NimbusML enables data scientists to use ML.NET to train models in Azure Machine Learning or anywhere else they use Python. NimbusML provides state-of-the-art ML algorithms, transforms and components, aiming to make them useful for all developers, data scientists, and information workers and helpful in all products, services and devices. The components are authored by the team members, as well as numerous contributors from MSR, CISL, Bing and other teams at Microsoft. NimbusML is interoperable with scikit-learn estimators and transforms, while adding a suite of highly optimized algorithms written in C++ and C# for speed and performance.
The trained machine learning model can be used in a .NET application with ML.NET. This presentation will outline the features of NimbusML and provide a notebook-based demonstration using Azure Notebooks.
This presentation is the third of four related to ML.NET and Automated ML. The presentation will be recorded with video posted to this YouTube Channel: http://bit.ly/2ZybKwI
2. Mark Tabladillo Ph.D.
• Science doctorate from Georgia Tech
• Analytics career based on SAS,
Microsoft, open source
• Tech Presentations:
• Seattle, Portland, Chicago, Boston,
Mountain View, San Francisco, San
Antonio, Charlotte, Orlando
• London, Hong Kong, Montreal
• Social Media
LinkedIn
Twitter @marktabnet
• Cloud Solution Architect
• US CTO Customer Success
5. Domain specific pretrained models
To simplify solution development
Azure
Databricks
Machine
Learning VMs
Popular frameworks
To build advanced deep learning solutions
TensorFlowPyTorch ONNX
Azure Machine
Learning
LanguageSpeech
…
SearchVision
Productive services
To empower data science and development teams
Powerful infrastructure
To accelerate deep learning
Scikit-Learn
Familiar Data Science tools
To simplify model development
CPU GPU FPGA
From the Intelligent Cloud to the Intelligent Edge
Azure Notebooks JupyterVisual Studio Code Command line
7. About
NimbusML
• NimbusML provides state-of-the-art ML
algorithms, transforms and components, aiming
to make them useful for all developers, data
scientists, and information workers and helpful
in all products, services and devices.
• The components are authored by the team
members, as well as numerous contributors
from MSR, CISL, Bing and other teams at
Microsoft.
• nimbusml is interoperable with scikit-learn
estimators and transforms, while adding a suite
of highly optimized algorithms written in C++
and C# for speed and performance.
8. NimbusML
Features
NimbusML trainers and transforms support
the following data structures for the fit() and
transform() methods:
• numpy.ndarray
• scipy.sparse_cst
• pandas.DataFrame.
NimbusML also supports streaming from files
without loading the dataset into memory,
which allows training on data significantly
exceeding memory using FileDataStream.
• With FileDataStream, NimbusML is able to handle up to
billion features and billions of training examples for
select algorithms