[Gzz-commits] manuscripts/Paper paper.tex

[Tuomas J. Lukka]

CVSROOT:        /cvsroot/gzz
Module name:    manuscripts
Changes by:     Tuomas J. Lukka <address@hidden>        03/03/31 07:29:22

Modified files:
        Paper          : paper.tex 

Log message:
        Twids

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/gzz/manuscripts/Paper/paper.tex.diff?tr1=1.86&tr2=1.87&r1=text&r2=text

Patches:
Index: manuscripts/Paper/paper.tex
diff -u manuscripts/Paper/paper.tex:1.86 manuscripts/Paper/paper.tex:1.87
--- manuscripts/Paper/paper.tex:1.86    Mon Mar 31 06:46:05 2003
+++ manuscripts/Paper/paper.tex Mon Mar 31 07:29:22 2003
@@ -45,8 +45,6 @@
 
 NO SEMANTIC CORRELATIONS!!!!!!!
 
-MOTIVATING EXAMPLE NOT YET USER-TESTED
-
 IMAGE: RENDERING MODES, TEXTURE COORDINATES!
 
 SPARSE CODING: A TEXTURE CONTAINING BOTH YELLOW TRIANGLE AND
@@ -544,9 +542,9 @@
 
 %XXX: shorten by one half column 
 
-In this section, we discuss the methods to generate unique
-background textures on an abstract level.
-
+In this section, we discuss the principles underlying
+our algorithm to generate unique
+background textures.
 To be useful, the unique backgrounds should be easily
 distinguishable and recognizable, and should not
 significantly impair the reading of black text on top of it.
@@ -560,14 +558,8 @@
 In order to design a distinguishable distribution of textures,
 we have to take into account the properties of the human
 visual system.
-% The seed for randomly choosing
-% an easily distinguishable unique background from a
-% distribution based on 
 
-%providing an infinite source of unique backgrounds.
-%generating textures based on seed numbers [identity]
-
-The simple model we use here assumes
+The simple model of texture perception we use assumes
 that at some point,
 the results from the different pre-attentive feature detectors,
 such as different shapes and colors, 
@@ -579,6 +571,9 @@
 in a simple perceptron-like 
 fashion\cite{rosenblatt62neurodynamics,widrow60adaptive}.
 This configuration is commonly used in neural computation.
+As seen for example in \cite{olson02vstm}, only a limited
+number of different features detected can be grouped into 
+objects, ... XXX SO!!! JVK
 
 \begin{figure}
 %\fbox{\vbox{\vskip 3in}}
@@ -665,8 +660,12 @@
     or if there were no curved lines, we would be wasting
     recognition potential by leaving some elements
     of the feature vector always zero.
+    However, the results cited above\cite{olson02vstm}  also
+    indicate that in {\em one} texture, only a limited range
+    of features should be used.
+
 
-Indeed, the entropy is maximized when the features are distributed
+Generally, the entropy is maximized when the features are distributed
 independently from each other:
 features orthogonal to human perception 
 (e.g.,~color, direction of fastest luminance change)