation<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>55<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>56<\/td>\n<td>9<\/p>\n<\/td>\n<td>AWS DeepRacer<\/p>\n<\/td>\n<td>Console to train and evaluate deep RL<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>57<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>58<\/td>\n<td>10<\/p>\n<\/td>\n<td>Amazon Forecast<\/p>\n<\/td>\n<td>ML to deliver highly accurate forecasts<\/p>\n<\/td>\n<td>Use case &#8211; predict future sales <br \/> Product demand planning, financial planning, resource<br \/>\n                        planning<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>59<\/td>\n<td>\n<\/td>\n<td><\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>60<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>61<\/td>\n<td>11<\/p>\n<\/td>\n<td>Amazon Mechanical Turk<\/p>\n<\/td>\n<td>Crowdsourcing marketplace <br \/> Distributed virtual workforce <br \/> Integrates with Amazon A2I,<br \/>\n                        SageMaker Ground Truth etc.,<\/p>\n<\/td>\n<td>Use case &#8211; label 1000000 images <br \/> Data collection, business processing etc.,<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>62<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>63<\/td>\n<td>12<\/p>\n<\/td>\n<td>Amazon Augmented AI<\/p>\n<\/td>\n<td>Human oversight of machine learning predictions in production<\/p>\n<\/td>\n<td>Can be own employees or AWS\/contractors<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>64<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>65<\/td>\n<td>13<\/p>\n<\/td>\n<td>Amazon Comprehend Medical and Transcribe<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>66<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>67<\/td>\n<td>14<\/p>\n<\/td>\n<td>Amazon&#8217;s Hardware for AI<\/p>\n<\/td>\n<td>AWS Trainium &#8211; Trn1 instance <\/p>\n<p> AWS Inferentia &#8211; ML chip built to deliver inference <br \/> 4x<br \/>\n                        throughput, 70% cost reduction <\/p>\n<p> EC2, EBS,EFS, ELB, ASG <br \/> EC2 user data\/firewall<\/p>\n<\/td>\n<td>EC2 GPU &#8211; P3, P4, P5,\u2026. G3,.. G6<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>68<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr><\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n\n<p><\/body><\/p>\n<p><\/html><\/p>\n<\/details>\n\n\n\n<p><\/p>\n\n\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\"><summary>Machine Learning Notes<\/summary>\n<p class=\"is-layout-flow wp-block-details-is-layout-flow\">\n\n<p><html><\/p>\n<p><head><\/p>\n<style>\n        \/* table td {<br \/>\n                border-top: thin solid;<br \/>\n                border-left: thin solid;<br \/>\n                border-right: thin solid;<br \/>\n                border-bottom: thin solid;<br \/>\n            }<\/p>\n<p>            table td:first-child {<br \/>\n                border-left: thin solid;<br \/>\n            }<\/p>\n<p>            table td:last-child {<br \/>\n                border-right: thin solid;<br \/>\n            } *\/<br \/>\n        table.stats4 {<br \/>\n            text-align: center;<br \/>\n            font-family: Verdana, Geneva, Arial, Helvetica, sans-serif;<br \/>\n            font-weight: normal;<br \/>\n            font-size: 12px;<br \/>\n            color: #fff;<br \/>\n            background-color: #666;<br \/>\n            border: 0px;<br \/>\n            border-collapse: collapse;<br \/>\n            border-spacing: 7px;<br \/>\n        }<\/p>\n<p>        table.stats4 td {<br \/>\n            background-color: #c9c4c4;<br \/>\n            color: #000;<br \/>\n            padding: 5px 7px 5px 7px;<br \/>\n            text-align: left;<br \/>\n            border: 1px #fff solid;<br \/>\n        }<\/p>\n<p>        table.stats4 tr {<br \/>\n            background-color: #f19a9a;<br \/>\n            color: #000;<br \/>\n            padding: 5px 7px 5px 7px;<br \/>\n            text-align: left;<br \/>\n            border: 1px #fff solid;<br \/>\n        }<\/p>\n<p>        table.stats4 td.hed {<br \/>\n            background-color: #666;<br \/>\n            color: #fff;<br \/>\n            text-align: left;<br \/>\n            border: 2px #fff solid;<br \/>\n            font-size: 12px;<br \/>\n            font-weight: bold;<br \/>\n        }<\/p>\n<p>        tr.trHeader {<br \/>\n            background-color: #9f0404;<br \/>\n            color: #fff;<br \/>\n            text-align: left;<br \/>\n            border: 2px #fff solid;<br \/>\n            font-size: 12px;<br \/>\n            font-weight: bold;<br \/>\n        }<br \/>\n    <\/style>\n<p><\/head><\/p>\n<p><body><\/p>\n<div>\n<table class=\"stats4\">\n<thead>\n<tbody>\n<tr class=\"trHeader\">\n<td>1<\/td>\n<td>SNo<\/p>\n<\/td>\n<td>Type<\/p>\n<\/td>\n<td>Used for<\/p>\n<\/td>\n<td>Name<\/p>\n<\/td>\n<td>Details<\/p>\n<\/td>\n<td>Use Cases<\/p>\n<\/td>\n<td>Comment\n                    <\/td>\n<\/tr>\n<tr>\n<td>2<\/td>\n<td>1<\/p>\n<\/td>\n<td>Supervised Learning<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Linear Regression<\/p>\n<\/td>\n<td>Model relationship&nbsp; between one or more input features <br \/>&nbsp; <br \/>One output&nbsp;<br \/>\n                        variable \u2014 target<\/p>\n<\/td>\n<td>Historical sales data, output &#8211; no of units to be produced <br \/>Predict House prices, stocks<br \/>\n                        prices, sales volume etc.,<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>3<\/td>\n<td>2<\/p>\n<\/td>\n<td>Supervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Binary classification<\/p>\n<\/td>\n<td>Binary outcome yes\/no, true\/false, +\/-<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>4<\/td>\n<td>3<\/p>\n<\/td>\n<td>Supervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Time series prediction<\/p>\n<\/td>\n<td>forecasts future values based on past and present data<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>5<\/td>\n<td>4<\/p>\n<\/td>\n<td>Supervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Regression<\/p>\n<\/td>\n<td>estimates <b>a continuous numerical value<\/b> based on the input features<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>6<\/td>\n<td>5<\/p>\n<\/td>\n<td>Supervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>recurrent neural network (RNN)<\/p>\n<\/td>\n<td>type of neural network that can process sequential data. suited for predicting future events<br \/>\n                        based on past observations <br \/>&nbsp; <br \/>NOTE: CNN is for images and RNN is for timeseries.<\/p>\n<\/td>\n<td>forecasting engine failures based on sensor readings<\/p>\n<\/td>\n<td>TensorFlow, PyTorch, Keras, MXNet<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>7<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>convolutional neural network (CNN)<\/p>\n<\/td>\n<td>Classify an object amongst a group <br \/>&nbsp; <br \/>NOTE:&nbsp; CNN is for images and RNN is for<br \/>\n                        timeseries.<\/p>\n<\/td>\n<td>an animal image as input and identify probability distribution of how likely amongst 10<br \/>\n                        types of animals<\/p>\n<\/td>\n<td>Softmax function transforms a a arbitrary real values&nbsp; into a range of<br \/>\n                        (0,1) <br \/>&nbsp; <br \/>TensorFlow, PyTorch, Keras, MXNet<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>8<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>WaveNet<\/p>\n<\/td>\n<td>Generative model for raw audio <br \/>&nbsp; <br \/>WaveNet is a deep autoregressive CNN with<br \/>\n                        stacked layers of dilated convolution, used for generating speech. To deliver a more human-like<br \/>\n                        voice,  <br \/>WaveNet: \u2022 \ud835\udde0\ud835\uddfc\ud835\uddf1\ud835\uddf2\ud835\uddf9\ud835\ude00 \ud835\ude01\ud835\uddf5\ud835\uddf2 \ud835\uddff\ud835\uddee\ud835\ude04 \ud835\ude04\ud835\uddee\ud835\ude03\ud835\uddf2\ud835\uddf3\ud835\uddfc\ud835\uddff\ud835\uddfa \ud835\uddfc\ud835\uddf3 \ud835\uddee\ud835\ude02\ud835\uddf1\ud835\uddf6\ud835\uddfc<br \/>\n                        \ud835\ude00\ud835\uddf6\ud835\uddf4\ud835\uddfb\ud835\uddee\ud835\uddf9\ud835\ude00, making the voice sound more natural and expressive <br \/>&nbsp; <br \/>In WaveNet, the<br \/>\n                        CNN takes a raw signal as an input and synthesises an output one sample at a time <\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>9<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>classification<\/p>\n<\/td>\n<td>KNN (K nearest neighbor)<\/p>\n<\/td>\n<td>finding the k most similar instances in the training data to a given query instance, and<br \/>\n                        then predicting the output based on the average or majority of the outputs of the k nearest<br \/>\n                        neighbors  <br \/>&nbsp; <br \/>handle time series data<\/p>\n<\/td>\n<td>Ex: air quality data and predict for next 2 days based on last 2 year<br \/>\n                        data <br \/>&nbsp; <br \/>Identify if imge has a logo amongst a larger group<\/p>\n<\/td>\n<td>can perform both classification and regression tasks<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>10<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>Latent Dirichlet Allocation (LDA)<\/p>\n<\/td>\n<td>suitable for topic modeling tasks (in NLP) <br \/>&nbsp; <br \/>discover the hidden topics and their<br \/>\n                        proportions in a collection of text documents,<\/p>\n<\/td>\n<td>&nbsp;news articles, tweets, reviews, etc<\/p>\n<\/td>\n<td>Gensim, Scikitlearn, Mallet  <br \/>&nbsp; <br \/>Not valid for images<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>11<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>Factorization Machines (FM) Algorithm<\/p>\n<\/td>\n<td>used for tasks dealing with high dimensional sparse datasets<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>12<\/td>\n<td>\n<\/td>\n<td>Unsupervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Topic Modeling<\/p>\n<\/td>\n<td>Topic modeling is a type of statistical modeling that uses unsupervised Machine Learning to<br \/>\n                        identify clusters or groups of similar words within a body of text<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>13<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>BERT based models<\/p>\n<\/td>\n<td>Google developed BERT to serve as a bidirectional transformer model that examines words<br \/>\n                        within text by considering both left-to-right and right-to-left contexts <br \/>&nbsp; <\/p>\n<\/td>\n<td>Missing words in&nbsp; text<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>14<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>15<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>16<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>17<\/td>\n<td>\n<\/td>\n<td>Unsupervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Principal component analysis (PCA)<\/p>\n<\/td>\n<td>reduce the dimensionality (number of features) within a dataset while still retaining as<br \/>\n                        much  <br \/>information as possible  <br \/>&nbsp; <br \/>Used when the features are highly correlated with<br \/>\n                        each other<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Using finding a new set of features called components<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>18<\/td>\n<td>6<\/p>\n<\/td>\n<td>Unsupervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Random Cut Forest (RCF)<\/p>\n<\/td>\n<td>assigns an anomaly score to each data point based on how different it is from the rest of<br \/>\n                        the data<\/p>\n<\/td>\n<td>Ex: realtime ingestion, identify anamoly\/malicious events<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>19<\/td>\n<td>7<\/p>\n<\/td>\n<td>Unsupervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Anomaly detection<\/p>\n<\/td>\n<td>identifies outliers or abnormal patterns in the data<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>20<\/td>\n<td>8<\/p>\n<\/td>\n<td>Unsupervised<\/p>\n<\/td>\n<td>\n<\/td>\n<td>K-means clustering<\/p>\n<\/td>\n<td>randomly assigning data points to a number of clusters, then iteratively updating the<br \/>\n                        cluster centers and reassigning the data points until the clusters are stable<\/p>\n<p>&nbsp; <br \/>result is a partition of the data into distinct and homogeneous groups<\/p>\n<\/td>\n<td>exploratory data analysis, data compression, anomaly detection, and feature extraction<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>21<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>RMSE Root mean square error<\/p>\n<\/td>\n<td>Goal &#8211; to predict a continuous value <br \/>&nbsp; <br \/>measures the average difference between<br \/>\n                        the predicted and the actual values  <\/p>\n<\/td>\n<td>Price of a house, temperature of a city<\/p>\n<\/td>\n<td>Good for regression <br \/>NOT good fo classification<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>22<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>regression<\/p>\n<\/td>\n<td>MAPE Mean absolute percentage error<\/p>\n<\/td>\n<td>Used for regression<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>23<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>ROC receiver operating characteristic (ROC) curve<\/p>\n<\/td>\n<td>used to understand how different classification thresholds will impact the models<br \/>\n                        performance.  <br \/>&nbsp; <br \/>A ROC curve can show the trade-off between the True positive rate TPR<br \/>\n                        and the FPR for different thresholds<\/p>\n<\/td>\n<td>predict whether or not a person will order a pizza<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>24<\/td>\n<td>9<\/p>\n<\/td>\n<td>\n<\/td>\n<td>Classification (binary)<\/p>\n<\/td>\n<td>Area Under ROC Curve (AOC)<\/p>\n<\/td>\n<td>Compare\/evaluate ML models <br \/>&nbsp; <br \/>AUC is calculated based on the Receiver <br \/>Operating<br \/>\n                        Characteristic (ROC) curve, which is a plot that shows the trade-off between the<br \/>\n                        true <br \/>positive rate (TPR) and the false positive rate (FPR) of the classifier as the decision<br \/>\n                        threshold is <br \/>varied. The TPR, also known as recall or sensitivity <\/p>\n<\/td>\n<td>Credit card transactions &#8211; identify 99k valid vs 1k fraudulent<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>25<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>Residual plots<\/p>\n<\/td>\n<td>used to understand whether a <br \/>regression model is more frequently overestimating or<br \/>\n                        underestimating the target<\/p>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>26<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>Confusion matrix<\/p>\n<\/td>\n<td>table that shows the counts of true positives, false positives, true negatives, and false<br \/>\n                        negatives for each class  <br \/>&nbsp; <br \/>indicate the accuracy, precision, recall, and F1-score of<br \/>\n                        the model for each class,<\/p>\n<\/td>\n<td>\n<\/td>\n<td>only applicable for classification models, not regression models. A confusion matrix cannot<br \/>\n                        show the magnitude or direction of the errors made by the model.<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>27<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>Precision<\/p>\n<\/td>\n<td>proportion of predicted positive cases that are actually positive. Precision is a useful<br \/>\n                        metric when the cost of a false positive is high  <br \/>Recall is not a good metric for imbalanced<br \/>\n                        classification problems<\/p>\n<\/td>\n<td>fraudulent transactions <br \/>spam detection or medical diagnosis  <\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>28<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>Classification<\/p>\n<\/td>\n<td>Recall<\/p>\n<\/td>\n<td>Same as TPR (true positive rate) <br \/>Recall is a useful metric when the cost of a false<br \/>\n                        negative is high  <br \/>&nbsp; <br \/>Recall = True Positives \/ (True Positives + False Negatives)<\/p>\n<\/td>\n<td>fraud detection or cancer diagnosis<\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>29<\/td>\n<td>\n<\/td>\n<td>Supervised<\/p>\n<\/td>\n<td>Classification (multi-class)<\/p>\n<\/td>\n<td>XGBoost<\/p>\n<\/td>\n<td>Can handle multiple features and multiple classes<\/p>\n<\/td>\n<td>Categorize new products when a dataset\/features is provided <br \/>Ex: with 15 features<br \/>\n                        (title\/weight\/price) categorize books\/games\/movies etc from a dataset of 1200<br \/>\n                        products <br \/>&nbsp; <br \/>&nbsp; <br \/>Credit card fraud detection (ex: with a large dataset of<br \/>\n                        historical data, find\/predict new txns)<\/p>\n<\/td>\n<td>can be used for classification, regression, ranking, and other tasks. It is based on the<br \/>\n                        gradient boosting algorithm, which builds an ensemble of weak learners (usually decision trees)<br \/>\n                        to produce a strong learner<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>30<\/td>\n<td>\n<\/td>\n<td>\n<\/td>\n<td>classification<\/p>\n<\/td>\n<td>Term frequency-inverse document frequency (TF-IDF)<\/p>\n<\/td>\n<td>assigns a weight to each word in a document based on how important it is to the meaning of<br \/>\n                        the document <br \/>&nbsp; <br \/>NOTE: The term frequency (TF) measures how often a word appears in a<br \/>\n                        document, while the inverse document frequency (IDF) measures how rare a word is across a<br \/>\n                        collection of documents.