Let's start. Hello everybody. Can everybody hear me? Because I don't know if this microphone

is working or not. Okay, let's see. So the lecture will be recorded and so I hope this

microphone works. Okay, good. So hello everybody. I'm Vincent Christlanser. I'm also on that

slide here in the long list of slide authors. And I'm in replacement here for Andreas Meyer,

who is currently on a conference trip and also next week probably he won't be here because

he has a health issue. Okay, so we are actually quite a big team. So some of them are also

here. Maybe you can just come quickly in front. So some of the tutors are here. Katharina

Breininger is I think also on conference trip. Suleyman is, can't see him. Okay. So the bottom

row is basically from our lab and the top row are apart from Andreas Meyer of course

are student tutors who will help you in the tutorials. Yeah, Noah and Florian are currently

here and the others you will get to know and will meet in the tutorials. Thanks. It was

just a short time so that you know their faces. Okay, yes. Good. Yeah, so as I said I'm currently

replacing Andreas Meyer. I'm also a postdoc at the lab and so for the next week probably

maybe also in two weeks I will replace him. And if Andreas cannot come then typically

Katharina Breininger will replace him or I will replace him so that not too many changing

faces are here. Okay, so you probably all heard of deep learning I guess at least. Otherwise

you won't be here. And it's always difficult to characterize or where actually to put deep

learning in terms of related fields. So big data is another big password, artificial intelligence,

machine learning, representation learning. And deep learning is basically, yeah, it needs

big data, it's a part of artificial intelligence. It's of course not only, so it's not artificial

intelligence, it's not a subset of deep learning. And machine and deep learning is also a subset

of machine learning and representation learning. So don't get confused if someone talks about

AI, so all these big companies, it's often not AI but just simple, normal pattern recognition

or what do you mean normal, yeah. It's also great but it can also be a logistic regression

or something like that, yeah. So however all these terms they change over time, nowadays

everybody talks about artificial intelligence again but actually means deep learning. So

yeah, it's a little bit complicated also for us as researchers. How do we frame now our

work? Is it now artificial intelligence? I think it's always difficult and not easy also.

For the lab for example, yeah, we had this old artificial intelligence which worked on

symbolic representations and then we had the pattern recognition on the other side and

machine learning and now it's again artificial intelligence which means completely something

completely different. Then you can characterize deep learning of course by the tasks, so classification

task, we have segmentation task, regression generation, detection task whatsoever and

we have different categories, so supervised versus unsupervised learning and neural networks

are of course a basic part of deep learning. Deep learning nearly always contains neural

networks because there you have these deep layers, so multiple layers and they are typically

represented by neural networks and we have features, of course feature learning is also

a part of deep learning. Okay, so today's lecture we will have a small or well quite big motivation

actually so all these slides are today. The main part is about the motivation and in general

what we are doing at our pattern recognition lab and then we come a little bit theory but

it's really tiny today, so the big theory part will be next week and yeah and then we

have some in the end we have some organizational matters about the tutorials and the lecture

and the exams and so on. Okay, let's start with a small motivation. I don't know, is

any one of you having stocks? It's a good time actually but yeah I was actually we have

this graph already several times now here in this deep learning lecture and we give

this lecture since 2016-17 and it always raised until about 2019. Okay, and has someone an

idea why it drops here the Nvidia stock market? So Nvidia, biggest GPU producer? Bitcoin,

yes, thank you and very good, exactly. So here we have the drop of Bitcoin, so it's

not all about deep learning, yeah. But of course it is a part here and when I saw these

graphs I always was annoyed by myself why I have not bought some Nvidia stocks. So yeah,