The primary topics in this part of the specialization are: greedy algorithms (scheduling, minimum spanning trees, clustering, Huffman codes) and dynamic programming (knapsack, sequence alignment, optimal search trees).
Correctness of Clustering Algorithm
Loading...
Del curso dictado por Stanford University
Greedy Algorithms, Minimum Spanning Trees, and Dynamic Programming
514 calificaciones
De la lección
Week 2
Kruskal's MST algorithm and applications to clustering; advanced union-find (optional).
Conoce a los instructores
Tim RoughgardenProfessor
Computer Science
This video will prove the correctness of our greedy algorithm for clustering.
We'll show that it maximizes the spacing over all possible K clusterings.
You might have hoped that we could deduce the correctness of this greedy algorithm
for clustering immediately from our correctness proofs for various greedy
minimum spanning tree algorithms. Unfortunately that doesn't seem to be the
case. In the minimum cost spanning tree
problem, we're focusing on minimizing the sum of the edge cost.
Here we're looking at different objective, maximizing the spacings.
We do need to do a proof from scratch. That's said, you know, the arguments
we'll use should look familiar to you not just from the sort of exchange type
arguments when we prove the cut property, but also it might remind you even more,
going back further, to our greedy algorithms for scheduling.
So let's now set up the notation for the proof.
As usual, we're going to look at the output of r algorithm.
It achieves some objective function value, some spacing.
We're going to look at an arbitrary competitor.
Some other proposed scheduling. We're going to show that we're at least
as good, our spacing is, at least, as large.
So specifically, we'll denote the clusters in the output of r algorithm by
C1 up to CK. Our clustering has some spacing, some
distance between the near, closest pair of separated points.
Call it capital S. We're going to denote our competitor,
some alternative K clustering by C-Hat one of the C-Hat K, what is it that we're
tryin to show? We want to show that this arbitrary other
clustering has spacing no larger than R's, if we can show that, then because
this clustering was arbitrary it means the greedy clustering has spacing as
large as any other, so it's maximizing the spacing, that's what we want to
proof. But differently we want to exhibit a pair
of points separated by this cluster and C one-half to C1K, such that the distance
between those separated points is S or smaller.
So, let me just quickly depose of a trivial case.
If the C hats are the same as the C's, possibly up to a renaming, then of course
exactly the same pairs of points are separated into each of the clustering, so
that the spacing is exactly the same. So that's not a case we have to worry
about. The interesting case, then, is when the c
hats differ fundamentally from the cs, when they're not merely a permutation of
the clusters in the greedy clustering. And the maneuver we're going to do here
is similar in spirit to what we did in our scheduling correctness proof.
Way back in our scheduling correctness proof, we argued that any schedule that
differs from the greedy one, suffers from, in some sense, a local flaw.
We identified an adjacent pair of jobs that was, in some sense, out of order
with respect to the greedy ordering. The analog here is, we're going to argue
that, for any clustering which is not merely a permutation of the greedy
clustering. There has to be a pair of points which is
classified differently in the c hats relative to the c's.
By differently, I mean they're clustered together in the greedy clustering.
These points, p and q, belong to the same cluster, c sub i.
Yet, in this alternative clustering, which is not just the permutation of the
greedy clustering. They're placed in different clusters.
One, maybe p and c hat i, and q and some other c hat j.
So I want to now split the proof into an easy case and a tricky case.
To explain why the easy case is easy lets, lets observe a property that this
greedy clustering algorithm has. Now the algorithm's philosophy is that
the squeaky wheel should get the grease. That is, the separated pair of points
that are closest to each other are the ones that should get merged.
So for this reason, because it's always the closest separated pair that get
merged, if you look at the sequence of point pairs that get merged together,
that determine the spacing in each subsequent iteration, the distances
between these sort of worst separated points is only going up over time.
At the beginning of the algorithm, the closest pair of points in the entire
point set are the ones that get directly merged.
Then those are out of the picture, and now that some further away pair of
points are separated, it determines the spacing,
then they get merged. Once they've been coalesced, then there
is still some further away pair of points, which is now the smallest
separated. They get merged, and so on.
So if you look at the sequence of distances between the pairs of points
that are directly merged by the greedy algorithm, that is only going up over
time. And this sequence culminates with the
final spacing S of the greedy algorithm. At some sense, the spacing of the output
of the greedy algorithm is the distance between the point period that would get
merged if we ran the greedy algorithm one more in moderation but unfortunately
we're not allowed to do that. Okay?
So the point is, for every pair of points directly merged by the greedy algorithm,
they're always a distance at most S away from each other.
So the easy case, then, is when this pair of points, pq,
which, on the one hand, lie in a common greedy structure,
but on the other hand, in different clusters with c hats.
