MLOps - End-to-End Operationalization of Machine Learning Algorithms

Automation has long been a corner stone of modern Software Engineering and has given rise to successful initiatives such as DevOps.

Automating the end-to-end lifecycle of a machine learning model, from training to deployment in production, is a lot more complex and requires careful consideration of multiple axis of change, such as data, code and configuration, but can yield fantastic results when done right.

MLOps, also known as Continuous Delivery for Machine Learning (CD4ML), has recently gotten a lot of traction and was mentioned by ThoughtWorks as a technique worth keeping a close eye on (https://www.thoughtworks.com/radar/techniques)

Join me in this session where we’ll walk through some common concepts related to MLOps, and look at tools and techniques to achieve this both on-premise and in the cloud.

A Deep Dive into Machine Learning and Deep Learning with ML.NET

It’s no exaggeration to say that Machine Learning (ML) and Deep Learning (DL) are quickly changing the way we live. Things we are accustomed to today are things we could only dream of 10 years ago. As ML is becoming more and more mainstream, you may be asking yourself how do I get started?

Historically, many of the tools and libraries used to train a model have been found in the Python and R communities (e.g. Scikit Learn, PyTorch, TensorFlow), but what if you’re a .NET Developer? Allow me the pleasure to introduce ML.NET!

ML.NET is an open-source, cross-platform machine learning library, specifically built to enable .NET developers to train their own custom machine learning models in C# or F#.
In this workshop we’ll cut through all the foreign jargon around machine learning and give you some hands-on experience training custom models in C# using interactive exercises.

What you’ll learn:
- Understand the concepts around Machine Learning and Deep Learning
- Learn how to train a custom machine learning model using ML.NET in C# and AutoML
- Learn how to use transfer learning to train deep neural networks for computer vision
- Understand when and how we can leverage Jupyter Notebooks with ML.NET
- Explore ways we can deploy our custom machine learning model to production
- Explore how to set up a real-time data pipeline in Azure for real-time inference

What you’ll need to bring
- A laptop
- A free Azure subscription
- An installation of the .NET Core SDK
- An installation of Visual Studio Code

Building a Recommendation System in ML.NET

Ever wondered how Netflix, Amazon, Spotify, Facebook and Instagram determines what to recommend for you? Interested in learning how to build your own recommendation system, whether it would be for movies or something more specific, serving your specific business need?

Join me in this session where we'll take a deep dive into recommendation systems and explore concepts such as collaborative- and content-filtering. We'll leverage the cross-platform, open-source machine learning framework ML.NET to build our own recommendation system in C#/.NET!

From Zero to Hero with Automatic Machine Learning in ML.NET

Imagine being able to train a custom machine learning model without any previous experience or the need to fully understand what algorithms exist and what they do? What if we could take it one step further and get all the needed artifacts to deploy the model as an Azure Function, or to integrate it into an ASP.NET Core app?

Sound too good to be true? It’s far from it.

In this session we'll take a deep dive into the Model Builder in the open-source, cross-platform machine learning library ML.NET. We'll learn how automatic machine learning is quickly transforming the data science community and democratizing AI and ML for everyone.

ML.NET <3 Jupyter Notebooks

ML.NET is brilliant for doing Machine Learning in .NET!

But is the framework really mature enough for a full-fledged Data Science project? How do we work with models trained on 1000+ features? How do I make sense of the data, and how can I create simple and intuitive visualizations? What about quickly prototyping a solution that you can share with a co-worker?

The questions are many.

In this session will explore the new C# kernel for Jupyter Notebooks, and look at the newly released DataFrame type to see how these can help us solve our problems. This session will teach you how you can use C# and .NET for the end-to-end machine learning process.

Model Explainability – Understanding the Decision Process of a Machine Learning Model

Just as a patient wouldn't be okay with a doctor diagnosing them with cancer without giving a thorough explanation of why, neither can we fully trust a machine learning model until we understand its internal decision process.

It's imperative for a model to not only explain what decision it came to, but also why. Exploring the decision process of machine learning models, have over the last couple of years become a hot research topic, as it aims to ensure that important traits such as fairness, privacy, reliability and trust are preserved.

Join me in this session where we'll take a deep dive in to machine learning explainability, and explore a couple of techniques available for data scientists today.

MLOps | Bringing the Ops to Machine Learning

Awesome, your shiny new machine learning model has been trained, now what?

How do you manage its lifecycle, and make sure that both the model and its data source are versioned so that they can be audited? How do you handle the infrastructure needed to serve your model, and wouldn't it also be nice if the model training, evaluation, and deployment all where automated steps as soon as a change was checked in?

Join me in this session where we will take a deep dive in to MLOps, DevOps for Machine Learning. We'll look at how we can leverage tools such as Azure Machine Learning Service and Azure DevOps to set up automated pipelines for your machine learning environment.

The Democratization of Machine Learning: What It Means for Software Engineers

Although academic research in the area of machine learning dates back to the 1950s, it's not up until recently we've seen major breakthroughs such as advanced natural language processing, self-driving vehicles and robotics.

