Ex Parte Könönen et al

Patent Trial and Appeal Board

Jan 23, 2019

13320265 (P.T.A.B. Jan. 23, 2019)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE
13/320,265 11/13/2011
44696 7590
Dr. Mark M. Friedman
Moshe Aviv Tower, 54th floor
7 Jabotinsky St.
Ramat Gan, 5252007
ISRAEL
01/25/2019
FIRST NAMED INVENTOR
Ville Kononen
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
ATTORNEY DOCKET NO. CONFIRMATION NO.
5070/4 5942
EXAMINER
TRAN,MAIT
ART UNIT PAPER NUMBER
2124
NOTIFICATION DATE DELIVERY MODE
01/25/2019 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):
patents@friedpat.com
friedpat.uspto@gmail.com
rivka_f@friedpat.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte VILLE KONONEN, JUSSI LIIKKA, and JANI MANTY JARVI
Appeal2018-000247
Application 13/320,265
Technology Center 2100
Before JOHNNY A. KUMAR, CHARLES J. BOUDREAU, and
JASON M. REPKO, Administrative Patent Judges.
REPKO, Administrative Patent Judge.
DECISION ON APPEAL
STATEMENT OF THE CASE
Appellants 1 appeal under 35 U.S.C. § 134(a) from the Examiner's
rejection of claims 21-23 and 25--40. App. Br. 5. 2 Claims 1-20 and 24 are
canceled. We have jurisdiction under 35 U.S.C. § 6(b ).
We affirm.
1 According to Appellants, the real party in interest is Teknologian
Tutkimuskeskus VTT of Finland. App. Br. 3.
2 Throughout this opinion, we refer to the Final Rejection ("Final Act.")
mailed Sept. 8, 2018; the Appeal Brief ("App. Br.") filed March 20, 2017;
the Examiner's Answer ("Ans.") mailed August 9, 2017; and the Reply Brief
("Reply Br.") filed October 9, 2017.
Appeal2018-000247
Application 13/320,265
THE INVENTION
Appellants' invention relates to context recognition in mobile devices.
Spec. 1:6-7. A typical context-recognition system uses a classification
algorithm that maps sensor observations to a context. Id. at 1 :35-36. A
context could be an environmental state, the device's state, or a user's
physiological state. Id. at 1: 18-19. The Specification explains that
classifiers used in this way have several limitations, including variable
accuracy and high computational requirements for training and recognition,
among other issues. Id. at 1 :35-2:5.
To improve classification for context recognition, the invention uses
feedback from the user. Id. at 3:30-36. Feedback can include some
indication of the classification's quality and correctness. Id. at 4:6-7. For
instance, the user can provide feedback using keys shown on a mobile
device's touchscreen. Id. at 4:9-11. The invention adapts the classifier to
take into account the user's feedback. Id. at 4:14--15.
Claims 21, 38, and 39 are independent. Claim 21 is reproduced below
with our emphasis:
21. A mobile device comprising:
a feature determination logic for determining a plurality
of representative feature values on the basis of sensing data
indicative of the context of the mobile device associated with
the user thereof, and a context recognition logic including an
adaptive linear classifier, configured to map, during a
classification action, the plurality of feature values to a context
class, wherein the classifier is further configured to adapt the
classification logic thereof on the basis of the feature values
and feedback information by the user of the mobile device,
wherein in the case of positive or negative feedback
regarding the performed classification, the classifier is
configured to adapt the classification logic thereof such that a
2
Appeal2018-000247
Application 13/320,265
prototype feature value vector of the recognized class is brought
closer to or farther away from the feature vector, the feature
vector determined by the plurality of feature values,
respectively.
THE EVIDENCE
The Examiner relies on the following as evidence:
Golding et al.
Chaddha
Botchen et al.
Huang et al.
US 6,323,807 Bl
US 6,404,923 B 1
US 2009/0278937 Al
US 2010/0087987 Al
Nov. 27, 2001
June 11, 2002
Nov. 12, 2009
