CS代写 Homework Assignment 3: ML Ops::Spam Detection Due Date: 04/19 11:59pm

Homework Assignment 3: ML Ops::Spam Detection Due Date: 04/19 11:59pm
In this assignment you will implement a machine learning model to predict whether a message is spam or not. Furthermore, you will create a system that upon receipt of an email message, it will automatically flag it as spam or not, based on the prediction obtained from the machine learning model.
Architecture Diagram :
This assignment has the following components:

Copyright By PowCoder代写 加微信 powcoder

• Complete tutorial for using Amazon SageMaker on AWS.
0. FollowthefollowingAWStutorialonhowtouse
Amazon SageMaker to implement the required model:
https://aws.amazon.com/getting-started/hands-

on/build-train-deploy-machine-learning-model-
sagemaker/ (Links to an external site.)
1. Thepurposeofthetutorialistofamiliarizeyouwith Amazon Sagemaker and the basic components of SageMaker.
There is a change that is to be made due to Sagemaker updates:
Change framework_version from 1.6 to 1.2
• Implement a Machine Learning model for predicting whether an SMS message is spam or not.
0. FollowthefollowingAWStutorialonhowto
build and train a spam filter machine learning model using Amazon SageMaker: https://github.com/aws-samples/reinvent2018- srv404-lambda- sagemaker/blob/master/training/README.md (Links to an external site.)
1. Theresultingmodelshouldperformwellonemailsas well, which is what the rest of the assignment will focus on.
2. Deploytheresultingmodeltoanendpoint(E1).
• Implement an automatic spam tagging system.
0. CreateanS3bucket(S1)thatwillstoreemailfiles.
1. UsingSES,setupanemailaddress,thatuponreceipt
of an email it stores it in S3.
0. Confirmthattheworkflowisworkingby sending an email to that email address and seeing if the email information ends up in S3.
2. ForanynewemailfilethatisstoredinS3,triggera Lambda function (LF1) that extracts the body of the email and uses the prediction endpoint (E1) to predict if the email is spam or not.
0. Youmightwanttostripoutnewline
characters “\n” in the email body, to match the data format in the SMS dataset that the ML model was trained on.
3. Replytothesenderoftheemail(itcouldbeyour email, the TA’s etc.) with a message as follows:
“We received your email sent at [EMAIL_RECEIVE_DATE] with the subject [EMAIL_SUBJECT].

Here is a 240 character sample of the email body: [EMAIL_BODY]
The email was categorized as [CLASSIFICATION] with a [CLASSIFICATION_CONFIDENCE_SCORE]% confidence.”
0. Replaceeachvariable“[VAR]”withthe corresponding value from the email and the prediction.
1. Thepurposeofthisstepistofacilitateeasy
• Create an AWS CloudFormation template for the automatic spam tagging system.
0. CreateaCloudFormationtemplate(T1)torepresentallthe infrastructure resources (ex. Lambda, SES configuration, etc.) and permissions (IAM policies, roles, etc.).
1. Thetemplate(T1)shouldtakethepredictionendpoint(E1)asa stack parameter.
Acceptance criteria:
1. TAsshouldbeabletoemailtheuniqueemailaddresssubmittedaspartof
the assignment and they should be able to get reasonable predictions
(spam/not spam) for the emails they send.
2. TAsshouldbeabletostanduptheCloudFormationtemplate(T1)withina
separate account, using their own prediction endpoint (E1’), and successfully test the system.
0. ThisalsoassumesthatyouprovidetheTAswiththecodeforthe Lambda function (LF1).
Extra credit (10 points):
Please find below the assignment prompt to receive extra credits:
In real-world applications, machine learning models are usually retrained on newly obtained data to stay updated. For extra credits, complement your spam classifier with a retraining service. To do that, user Cloudwatch and Lambda function that does the retraining and code deployment. For simplicity, retrain the model on the
same data from scratch.

程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com