We now turn to the questions of generating good approximations for
term approximation from a general dictionary We shall assume
that the dictionary
is complete in the Hilbert
space
. This means that every element in
can be approximated arbitrarily well by linear combinations of the
elements of
. Since the dictionary is no longer an
orthogonal basis as was considered above, we need to revisit howto find good
term approximations. Because of redundancy
within the dictionary, we cannot simply pick the largestcoefficients as we saw with a basis. Greedy algorithms are a
method to generate good
term approximations.
- General Greedy Algorithm
Given
, we want to generate an
-term approximation to
.
The general steps are as follows:
- Initialize: (approximation)
, (residual)
,
approximation collection
- Search
for some
, then add
to the
set
.
- Use
to find new
approximation for
.
At stage
, we have
,
, and
.
There are many types of greedy algorithms. We describe the threemost common in the case
is a Hilbert space. However,
there are anlaogues of these for
.
- Pure Greedy Algorithm (PGA)
Note:>From
choose
(the
that causes the largest inner product).
This method is similar to a steepest decent algorithm for
decreasing the error.
- Orthogonal Greedy Algorithm (OGA)>From
choose
as in the PGA.
where
denotes the orthogonal projection onto the space
.
We can find
by solving the linear system of
equations
Then,
.
- Relaxed Greedy Algorithm (RGA)>From
choose
in some way (for example, our earlier methods) and then define
Unlike PGA, here we do not make a full step in the correct
direction. For example, one way to proceed is to define
This type of greedy algorithm is known to perform the best as
compred with the previous two.
Given
,
, it is not practical to
minimize
by searching over all the
possibilities.The greedy approximation gives an
-term
solution with less computation, but does it perform well?
Let
where the smallest
is the
norm of
.
For OGA or RGA as described above, we have
Remark 5 This is similar to
(
-term approximation) but its
not always quite as good.