Permalink
Find file
Fetching contributors…
Cannot retrieve contributors at this time
36 lines (23 sloc) 1.03 KB
@findex bp-teach
@anchor{Function bp-teach}
@deffn {Function} {bp-teach}
bp-teach (fn, nodes, &key iteration, nn) @code{->} nn
@code{fn} @code{:} the target function. @code{( input-arguments* -> output-arguments* )}
@code{iteration} @code{:} a =fixnum=
@code{nodes} @code{:} (@{ number-of-nodes-in-layer @}*)
@code{number-of-nodes-in-layer} @code{:} a =fixnum=
@code{nn} @code{:} an instance of =neural-network=
let I @code{=} [0.0d0,1.0d0] .
function @code{fn} should accept n =double-float= arguments within I
and is expected to return m =double-float= values also within I.
n should match the first =fixnum= in the @code{nodes} , and m should
match the last =fixnum= in the @code{nodes}.
if @code{nn} is unspecified, @code{nodes} argument is used to create
a new instance of =neural-network=
otherwise, @code{nodes} will be ignored and it will
conduct further teaching on @code{nn}.
@code{iteration} determines iteration number of
back-propagation algorhithm, defaulted to 10000.
restarts:
@code{+} /Restart/ add-iteration (new-i)
@end deffn