There are no technical questions.

I would like to, what's the time?

Oh, yes, I'll just situate and then we'll make a break.

Remember our agent.

We have this kind of an agent.

An agent has a world model, which currently

is a Beijing network.

And now, essentially, we have percepts coming in.

And essentially, those give us evidence.

That's really stuff where if we have,

this is the question we want to know, what we believe

at the moment about x.

Now that we know e, that's where e comes in, essentially.

Or we compute e.

And that's really what this is the inference here.

And we have the Beijing network here.

This is half an agent.

What's missing?

Yes.

The part that makes from the x-axis and the y-axis,

right?

We need another box here that kind of closely

is connected with the world model that decides,

what am I going to do?

So we need two things here.

We need a world model, and we need a decision theory.

So now we're going to, after the break,

we're going to make a decision theory.

Yes?

In this case, the Beijing network

is not something to change.

Yes, we're not learning yet.

There must be something for July and maybe June.

Scary how much we were already done.

OK, so but yes, the Beijing network,

you're born with the Beijing network.

Completely unrealistic assumption,

but one which we'll make in order to survive.

We're going to look into learning later.

So yes, life is more complicated.

But let's assume we have a built invasion network,

and it's not changing.

And the next step will be that we're basically

changing weights and stuff like that.

But first we do decision theory.

This is the first step.

This is done.

Any questions?

Now we'll start with this.

And particularly, I've been very negligent