Intro to ML on Cloud

Course description

Give an overview of pretrained ML models and services, which could be used for fast prototyping (using GCP as an example). Prerequisites

• Basic knowledge of Python and Jupyter notebooks
• Brief understanding of how ML and DL works (really brief 🙂 )
• Persistent internet access (everything is online)
• GCP account and complete pre-course activities

Pre-course activity All the participants will get a set of instructions, how to create a free GCP account and activate CloudML services. Also, participants will need to activate the access tokens to API services. Should take no more than 1.5 – 2. Finally, it would be cool to have a short summary of participants backgrounds to design more interesting use cases Main risk If people will not set up their accounts before to the course, we could lose half of the time, resolving issues. Therefore, we should have a slack channel a few days before the course where people could ask questions regarding the setup.

The course agenda

Session1 (1.5h)

• The course agenda
• Motivation to take the course:
  • For engineers: you will be able to safely use advanced ML techniques and services without a deep understanding of how they work
  • For data scientists/ ML engineers: you will be able to save time and effort on building everything from scratch, you could use Cloud API to automate simple tasks and concentrate efforts on advanced custom ones
  • For StartUpers and entrepreneurs: you could cheaply and fastly validate your idea and significantly reduce time to market of your product.
• What is Cloud ML and who are the main players (AWS, GCP, Azure)
• Why GCP
• GCP services overview
• Cloud Vision Overview and Web Demo
  • Images Labeling
  • Logo and landmarks detection
  • Web search keywords
  • Safe Search (content validation)
  • Face and mood recognition
• Video Intelligence Overview and possible use cases
  • Scene change
  • Video annotation with labels
  • Searching in video or video collection
  • Live video navigation
• Sharing Notebooks for the first use case (Super fast Vision API Startup)

Session2 (1.5h)

• (1h) Notebook walk through – building your own app
• (15min) Speech API overview
  • Searching in audio files
  • Language detection
  • Audio to text
• (15min) Cloud Natural Language overview
  • Semantic and syntax analysis
  • Content Classification
  • Sentiment Analysis
  • Translation

Session3 (2h)

• AutoML Intro, Services and Use Cases Overview (20m)
• AutoML Demo (15m)
• Winning a Competition with AutoML (15m)
• Beating AutoML (trying to 🙂 ) (1h)
• Q&A and closure (10m)

About the lecturer

Andriy Kusyy Data Scientist with in-depth engineering & research background and strategic business approach to problems. Presently, leading multiple DL and CV related commercial projects, social and research initiatives. Main areas of interest: Computer Vision, Generative Modeling and Finance More at: https://www.linkedin.com/in/andriy-kusyy-806428119/