Breaking News

AWS, aim to bridge scaling gap with machine learning models via Coursera specialization

Side view of hands using laptop with glowing circuit head outline. AI and machine learning concept. Multiexposure” data-id=”59″ data-m=”{“i”:59,”p”:58,”n”:”openModal”,”t”:”articleImages”,”o”:1}”>

© Getty Images/iStockphoto
Side view of hands using laptop with glowing circuit head outline. AI and machine learning concept. Multiexposure

special feature

© Provided by ZDNet
Managing AI and ML in the Enterprise

Managing AI and ML in the Enterprise

The AI and ML deployments are well underway, but for CXOs the biggest issue will be managing these initiatives, and figuring out where the data science team fits in and what algorithms to buy versus build.

Read More

Amazon Web Services,, and Coursera are looking to bridge the gap between creating and testing machine learning and models and scaling them in production via a three-course specialization.

“Machine learning has a proof of concept to production gap,” explained Andrew Ng, founder of DeepLearning.AI and a top instructor on Coursera. The specialization is designed to help developers take a model from a prototype on a laptop to the cloud. “There’s just so much stuff to be done when going from 10 users to 1 million,” added Ng.

Bratin Saha, vice president and general manager of machine learning services at AWS, said customers have gone from deploying a handful of models to millions in just a few years. “ML is no longer a niche,” said Saha, who oversees SageMaker, the machine learning platform that is the fastest growing product at AWS.

Must read:

The specialization course gives an overview of the moving parts (MLOps, DevOps) required to move models into production as well as topics covering accuracy, costs and optimization as prototypes scale.


Load Error

Primers: What is AI? | What is machine learning? | What is deep learning? | What is artificial general intelligence? 

In an interview with Ng and Saha, we touched on a few notable points about models. A few highlights of our chat:

Should models begin in the cloud from the start to account for scale? Ng said his approach to machine learning models is one based on “use the right tool for the right job.” “It’s fine to do a proof of concept on a laptop. You need the proof of concept to decide go or no go,” said Ng.

Ng added that planning for scale before there’s a proof of concept can muddle the process.

Scaling requires skill. Saha said the specialization course is designed to broaden the talent base for machine learning. Both Saha and Ng said there’s a shortage of talent that understands how to scale models. “Two years back we’d train models with 20 million parameters. Today it’s 100 million. We make 100s of billions of predictions per month,” said Saha.

Ng said there is high demand for skilled machine learning practitioners and those people who have deployed a meaningful service in the cloud are even in more short supply. As a result, Saha said all engineers joining Amazon have mandatory machine learning courses.

Machine learning is early in its evolution. Ng said in many ways machine learning rhymes with software development in its early days. “I vaguely remember when software engineering was a mess and now version control is now more mature,” said Ng. “I take inspiration from how software emerged as an industry.” 

Also: Andrew Ng sees an eternal springtime for AI


As for the Practical Data Science specialization, here are the key points.

  • The specialization is designed for data-focused developers, scientists, and analysts who are familiar with the Python and SQL programming languages and want to build end-to-end machine learning pipelines.
  • Algorithms for natural language processing and natural language understanding including BERT, GLoVe, ELMo and FastText.
  • The first course covers foundational concepts and exploratory data analysis using Amazon SageMaker Studio as well as other SageMaker services. Automated machine learning will be covered.
  • In the second course, learners will build, train and deploy and end-to-end machine learning pipeline.
  • The third course will cover advanced model training, tuning and deployment techniques. Distributed training, hyper-parameter tuning, and A/B tests will be covered.
  • Managed online lab environments are provided by AWS Partner Vocareum.

View Now at Coursera

Continue Reading