Tip:
Highlight text to annotate it
X
to go into some detail let's take a look at how one might do
locality sensitive hashing
for fingerprint back
this example is covered in rhizome and coleman
so you can read about it there as well
fingerprints match
if they're minutiae mac
minutiae is
a technical term in the world of fingerprints
which i don't claim to understand great detail things like
islands ridge's bifurcations quarters crossover as these are
drums used in that field
we won't get into the details of the different types of leadership
instead with only worry about weather
particular print
has
some minutiae
a particular craig point
so we assume
grid of cells to be placed on the fingerprint and
we only measuring whether or not
a cell contains i mean you ***
of some kind
now we define a function
af of x
for any print x
which is one
if the print has minutiae
unspecified
set of kate my positions
so the function of
depends on these
particular k permissions
millions he equals three
and example going forward
noticed that the function f
is dependent on the choice of decay grid positions if you choose a different set
of new positions you need a different function
now let's consider
the probability that any print
has abrupt has then you shake
a particular position
let that be p
not every print has minutiae in every position
but across all possible prints
a particular position
as minutiae a would probably be p let's assume that that distribution as
uniform and the brody's beat
not necessarily a
uh... graders oceana but for example it will suffice
the probably gave that as alexis equal to one
that is that
print x has meenu shaking in all k positions
peers about k
so if you take
the probably be dead
a particular sale has
minutiae
these point two
the probably do that
a set of
three chosen cells
all has initiate is obviously point to race to follow three
which is point zero zero eight
now let's consider
another print y
but from the same person
quite likely that this print will have minutiae in the same position that
blacks
snot always the case so that's a sign of probability q
will be quite high
that the print y
we'll have near shea
if x also does in a particular red cell
now let's look at
the function f
what is the probably beat that
af of x
is one
an f_ of why is also one
well first of all of the problem with the
that f_ of x_ is one the speed of the k so
that needs to happen
but then
you have the probably believe that
why also has to happen you sharing the same k positions so u multiplied by
k times
so you get he q to decay
if you use point nine that means there's a ninety percent chance that
why will have new chain excel so does
in particular cell
then he came to the
three
works out to point zero zero six
that's not so nice that uh...
both
x and y you get
uh... yes match with the function af
is only
happening with the problem of the
point zero zero six