Ex parte Kovacs

Board of Patent Appeals and Interferences

Jun 21, 2001

08373937 (B.P.A.I. Jun. 21, 2001)

According to the examiner (paper number 12), the amendment had the1
effect of overcoming the rejections under 35 U.S.C. § 112.
1
The opinion in support of the decision being entered today was not written for
publication and is not binding precedent of the Board.
Paper No. 19
UNITED STATES PATENT AND TRADEMARK OFFICE
_____________
BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
_____________
Ex parte ZSOLT T. M. KOVACS
_____________
Appeal No. 1998-2718
Application No. 08/373,937
______________
ON BRIEF
_______________
Before HAIRSTON, JERRY SMITH, and FLEMING, Administrative
Patent Judges
HAIRSTON, Administrative Patent Judge.
DECISION ON APPEAL
This is an appeal from the final rejection of claims 14
through 31. In an Amendment After Final (paper number 11),1
claim 27 was amended.
The disclosed invention relates to a method and apparatus
Appeal No. 1998-2718
Application No. 08/373,937
2
that uses a plurality of classifier networks that are neural
networks. Feature vectors are generated from a scanned
character image, and each of the neural networks in the
classifier network is configured to receive at least two
feature vectors.
Claim 14 is illustrative of the claimed invention, and it
reads as follows:
14. Apparatus for recognizing an alphanumeric
character from an image of said character, said
apparatus comprising:
a first classifier network having a first input
terminal coupled to receive a first feature vector
of said character, a second input terminal coupled
to receive a second feature vector of said
character, and a plurality of output terminals;
a second classifier network having a first input
terminal coupled to receive said second feature
vector, a second input terminal coupled to receive a
third feature vector of said character, and a
plurality of output terminals;
a third classifier network having a plurality of
input terminals coupled to said pluralities of
output terminals of said first and second classifier
networks, and a plurality of output terminals
operable to carry a statistical value corresponding
to a predetermined classification of said image; and
wherein said third classifier network is
operable to carry out consecutive statistical
Appeal No. 1998-2718
Application No. 08/373,937
The last paragraph on page 7 of the answer makes clear that claim 212
should be included under this rejection.
3
operations on information received from said first
and second classifier networks until said
statistical value is generated.
The references relied on by the examiner are:
Fukumizu 5,060,278 Oct. 22,
1991
Lincoln 5,155,801 Oct. 13,
1992
Matsuba et al. (Matsuba) 5,255,347 Oct.
19,
1993
Claims 14, 18, 19, 22 and 25 stand rejected under 35
U.S.C. § 102(b) as being anticipated by Fukumizu.
Claims 15 through 17 and 26 through 29 stand rejected
under 35 U.S.C. § 103(a) as being unpatentable over Fukumizu
in view of Matsuba.
Claims 20, 21 , 30 and 31 stand rejected under 35 U.S.C. 2
§ 103(a) as being unpatentable over Fukumizu in view of
Lincoln.
Claims 23 and 24 stand rejected under 35 U.S.C. § 103(a)
as being unpatentable over Fukumizu.
Appeal No. 1998-2718
Application No. 08/373,937
4
Reference is made to the briefs and the answer for the
respective positions of the appellant and the examiner.
OPINION
We have carefully considered the entire record before us,
and we will reverse all of the rejections of record.
Turning first to the 35 U.S.C. § 102(b) rejection of
claims 14, 18, 19, 22 and 25, the examiner is of the opinion
(answer,
page 5) that each of the neural networks (i.e., NET 0 through
NET 9) in Fukumizu (Figure 1) has “a first input terminal
coupled to receive a first feature vector of a character and a
second feature vector of a character (col. 3, lines 44-57 and
fig. 1, element 10).” Appellant argues (brief, page 10; reply
brief, page 4) that each of the neural networks NET 0 through
NET 9 in Fukumizu is configured to receive “one and only one
feature vector at a time.”
Inasmuch as Fukumizu clearly teaches (Abstract; column 2,
line 39; column 3, line 48; column 9, line 61; column 10, line
33; column 11, lines 7 and 54; column 12, line 34; and column
Appeal No. 1998-2718
Application No. 08/373,937
5
13, lines 13 and 61) that each of the neural networks receives
one feature vector, we agree with appellant’s argument. Once
that single feature vector is received at each of the neural
networks, multiple input units within each of the neural
networks are used to process the feature vector (answer, page
9). We likewise agree with appellant’s argument (reply brief,
page 4) that “it is clear that neither Figure 8 nor any other
section of Fukumizu discloses a neural network that receives
multiple feature vectors in parallel.” For this reason, the
35 U.S.C. § 102(b) rejection of claims 14, 18, 19, 22 and 25
is reversed.
For the same reason, the 35 U.S.C. § 103(a) rejection of
claims 23 and 24 based upon the teachings of Fukumizu is
reversed.
Turning next to the 35 U.S.C. § 103(a) rejection of
claims 15 through 17 and 26 through 29, appellant argues
(brief, page 13) that :
Matsuba at most suggests providing more of
Fukumizu’s preprocessing blocks 13 to extract
additional feature vectors, and then providing more
of Fukumizu’s network groups NET0-NET9 to each
process one and only one of these additional feature
Appeal No. 1998-2718
Application No. 08/373,937
6
vectors. Thus, it is clear that because Matsuba
does not supply the teaching missing from Fukumizu--
namely processing multiple feature data in parallel
with a single neural network--the combination of
these references does not and cannot render claim 26
obvious.
We agree with appellant’s argument. Accordingly, the 35
U.S.C.
§ 103(a) rejection of independent claim 26 and dependent
claims 15 through 17 and 27 through 29 is reversed.
Turning lastly to the 35 U.S.C. § 103(a) rejection of
claims 20, 21, 30 and 31, appellant argues (brief, page 14)
that “the neural networks 11, 12, and 13 of Lincoln’s Figure 1
. . . are, like Fukumizu’s neural networks, each coupled to
receive one and only one feature vector, and thus also cannot
process a
respective pair of multiple feature vectors.” We agree. In
summary, the 35 U.S.C. § 103(a) rejection of claims 20, 21, 30
and 31 is reversed.
DECISION
The decision of the examiner rejecting claims 14, 18, 19,
Appeal No. 1998-2718
Application No. 08/373,937
7
22 and 25 under 35 U.S.C. § 102(b) is reversed, and the
decision of the examiner rejecting claims 15 through 17, 20,
21, 23, 24 and 26 through 31 under 35 U.S.C. § 103(a) is
reversed.
REVERSED

KENNETH W. HAIRSTON )
Administrative Patent Judge )
)
)
) BOARD OF PATENT
JERRY SMITH ) APPEALS AND
Administrative Patent Judge ) INTERFERENCES
)
)
)
MICHAEL R. FLEMING )
Administrative Patent Judge )
kwh/vsh
Appeal No. 1998-2718
Application No. 08/373,937
8
BRYAN A. SANTARELLI
SEED AND BERRY
6300 COLUMBIA CENTER
701 FIFTH AVENUE
SEATTLE, WA 98104