Ex Parte Vafin et al

Patent Trial and Appeal Board

Feb 27, 2017

12838118 (P.T.A.B. Feb. 27, 2017)

United States Patent and Trademark Office
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O.Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
12/838,118 07/16/2010 Renat Vafin 335666.02 5600
69316 7590 03/01/2017
MICROSOFT CORPORATION
ONE MICROSOFT WAY
REDMOND, WA 98052
EXAMINER
HESS, MICHAEL J
ART UNIT PAPER NUMBER
2481
NOTIFICATION DATE DELIVERY MODE
03/01/2017 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
u sdocket @ micro soft .com
chriochs @microsoft.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte RENAT VAFIN, LAZAR BI VOL ARSKY, MATTIAS NILSSON,
and SOREN VANG ANDERSEN
Appeal 2016-004907
Application 12/838,118
Technology Center 2400
Before JOHN A. JEFFERY, CARL L. SILVERMAN, and
MELISSA A. HAAPALA, Administrative Patent Judges.
JEFFERY, Administrative Patent Judge.
DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from the Examiner’s
decision to reject claims 1—10, 12—16, 20, and 22—25. We have jurisdiction
under 35 U.S.C. § 6(b). We reverse.
STATEMENT OF THE CASE
Appellants’ invention is a variable length encoder for encoding an
input signal with different symbols. Each portion of the input signal is
encoded using a probability mass function (PMF) generated dynamically at
the encoder depending on the frequencies at which different symbols are
found within each portion, where the frequencies are detected dynamically
Appeal 2016-004907
Application 12/838,118
by a counter before the symbols are encoded. See generally Abstract. Claim
1 is illustrative:
1. An encoder stage for encoding an input signal
comprising a plurality of different symbols, the encoder stage
comprising:
a variable length encoder for encoding the input signal;
and
a counter configured to dynamically detect a frequency at
which the different symbols are found within each of a plurality
of predetermined portions of the input signal, prior to the
symbols of each respective portion being encoded by the variable
length encoder;
wherein the variable length encoder is configured to
encode the symbols of each portion using variable length coding
performed using a probability mass function (PMF) generated
dynamically at the encoder in dependence on the frequencies
detected dynamically by the counter within the respective portion
of the input signal including the frequencies detected by the
counter for symbols of the respective portion that have not yet
been encoded, to generate an encoded bitstream comprising the
encoded symbols along with an additional element indicating
information regarding the frequencies detected by the counter for
each portion, and to output the encoded bitstream to at least one
of a storage medium and a transmission medium for supply to a
decoder.
THE REJECTIONS
The Examiner rejected claims 1—3, 7—10, 12—16, and 22—25 under 35
U.S.C. § 102(b) as anticipated by Karczewicz (US 2004/0021592 Al; Feb.
5,2004). Ans. 2—9.1
1 Throughout this opinion, we refer to (1) the Appeal Brief filed August 31,
2015 (“App. Br.”); (2) the Examiner’s Answer mailed February 11, 2016
(“Ans.”); and (3) the Reply Brief filed April 11, 2016 (“Reply Br.”).
2
Appeal 2016-004907
Application 12/838,118
The Examiner rejected claims 4—6 and 20 under 35 U.S.C. § 103(a) as
unpatentable over Karczewicz and Smirnov (US 2007/0116370 Al; May 24,
2007). Ans. 9-11.
THE ANTICIPATION REJECTION
The Examiner finds that Karczewicz discloses every recited element
of claim 1 including a variable length encoder configured to encode each
portion of an input signal using a PMF generated dynamically at the encoder
depending on the frequencies detected dynamically by a counter. Ans. 2-4.
Although the Examiner acknowledges that Karczewicz uses training
material to determine symbol probabilities in a “one-pass” encoding scheme
in paragraph 22, the Examiner nonetheless finds that “source material” is
used to determine actual probabilities for each image to be decoded in an
adaptive “two-pass” encoding scheme in paragraphs 56 and 57. Ans. 3, 12—
14. This latter “two-pass” technique in Karczewicz is said to anticipate the
recited variable length coding scheme that depends on pre-encoded symbol
frequencies determined dynamically by the recited counter. See id.
Appellants argue that the encoding schemes of Karczewicz’s
paragraphs 56 and 57—like those in paragraph 20—assign codewords to
symbols based on a probability of their occurrence in a signal. App. Br. 12—
14; Reply Br. 5—6. According to Appellants, this estimate of the probability
that a symbol is to occur in a signal is based on the symbol’s frequency in
training material that is determined in advance, and is not equivalent to a
frequency that the symbol actually occurs in a signal as claimed.
3
Appeal 2016-004907
Application 12/838,118
ISSUE
Under § 102, has the Examiner erred in rejecting claim 1 by finding
that Karczewicz discloses the variable length encoder configured to encode
each portion of an input signal using a PMF generated dynamically at the
encoder depending on the frequencies at which different symbols are found
within each portion, where the frequencies are detected dynamically by a
counter before the symbols are encoded?
ANALYSIS
We begin by noting that it is undisputed that the encoding scheme in
