https://github.com/secretshardul/aws_ml_webinar

Webinar on machine learning with AWS for Smart India Hackathon
https://github.com/secretshardul/aws_ml_webinar

Last synced: 4 months ago
JSON representation

Webinar on machine learning with AWS for Smart India Hackathon

• Host: GitHub
• URL: https://github.com/secretshardul/aws_ml_webinar
• Owner: secretshardul
• License: mit
• Created: 2020-02-04T05:24:27.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2020-02-04T06:40:37.000Z (over 6 years ago)
• Last Synced: 2025-01-19T07:24:45.566Z (over 1 year ago)
• Size: 897 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# AWS_ML_webinar

Webinar on machine learning with AWS for Smart India Hackathon

# Intro

![](images/2020-02-04-11-05-43.png)

AI- anything that mimics human intelligence

ML- abstracts away stat

# ML

- ML is all about prediction.

- Don't code, let the system learn through data. With more data, accuracy of the prediction function increases.

# Types of ML

- Supervised learning: Data is manually labelled

    - Classification: Yes/No

    - Regression: Extrapolating graph

- Unsupervised: No manual labelling but some grouping policy is needed.

    - Clustering

- Reinforcement

# AWS ML stack

1. AI services: rekognition, polly etc

2. ML services: sagemaker

3. ML frameworks and infra(need detailed knowledge): Deep learning AMIs etc

# AI services

1. Rekognition: image recognition

2. Polly: text to speech

3. Comprehend: Extract insights from text(keyphases, sentiments, entities)

# ML services

Sagemaker: 

![](images/2020-02-04-11-23-22.png)

1. Groundtruth: easy instructions for labelling

2. Notebooks: Like Jupyter notebooks but come with ML libraries

3. Training: Readymade algorithms