<\/p>\n<\/td>\n<td>&nbsp; <\/p>\n<\/td>\n<td>\n<\/td>\n<\/tr>\n<tr>\n<td>31<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Classification\/ categorize<\/p>\n<\/td>\n<td>Word2vec<\/p>\n<\/td>\n<td>technique that can learn distributed representations of words, also known as word<br \/>\n                        embeddings, from large amounts of text data <br \/>&nbsp; <br \/>&nbsp;<\/p>\n<\/td>\n<td>when tuning parameters doesn&#8217;t help a lot. Transfer learning would be better solution<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>32<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Collaborative Filtering<\/p>\n<\/td>\n<td>recommends products or services to users based on the ratings or preferences of other users<\/p>\n<\/td>\n<td>customer shopping patterns and preferences based on demographics, past visits, and locality<br \/>\n                        information<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>33<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Decision tree<\/p>\n<\/td>\n<td>perform classification tasks by splitting <br \/>the data into smaller and purer subsets based<br \/>\n                        on a series of rules or conditions<\/p>\n<\/td>\n<td>binary classifier based on two features: age of account and transaction month<\/p>\n<\/td>\n<td>both linear and non-linear data, and can capture complex patterns and interactions<br \/>\n                        among <br \/>the features<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>34<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>35<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Preprocessing technique<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Data normalization<\/p>\n<\/td>\n<td>Scale the feature to a common range (0,1) or (-1,)<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>min-max scaling, z-score standardization, or unit vector <br \/>normalization<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>36<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Preprocessing technique<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Dimensionality reduction<\/p>\n<\/td>\n<td>Reduce number of features<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>37<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Preprocessing technique<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Model regularization<\/p>\n<\/td>\n<td>adds a penalty term to the cost function to prevent overfitting<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>38<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>L1\/L2 regularization<\/p>\n<\/td>\n<td>Overfitting problem can be addressed by applying regularization techniques such as L1 or L2<br \/>\n                        regularization and dropouts.  <br \/>&nbsp; <br \/>Regularization techniques add a penalty term to the<br \/>\n                        cost function of the model, which helps to reduce the complexity of the model and prevent it<br \/>\n                        from overfitting to the training data. Dropouts randomly turn off some of the neurons during<br \/>\n                        training, which also helps to prevent overfitting<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>39<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Preprocessing technique<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Data augmentation<\/p>\n<\/td>\n<td>increases the amount of data by creating synthetic <br \/>samples<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>40<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Poisson distribution<\/p>\n<\/td>\n<td>suitable for modeling the number of events that occur in a fixed interval of time or space,<br \/>\n                        given