Dear students

welcome to this first lecture of the seminar advances in deep learning for

time series.

My name is Dario Zanka.

I'm the head of the applied machine learning group at the machine learning and data analytics

lab here at FAU.

Together with Giteen and Christopher

I will be giving you lectures and supervising you

on projects during the semester.

So I'm from FAU as I said

also Giteen is from the same lab

but you may notice that

Christopher is from a different university from this slide.

In fact

Christopher is professor at the Pontificia Universidad Catolica de Valparaiso in Chile.

This seminar is provided to you by the joint effort of the two universities.

The seminar is also offered to students in Chile and therefore it's a nice interchange

of expertise and opportunity for also cultural exchanges with other countries and other universities

that I hope you will enjoy.

So both me

Giteen and Christopher will give you lectures.

So you will see our phases in the recorded lectures from FAU TV.

But let's dive a little bit more into the organizational information for this year seminar.

So the seminar is worth 5 ECTS.

It's a team-based project

which means you will be assigned to a group and you will work

in teams and push forward a project together with your teammates.

The evaluation for FAU students is 60% based on the written report and 40% based on the

oral presentation where you will present to an audience the results of your project.

Right

so this is the list of exciting topics that is covered by the seminar.

First you will have an introduction, which is today's lecture.

And then in the second lecture already we will talk about the real-world time series

datasets, the tool tracking datasets.

This is one of the datasets that you will be able to work with during this semester.

We focus on real-world datasets because we think it is important within this seminar

to develop the skills to deal with datasets coming from a real industrial setting that

presents many challenges that oftentimes are just simplified or absent in the typical time

series benchmarks.

In the third lecture we will quickly present the deep learning architectures for time series.

Due to the nature of this seminar

so the focus on advances of deep learning

we need

to assume that students have some deep learning knowledge.

And therefore we will not dive into details about those architectures

but we will give

pointers to literature in case some of those concepts are unfamiliar to you.

And we are also here to assist

so in case you need future information or future support

we will be happy to provide more content and more explanations than those given in lecture