a simple implementation of multiple-layered neural network in cl.
Common Lisp
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
cl-simple-neuralnet
include
src
t
.gitignore
README.org
bp-ans.png
bp-fn2-ans.png
bp-fn2.png
bp.png
cl-simple-neuralnet-test.asd
cl-simple-neuralnet.asd
cl-simple-neuralnet.texinfo
doc.html
doc.org

README.org

Cl-Simple-Neuralnet

a simple implementation of multiple-layered neural network.

Usage

(defun fn3 (x y z w)
  (values
     (sigmoid1 (* x y z w))
     (sigmoid1 (- (+ x y z) w))))

(defvar *nn*
	(bp-teach #'fn3 '(4 10 15 2) :iteration 100000))

(let ((in (make-input 0.5d0 0.8d0 0.2d0 0.1d0)))
  (j-at in (make-output-from-input #'fn3 in) *nn*))
;; --> 0.016671963009814696d0

(let ((in (make-input 0.5d0 0.3d0 0.2d0 0.1d0)))
  (j-at in (make-output-from-input #'fn3 in) *nn*))
;; --> 0.005211269116245283d0

Installation

asdf-loadable. just (require :cl-simple-neuralnet)

Comments

Part of a project in the university. It’s just a hack in two weeks or so.

documentation is not yet available because I’m not so into it. just read teh source.

Author

Masataro Asai (guicho2.71828@gmail.com)

Copyright

Copyright (c) 2012 Masataro Asai (guicho2.71828@gmail.com)