a known average rate of occurrence<\/p>\n<\/td>\n<td>waiting for a bus, the interval is 10 minutes, and the average rate is 3 minutes<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>41<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Normal distribution<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>42<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Binomial distribution<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>43<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>Uniform distribution<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>44<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<tr>\n<td>45<\/td>\n<td>10<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<td>&nbsp;<\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/thead>\n<\/table>\n<\/div>\n\n<p><\/body><\/p>\n<p><\/html><\/p>\n<div>\n        <b>Other Notes<\/b><\/p>\n<p>        Data preprocessing &#8211; is the process of generating raw data for machine learning models<br \/>&nbsp;<br \/>\n        Feature engineering &#8211; refers to manipulation \u2014 addition, deletion, combination, mutation \u2014 of your data set to<br \/>&nbsp;<br \/>\n        improve machine learning model training, leading to better performance and greater accuracy.<br \/>&nbsp;<br \/>\n        Exploratory data analysis (EDA) &#8211; is used by data scientists to analyze and investigate data sets and summarize<br \/>\n        their main characteristics, often employing data visualization methods.<br \/>&nbsp;<br \/>\n        Hyperparameter tuning &#8211; is the process of selecting the optimal values for a machine learning model&#8217;s<br \/>\n        hyperparameters<br \/>&nbsp;<br \/>\n        Transfer learningis a strategy for adapting pre-trained models for new, related tasks without creating models<br \/>\n        from scratch.<br \/>&nbsp;<br \/>&nbsp;<\/p>\n<p>        Epochs &#8211; helps to improve accuracy<br \/>&nbsp;<br \/>\n        &#8211; Increasing the number of epochs during model training allows the model to learn from the data over more<br \/>&nbsp;<br \/>\n        iterations, potentially improving its accuracy up to a certain point. This is a common practice when attempting<br \/>\n        to reach a specific level of accuracy. Increasing epochs allows the model to learn more from the data, which can<br \/>\n        lead to higher accuracy.<br \/>&nbsp;<br \/>\n        &#8211; Decreasing the epochs would reduce the training time, possibly preventing the model from reaching the desired<br \/>\n        accuracy.<br \/>&nbsp;<br \/>&nbsp;<\/p>\n<p>        Batch Size &#8211; Affects training speed<br \/>&nbsp;<br \/>\n        &#8211; Decrease the batch size affects training speed and may lead to overfitting but does not directly relate to<br \/>\n        achieving a specific accuracy level.<br \/>&nbsp;<br \/>&nbsp;<\/p>\n<p>        Temperature &#8211; Affects randomness of predictions<br \/>&nbsp;<br \/>\n        &#8211; Increase the temperature parameter affects the randomness of predictions, not model accuracy.<br \/>&nbsp;<br \/>\n        &#8211; Decrease the temperature to produce more consistent responses to the same input prompts<br \/>&nbsp;\n    <\/div>\n<\/p>\n<\/details>\n<\/details>\n<!-- \/wp:details -->\n\n<!-- wp:paragraph -->\n<p><\/p>\n<!-- \/wp:paragraph -->\n\n<!-- wp:paragraph -->\n<p><\/p>\n<!-- \/wp:paragraph -->","protected":false},"excerpt":{"rendered":"<p>Howdy! I was able to crack AWS Certified AI Practitioner Certification | AWS Certification | AWS (amazon.com) recently.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"_links":{"self":[{"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/posts\/195"}],"collection":[{"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/comments?post=195"}],"version-history":[{"count":9,"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/posts\/195\/revisions"}],"predecessor-version":[{"id":207,"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/posts\/195\/revisions\/207"}],"wp:attachment":[{"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/media?parent=195"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/categories?post=195"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.codeandtech.com\/blog\/wp-json\/wp\/v2\/tags?post=195"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}