Ex parte Tsuboka

Board of Patent Appeals and Interferences

Feb 8, 2002

08864460 (B.P.A.I. Feb. 8, 2002)

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for publication and is not binding precedent of the Board.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. 31
UNITED STATES PATENT AND TRADEMARK OFFICE
____________
BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
____________
Ex parte EIICHI TSUBOKA
____________
Appeal No. 2000-0189
Application No. 08/864,460
____________
HEARD: Jan. 17, 2002
____________
Before KRASS, BARRETT, and BARRY, Administrative Patent Judges.
BARRY, Administrative Patent Judge.
DECISION ON APPEAL
The examiner rejected claims 1-13. The appellant appeals
therefrom under 35 U.S.C. § 134(a). We reverse.
BACKGROUND
The invention at issue in this appeal is a Hidden Markov
Model (“HMM”) generator that is applicable to such pattern
recognitions as speech recognition. The HMM generator
includes a vector quantizing means for quantizing vectors of a
training pattern having a vector series and converting the
Appeal No. 2000-0189 Page 2
Application No. 08/864,460
vectors to a label series of clusters to which they belong. A
continuous distribution probability density HMM generating
means generates a continuous distribution probability density
HMM from a quantized vector series corresponding to each label
of the label series. Incidence of a label in each state is
calculated from the training vectors classified in the same
clusters and the continuous distribution probability density
HMM.
Claim 1, which is representative for present purposes,
follows:
1. An HMM (Hidden Markov Model) generator,
comprising:
vector quantizing means for generating a model
by quantizing vectors of a training pattern having a
vector series, and converting said quantizing
vectors into a label series of clusters to which
they belong,
continuous probability distribution density HMM
generating means for generating a continuous
probability distribution density HMM from a
quantized vector series corresponding to each label
of said label series of clusters, and
label incidence calculating means for
calculating the incidence of the labels in each
state from said quantizing vectors of a training
pattern classified in the same label series of
Appeal No. 2000-0189 Page 3
Application No. 08/864,460
clusters and the continuous probability distribution
density HMM.
(Appeal Br. at A1.)
The prior art applied by the examiner in rejecting the
claims follows:
Thomson 5,023,910 June 11, 1991
J. N. Holmes, Speech Synthesis and Recognition, 124, 125,
142, 143 (Chapman & Hall 1988).
Claims 1-13 stand rejected under 35 U.S.C. § 103(a) as obvious
over Holmes in view of Thomson.
OPINION
After considering the record, we are persuaded that the
examiner erred in rejecting claims 1-13. Accordingly, we
reverse.
Rather than reiterate the positions of the examiner or
appellant in toto, we address the main points of contention
therebetween. The examiner makes the following assertions.
Appeal No. 2000-0189 Page 4
Application No. 08/864,460
Holmes teaches generating a continuous distribution
probability density HMM from a quantized vector
series for training and recognition: “A more widely
used method for coping with the fact that particular
sets of finely quantized feature values will occur
only very rarely is to represent the distribution of
feature vectors by some simple parametric model, and
to use the calculated probabilities from this model
to supply the probability distributions in the
training and recognition processes. The Baum-Welch
re-estimation must then be used to optimize the
parameters of the feature distribution model, rather
than the probabilities of the particular feature
vectors" (p. 143). Said computation of optimum
parameters of the feature distribution model (for
each state, tacitly understood) is just the recited
calculation of the incidence of the labels in each
state, from the HMM state likelihood functions
described by said parameters (claim 3), determined
from the training vectors.
Holmes also teaches clustering and using
nearest-neighbor templates representing the average
properties in each cluster (p. 125), and vector
quantizing training (and test) patterns into a label
series of clusters to which they belong ("It is
possible to make a useful approximation to the
feature vectors that actually occur by choosing only
a small subset (perhaps about 100) of feature
vectors, and replacing each measured vector by the
one in the subset that is `nearest' to it according
to a suitable distance metric. This process is
known as vector quantization", p. 142, emphasis in
original). As discussed above, since the
Specification does not teach a two-step
quantization, the examiner has interpreted the
recited "vectors so quantized" as a reference to the
inherent quantization involved in the measurement of
continuous data.
Appeal No. 2000-0189 Page 5
Application No. 08/864,460
(Paper No. 16 at 5-6.) The appellant argue, "[i]n claim 1,
the quantizing vectors are converted into a label series of
clusters, for example. Where is this limitation taught in the
two quoted sentences? Claim 1 also specifies that the
incidence of the labels in each state are calculated from the
quantizing vectors of a training pattern classified in the
same label series of clusters, for example. Where is this
limitation taught in the two quoted sentences?" (Appeal Br.
at 5-6.)
In deciding obviousness, “[a]nalysis begins with a key
legal question -- what is the invention claimed?” Panduit
Corp. v. Dennison Mfg. Co., 810 F.2d 1561, 1567, 1 USPQ2d
1593, 1597 (Fed. Cir. 1987). Here, independent claim 1
specifies in pertinent part the following limitations: "vector
quantizing means for generating a model by quantizing vectors
of a training pattern having a vector series, and converting
said quantizing vectors into a label series of clusters to
which they belong” and “label incidence calculating means for
calculating the incidence of the labels in each state from
Appeal No. 2000-0189 Page 6
Application No. 08/864,460
said quantizing vectors of a training pattern classified in
the same label series of clusters and the continuous
probability distribution density HMM." Similarly, claim 9,
the other independent claim, specifies in pertinent part the
following limitations: "vector quantizing means connected to
said word pattern memory means for quantizing vectors of a
training pattern received from said word pattern memory means
and converting said quantizing vectors into a label series of
clusters to which they belong . . . and label incidence
calculating means connected to said parameter memory means for
calculating the incidence of the labels in each state from
said quantizing vectors of a training pattern classified in
the same label series of clusters and the continuous
probability distribution density HMM.”
Having determined what subject matter is being claimed,
the next inquiry is whether the subject matter is obvious.
"’A prima facie case of obviousness is established when the
teachings from the prior art itself would appear to have
suggested the claimed subject matter to a person of ordinary
skill in the art.’" In re Bell, 991 F.2d 781, 782, 26 USPQ2d
Appeal No. 2000-0189 Page 7
Application No. 08/864,460
1529, 1531 (Fed. Cir. 1993) (quoting In re Rinehart, 531 F.2d
1048, 1051, 189 USPQ 143, 147 (CCPA 1976)).
Here, the examiner fails to show that Holmes or Thomson
teaches or would have suggested the claimed limitations.
Rather than comparing the language of the claims with the
references, he merely describes the references. We will not
“resort to speculation,” In re Warner, 379 F.2d 1011, 1017,
154 USPQ 173, 178 (CCPA 1967), as to where the claimed
limitations might be found in Holmes or Thomson. Therefore,
we reverse the rejection of claim 1-13 as obvious over Holmes
in view of Thomson.
CONCLUSION
In summary, the rejection of claims 1-13 under § 103(a)
is reversed.
REVERSED
Appeal No. 2000-0189 Page 8
Application No. 08/864,460
ERROL A. KRASS )
Administrative Patent Judge )
)
)
)
) BOARD OF PATENT
LEE E. BARRETT ) APPEALS
Administrative Patent Judge ) AND
) INTERFERENCES
)
)
)
LANCE LEONARD BARRY )
Administrative Patent Judge )
Appeal No. 2000-0189 Page 9
Application No. 08/864,460
FELIX J. D'AMBROSIO
P. O. BOX 2266
EADS STATION
ARLINGTON, VA 22202