3rd #MLDD Expedition - This is not what I've been trained for!

Hello Everyone,

YES, you heard correctly! It's us again and we are here to invite you to our first on-site event in a really long time.

The last two years were quite challenging for everyone, to say the least and while we tried online meetups with MLDD at the end of 2020 with some success, we never really grew fond of the idea of continuing online.

But since vaccination numbers are high enough and spring is around the corner we feel it is safe to meet in person again. And not just anywhere: We'll meet 10 floors above the buzzing ground of Dresden Neustadt with an exceptional view across the whole city.

But of course we won't be there to take a city tour. Instead, Luca Biemann from elevait[1] will discuss what happens when machine learning classifiers have to face input that is very different from the samples they saw during training. According to Luca, this happens quite frequently and can lead to silent misclassification, with potentially high confidences. To quote his abstract: "Prepare your models for the messy world out there, out of domain detection is a problem that concerns everyone!"

We can't tell you how much we are looking forward to seeing you all again!
Best,
Alex & Stefan

[1] elevait represents a team of experts from all over the world. United at two locations in Triberg and Dresden, half of all employees at the development site in Dresden work in the field of machine learning (ML). They continuously develop the concept of knowledge- and ML-based automation of business processes. Their elevait Suite builds on the paradigms of abstraction and generalization of data, state of the art ML algorithms in the fields of computer vision and natural language processing, data-driven workflows, and microservices.