As machine learning becomes more mainstream, it's also becoming more and more accessible to software engineers. We're in the middle of the democratization of AI. There are numerous ways to get started, ranging from variety of open-source libraries to specific offerings by cloud providers (e.g. AWS and Azure). It's easy to get lost.

Join me in this session in which we'll cut through the machine learning jargon, and explore what options are available to you, so you can get started today!

Real-Time Data Streaming with Azure Stream Analytics

It’s imperative in today's world to be able to make split second decisions based on real-time data. Reports based on batch data are great for looking back at trends and potentially making long-term decision, but old data is in many cases already obsolete, and the opportunity to have an actionable impact on the success of a specific process may have been lost.

What if we easily could set up a near real-time data pipeline, that could be used to provide complex analytics, and make intelligent actions based on the result? Allow me to introduce Azure Stream Analytics! In this talk, we will take a closer look at the Azure Stream Analytics ecosystem, and look at real world examples streaming twitter feeds as well as sensor data from Raspberry Pi's, demonstrating how you can build your own burglar alarm.

#azure #eventhub #streamanalytics #raspberrypi #powerbi #webjobs #azurefunctions

Machine Learning made Easy - An Introduction to ML.NET

Are you a .NET Developer who has always wondered what the ML and AI hype is all about? Do you want to get involved in the community and utilize ML algorithms to make your application smarter? Join this session for a deep dive into the cross-platform open-source repository of ML.NET!

In the session we will touch upon what Machine Learning is, and how it can make our applications smarter. What ML.NET is, and why we should be excited. What some ways are in which we can make our trained models operational and how to do deep learning in ML.NET.

Beware, there will be live coding!

A Deep Dive into Machine Learning in .NET

It’s imperative in today’s world to be able to make split-second decisions based on real-time data. Reports based on batch data are great for looking back at trends and potentially making long-term decision, but old data is in many cases already obsolete, and the opportunity to have an actionable impact on the success of a specific process may have been lost. Furthermore, incorporating machine learning algorithms in your real-time data pipeline enables you to derive great insight on the fly and truly set your organization up for your success.

The best part, it is not as difficult as you may think!

In this workshop, we will cut through all the foreign jargon and give participants a solid machine learning and real-time stream processing foundation.

By the end of the workshop you will be able to:

- Understand the basics of Machine Learning and Deep Learning
- Train custom machine learning models using
- ML.NET
- AutoML
- Understand how to use Jupyter Notebooks with ML.NET
- Deploy your machine learning models to an Azure Function
- Setup a real-time data pipeline with Azure Stream Analytics
- Understand the concept of temporal windows and temporal SQL
- Make real-time predictions on a moving data stream

Prerequisites
-A laptop
- A free Azure subscription
- .NET Core SDK
- Visual Studio Code

David vs Goliath | Machine Learning in ML.NET (C#) and Scikit Learn (Python)

Machine Learning, Deep Learning, AI. New breakthroughs are made every day, continuously pushing the frontier forward. For years, Python and R have been the go-to languages to train machine learning models in. Can an object-oriented language such as C# really compete in this environment? Yes, it sure can!

In this workshop, attendees will receive hands-on experience in training custom machine learning models to detect fraudulent transactions in both C# and Python.

To demonstrate the similarities and differences, the same model will be trained using:
• ML.NET in a Jupyter Notebook and Visual Studio Code
• Scikit Learn in a Jupyter Notebook

Attendees will leave the workshop with a thorough understanding of fundamental machine learning concepts and terminology.

Prerequisites
Although it is fully possible just to follow along, please make sure to bring the following if you would like to participate in the workshop:
• A laptop
• Jupyter Notebooks (e.g. via Anaconda)
• Visual Studio Code

Link to workshop material: https://github.com/aslotte/mldotnet-real-time-data-streaming-workshop

Building Deep Neural Networks in .NET

Tensorflow, Keras, PyTorch, Theano. There are numerous ways to build deep neural networks in Python to accomplish amazing tasks such as computer vision, voice assistance and much more.

But what about in .NET? Let's explore that!

Join me in this session where we'll explore the ins and outs of deep learning and neural networks. We'll take a deep dive in to the hot open-source, cross-platform library ML.NET and build our own convolutional neural network in C# using the very latest release.

Attendees will leave with a thorough understanding of deep learning and hands-on experience accomplishing it in .NET/C#

Helm, Warp One, Engage - Set coordinates for ML.NET

It’s no exaggeration to say that Machine Learning (ML) is quickly changing the way we live. Things we are accustomed to today are things we could only dream of 10 years ago. As ML is becoming more and more mainstream, you may be asking yourself how to get started.

Historically, many of the tools and libraries used to train a model have been found in the Python and R communities (e.g. Scikit Learn, PyTorch, TensorFlow), but what if you’re a .NET Developer?

This session will be introducing ML.NET! ML.NET is an open-source, cross-platform machine learning library, specifically built to enable .NET developers to train their own custom machine learning models in C# or F#.

In this session we’ll introduce some fundamental ML concepts and terms before transitioning into the world of ML.NET and building our first model from scratch.

Beware, there will be live coding!