Small data, great insights: ML/DL tricks with restricted data // ONLINE on 23,24 November 2020
!!!!!! The workshop is going to be held ONLINE on 2 afternoons: 1st session is on 23 November, the 2nd session is on 24 November. On both days we start at 13.30 and finish around 17.00 UTC+1 !!!!!
In this workshop Miklós and Levente (the workshop leaders) demonstrate a wide variety of techniques combatting "small data", and also along the way try to draw some conclusions about the generalization abilities of (deep) machine learning models, as well as demonstrate with hands on exercises the application of these methods.
With all the talk about big data, in practical use cases we very often find ourselves in settings, where specific data is very limited, so we have to resort to special techniques to be able to train reasonably performing models. In this workshop - building on their hands on experience as well as theoretical reflections accumulated during teaching and mentoring - Miklós and Levente demonstrate a wide variety of techniques combatting "small data", and also along the way try to draw some conclusions about the generalization abilities of (deep) machine learning models, as well as demonstrate with hands on exercises the application of these methods.Who is this for: data scientists with practical experience, who would like to broaden the palette of their tools and gain some theoretical insights as well.
Good command of Python and its modeling ecosystem, some experience and knowledge in machine learning and deep learning systems.
MLOps best practices with MLflow // ONLINE on 2,9 December 2020
!!!!!! The workshop is going to be held ONLINE on 2 afternoons: 1st session is on 2 December, the 2nd session is on 9 December. On both days we start at 13.30 and finish around 17.00 UTC+1 !!!!!
Tracking and productionizing Machine Learning models has never been more important than today. As part of the Data Engineering / Data Science work, you will need to track ML experiments, manage the lifecycle of your models, productionize your models and serve them. MLflow, an open source Model Management tool is here to help you with all of these operations. In this class you will be able to acquire hands-on knowledge about MLOps and MLflow.
This hands-on course is about the management of Machine Learning models.
As you develop your ML applications, several questions arise:
- How do I track experiments in the research phase?
- How do I publish my models?
- How can a Data Scientist hand over a model to a Data Engineer?
- How can you productionize a model and track versions?
- What’s the lifecycle of a machine learning model and how can I track and manage it?
- How can I serve a trained model either through my own application or through docker?
In this course, you will learn about these topics. We are taking a look at the general MLOps tasks and use a popular open-source tool, MLflow for tackling the problems that arise when you develop and productionize your machine learning models.This course is completely hands-on and all the code will be distributed to participants in a self-contained plug-and-play to work format.
- Basic Python Knowledge
- Firefox or Chrome web browser with unrestricted internet access
Who is this class for?
Data Engineers, Data Scientists, BI Analysts and Developers
At the end of this course, you will be able to track, manage, productionize and server Machine Learning Models with MLflow. You will also have an understanding of the problems that are faced by conducting these MLOps operations.