If they were, at some point, not merely in the same cluster, but actually
directly merged by the greedy algorithm. If, at some iteration, they determined
the spacing, and were picked by the greedy algorithm to have their, clusters
merged. Then we just argued that the distance
between p and q is no more than the space in capital s of the greedy clustering.
And since p and q lie in different clusters of the c hats.
It's separated by the C hats and therefore they upper bound the spacing of
the C hats. Maybe there's some even closer separated
pair by the C hats. But the very least P and Q are separated
so they upper bound the spacing of the C hat clustering.
So that's what we wanted to prove. We wanted to show that this alternative
spacing didn't have better spacing than our greedy spacing.
It had to be at most as big. It had to be at most capital S.
So in this easy case, when P and Q are directly merged by the greedy algorithm,
we're done. So the tricky case is when P and Q are
only indirectly merged, and you may be wondering at the moment, what does that
mean? How did two people wind up in the same
cluster if they weren't, at some point, directly merged?
So let's draw a picture and see how that can happen.
So the issue is that two points P and Q might wind up in a common greeting
cluster, not because the greedy algorithm ever
explicitly considered that point pair, but rather because of a path or cascade
of direct mergers of other point pairs. Imagine, for example, that at some
iteration of the greedy algorithm the point P was considered explicitly along
with the point A1, where here A1 is meant to be different than Q.
So that's a direct merger, and P and A1 wind up in the same cluster. Their
clusters are merged. Maybe the same thing happened to the
point Q at some point A sub L which is different than P.
Sooner or later maybe, you know, at some other time, some totally unrelated pair
of points A2 and A3 are directly merged and then at some point A1 and A2 are
considered by the greedy algorithm. Algorithm, because the other closest pair
of separated points, and, they get merged.
And so on. So the edges in this picture are meant to
indicate direct mergers, pairs of points that are explicitly fused because they
determine the spacing of some point of the greedy iteration.
But at the end of the day the greedy clustering is going to have the results
of all of these mergings. So in case you're feeling confused, let
me just point out that we really saw this exact same exact thing going on when we
were talking about minimum spanning trees in Kruskal's rhythm.
So, at an intermediate point in Kruskal's rhythm, after it's added some edges, but
before it's constructed a spanning tree. As we discussed, the intermediate state
is a bunch of different connected components.
And there are vertices that Have an edge chosen between them.
They, of course, are going to be in the same kinetic component.
But then again, a kenetic component could have long paths in it.
So you could have vertices that are in the same kinetic component in an
intermediate state of Kruskal's Algorithm, despite the fact that we've
haven't chosen an edge directly between them.
There's rather, a path of chosen edges between them.
It's exactly the same thing going on here.
Now, what we have going for us is that, if a pair of points, as discussed, was
directly merged, we know they're close. The distance between them is, at most,
this spacing, capital S. We really don't know anything, frankly,
about the distance between pairs of vertices that were not directly merged.
They just, sort of, accidentally wound up in a common cluster.
But this turns out to be good enough. This is actually sufficient to argue that
this competitor clustering with the c-hat has spacing no more than s?
No better than ours. Let's see why.
So given that P and Q are in a common greedy cluster it must mean there was a
path of direct mergers that forced them to be in the same cluster.
So let's let the intermediate points involved in that path denoted A1 of two
AL. So here's the part of the proof where we
basically reduce the tricky case to the easy case.
So we've got this pair of points, PQ. Now, remember, not, not only are they in
a common greedy cluster. But they're in different clusters in our
competitor in the c hats. So the point p is in some cluster.
Call it c hat i. And Q is in something else.
In particular, it's not in c hat i. Now, imagine you go on a hike.
You start at the point p, and you hike along this path.
You traverse these direct mergers toward q.
Now, you're starting inside c hat I, and you end up outside.
So at some point on your hike, you will traverse the boundary.
You will, for the first time, escape from c hat I, and wind up in some other
cluster. So that has to happen.
And let's call ai and ai+1 the consecutive pair of points at which you
go from inside this cluster to outside this cluster.
And now we're back in the easy case. Now we're dealing with a separated pair
that would directly merge by the greedy algorithm.
Remember that we set up this path to be a path of direct mergers so in particular,
AJ and AJ + one were direct mergers, therefore their distance is at most S.
And again, by virtue of being direct mergers, their distance is at most the
spacing of the greedy clustering and yet as a separated point by the C hats.
It's also an upper bound on the spacing of the C hats.
This means the spacing S of our greedy clustering is as good as the competitor.
Is the competitor was arbitrary or optimal.
That completes the proof.
Explora nuestro catálogo
Inscríbete de manera gratuita y obtén recomendaciones personalizadas, actualizaciones y ofertas.
Coursera brinda acceso universal a la mejor educación del mundo, al asociarse con las mejores universidades y organizaciones, para ofrecer cursos en línea.
© 2018 Coursera Inc. Todos los derechos reservados.
