Okay, quiz seems to have ended.

It's very interesting to see the study groups that are kind of forming in the audience.

Some of you have time for group work.

It's not exactly how it's planned, but just make sure that you actually can do it yourselves.

Because group work in the exam is being frowned upon and penalized.

I would like to make two announcements. One is concern study groups. Some of you have already seen this.

We have a new feature. We go to the AI course and you go to study body.

This is a kind of a sign up service, a very basic one at the moment, for forming study groups.

Because there are still people who are coming to me and saying, how do I get somebody to study with?

This is one way. You can just basically give a little bit of information.

Your email, something, I'm in AI, I'm I think in my something like 10th semester at least,

preferred Wednesday and Tuesday, I don't know, and I can do, who knows what I can do.

I understand that this data is being transferred, but only for the purposes of making study bodies.

Okay, we've made progress.

Then there is a profile and I have a couple of people who are looking for a study body.

Actually quite a few of you and I could just basically make a connection here.

Then they can make a connection and then we can actually become a study group.

Okay, I think you know the principle from somewhere else, whereas just swiping left or right, it's the same idea.

Eventually we'll try to match people upon their, based on their competencies and so on,

and you'll see that you're somewhat evenly matched, which is where learning actually profits most.

But if you are in search for a study group, then you can actually use this feature.

By the way, if you're trying to hook me up as a study buddy, Abhishek has promised to, if I give the demo, delete my profile directly afterwards.

So even if I contact with you now, that'll go away.

Okay, so that's this feature, study buddy, and feel free to use it.

We've done it for you. Okay, good. So that's one thing.

I wanted to say something else, yes.

There was an incident in the exam review where one of the students actually,

well, where something happened that we've been suspecting in old exam reviews all the time, that somebody actually

put in, corrected their exam sheet during the review.

I know it's tempting. I just wanted to tell you we have the scans, so now we can prove it.

This person is in trouble. It could actually lead to ex-metriculation.

And even more, it's an offense against the law.

Changing documents, changing official documents, and your AI exams are such, is actually

sanctioned by the penal code, which is not something you want.

Okay, so just to warn you, we can prove it now. It's easy.

Okay, any questions?

Oh yes, and a big thank you to all the people who are sending in corrections and typos and so on in the alias system.

Just to show you, if you go to the notes, there's a typo here, and you can just basically report a problem,

then this makes a good lab issue, which I will try to fix as soon as possible.

And to everybody who is complaining that you have everything printed on top of each other,

there were quite a few of you, thank you very much, but we've fixed it mostly.

Except for the picture with the Monte Carlo tree search, which is so complex I haven't been able to wrap my head around how it's actually done

to integrate it into a better presentation.

So with the next update, you should be seeing better micro slide shows in there.

Yes, thank you to everybody who's contributing to the quality of the slides.

Are there any questions so far?

So remember, we're somewhere near the end of constrained satisfaction problems.

Remember, constrained satisfaction problems are, if you will, search problems in which we have a factored representation of the states,

which allows us to lift the steps from single actions to the action groups induced by the, you could think of this as a meta action,

of choosing a variable, giving it a value.