Karczewicz’s paragraph 22 determines symbol probabilities based on
training material and, therefore, does not anticipate a variable length encoder
that uses a PMF generated dynamically depending on the particular detected
pre-encoded symbol frequencies recited in claim 1. See Ans. 12—13
(acknowledging that Karczewicz’s “one-pass” encoding scheme in
paragraph 22 determines symbol probabilities based on training material).
The question, then, is whether the Examiner’s additional reliance on
Karczewicz’s encoding scheme in paragraphs 56 and 57, that the Examiner
characterizes as a “two-pass” encoding scheme (Ans. 3, 12—13), anticipates
the disputed limitation. On this record, we find that it does not.
Karczewicz’s adaptive variable length coding (VLC) scheme
calculates initial VLC codes and mappings in both the encoder and decoder
based on symbol probability estimates determined in advance using a
database of training images representative of those to be encoded and
transmitted. Karczewicz | 57. As Appellants explain, the probability that a
signal is to occur in a signal as determined by Karczewicz’s estimate is not
4
Appeal 2016-004907
Application 12/838,118
equivalent to a frequency that the symbol actually occurs in a signal. Reply
Br. 6.
We reach the same conclusion regarding the subsequent updates to the
symbol probability estimates in Karczewicz’s paragraph 57. Notably, these
updates occur as further encoded data symbols are transmitted, and the
encoder and decoder use these updated estimates to recalculate VLC
codewords and assignments—a recalculation that can occur after receiving
each new symbol. Karczewicz | 57. But these updates are not necessarily
based on a counter that dynamically detects a frequency at which the
different symbols are found within each portion of the input signal before the
symbols are encoded—a key aspect of the recited coding scheme. As
Appellants’ Specification explains, the disclosed counter counts the actual
number of occurrences of different possible symbol values within each
frame to determine a PMF used for encoding. Spec. 11:14— 18. Not only do
Karczewicz’s updates occur after encoding the transmitted data symbols, the
updates merely involve estimates of probability: they do not necessarily
reflect, nor are they necessarily based on, a detected frequency at which the
symbols are found within each portion of an input signal before the symbols
are encoded which, as noted above, reflects the actual number of
occurrences in each portion that are so detected.
We reach this conclusion even assuming, without deciding, that
Karczewicz equates the term “frequency of occurrence” with “probability”
in light of the parenthetical expression in paragraph 55 as the Examiner
indicates. Ans. 14. That is, even if Karczewicz’s probability estimates in
paragraph 57 can somehow be equated to frequency estimates in light of
Karczewicz’s nomenclature in paragraph 55, they are still estimates based
5
Appeal 2016-004907
Application 12/838,118
primarily on training material as noted previously. To the extent that the
Examiner finds that Karczewicz’s probability estimate updates are
necessarily based on detected frequency at which the symbols are found
within each portion of an input signal before the symbols are encoded, the
Examiner fails to substantiate that finding.
That Karczewicz’s encoding scheme in paragraph 57 is adaptive as
the Examiner indicates (Ans. 13) is of little consequence here, for even
adaptive schemes can suffer from the very deficiencies from pre-trained
PMFs that the claimed invention was designed to alleviate. See, e.g., Spec.
3:21—4:19 (noting the drawbacks of the adaptive encoding scheme in Figure
1); see also id. 5:13—15 (“Even in the adaptive case, [retraining the PMF]
only involves selecting between a few pre-stored, pre-trained VLC look-up
tables based on certain previously encoded symbols, and/or by updating a
pre-trained PMF based on previously encoded symbols.") (emphasis added).
Therefore, we are persuaded that the Examiner erred in rejecting (1)
independent claim 1; (2) independent claim 22 which recites commensurate
limitations; and (3) dependent claims 2, 3, 7—10, 12—16, and 22—25 for
similar reasons.
THE OBVIOUSNESS REJECTION
Because the Examiner has not shown that Smirnov cures the
foregoing deficiencies regarding the Examiner’s reliance on Karczewicz in
rejecting independent claims 1 and 22, and similar deficiencies regarding
independent claim 20 (see Ans. 11, 17—18), we will not sustain the
obviousness rejection of claims 4—6 and 20 (Ans. 9—11) for similar reasons.
Because this issue is dispositive regarding our reversing the Examiner’s
6
Appeal 2016-004907
Application 12/838,118
rejection of these claims, we need not address Appellants’ other associated
arguments.
To the extent that the Examiner concludes, for the first time on appeal,
that Smirnov cures Karczewicz’s above-noted deficiencies to render
independent claims 1 and 22 obvious, and invites us to so reject the claims in
the first instance on appeal (see Ans. 15), we decline to do so. Rather, we
leave that issue to the Examiner to resolve after this decision.
CONCLUSION
The Examiner erred in rejecting claims 1—3, 7—10, 12—16, and 22—25
under § 102, and claims 4—6 and 20 under § 103.
DECISION
The Examiner’s decision rejecting claims 1—10, 12—16, 20, and 22—25
is reversed.
REVERSED
7