Apr. 8, 2010
Dwi H. Widyantoro et al., An Adaptive Algorithm for Learning
Changes in User Interests, CIKM '99, 405--412 (1999) ("Widyantoro").
Kristof Van Laerhoven, Combining the Self-Organizing Map and K-
Means Clustering for On-Line Classification of Sensor Data, ICANN 2001,
464--469 (2001) ("Van Laerhoven").
Jon Froehlich et al., MyExperience: A System for In situ Tracing and
Capturing of User Feedback on Mobile Phones, Mobisys '07 (2007)
("Froehlich").
Erich Bruns et al., Mobile Phone-Enabled Museum Guidance with
Adaptive Classification, Projects in VR, 98-102 (2008) ("Bruns").
THE REJECTIONS
Claims 21-23, 31, and 38--40 stand rejected under 35 U.S.C. § I03(a)
as unpatentable over Van Laerhoven and Widyantoro. Final Act. 3-7.
Claims 25 and 30 stand rejected under 35 U.S.C. § I03(a) as
unpatentable over Van Laerhoven, Widyantoro, and Chaddha. Final Act. 7-
9.
Claim 26 stands rejected under 35 U.S.C. § I03(a) as unpatentable
over Van Laerhoven, Widyantoro, and Botchen. Final Act. 9-10.
3
Appeal2018-000247
Application 13/320,265
Claims 27-29 stand rejected under 35 U.S.C. § 103(a) as unpatentable
over Van Laerhoven, Widyantoro, and Bruns. Final Act. 11-13.
Claims 32 and 33 stand rejected under 35 U.S.C. § 103(a) as
unpatentable over Van Laerhoven, Widyantoro, and Golding. Final Act. 13-
15.
Claims 34--36 stand rejected under 35 U.S.C. § 103(a) as unpatentable
over Van Laerhoven, Widyantoro, and Froehlich. Final Act. 16-18.
Claim 37 stands rejected under 35 U.S.C. § 103(a) as unpatentable
over Van Laerhoven, Widyantoro, and Huang. Final Act. 18-20.
THE OBVIOUSNESS REJECTION OVER VAN LAERHOVEN AND
WIDYANTORO
The Examiner finds that Van Laerhoven teaches every limitation
recited in representative3 claim 21 except for using the recited user feedback.
Final Act. 4--5. In concluding that the subject matter of claim 21 would have
been obvious, the Examiner finds Widyantoro teaches this feature. Id. at 5.
Specifically, the Examiner finds that Widyantoro uses the feedback about a
user's interests to adapt logic for updating feature-vector descriptors.
Id. at 4--5 (citing Widyantoro 405, 407--408).
Appellants argue that Widyantoro' s classification logic does not adapt
as required by claim 21. App. Br. 24--25. Instead, according to Appellants,
3 Appellants argue claims 21-23, 31, and 38--40 as a group. See App. Br.
20-25. In particular, Appellants argue claims 22, 23, 31, and 38--40 would
not have been obvious over Van Laerhoven and Widyantoro for the same or
similar reasons as those presented for claim 21. Id. at 25. Accordingly, we
select independent claim 21 as representative of claims 22, 23, 31, and 38-
40. See 37 C.F.R. § 4I.37(c)(l)(iv).
4
Appeal2018-000247
Application 13/320,265
Widyantoro uses the same logic for each article-sorting cycle. Id. We
disagree.
Claim 21 recites, in relevant part, "the classifier is further configured
to adapt the classification logic thereof on the basis of the feature values and
feedback information by the user of the mobile device" ( emphasis added).
The Examiner interprets the term "adapt" as adjusting the logic to new
conditions. Ans. 25. Appellants do not rebut the Examiner's interpretation.
See Reply Br. 1--4.
We find the Examiner's interpretation reasonable in view of the
Specification, which discloses adjusting logic to new conditions. See, e.g.,
Spec. 19:6-17. For example, the Specification discusses adjusting an
adaptive linear classifier to the condition of the user providing feedback
about the correct context class. Id.
We agree with the Examiner that, like Appellants' logic,
Widyantoro's logic is also adjusted to new user feedback. Ans. 25. In
particular, Widyantoro relates to automated information filtering.
Widyantoro 405, Abstract. Potential applications include personalizing a
news agent to gather online articles and filter them according to a user's
interests. Id., § 1. Widyantoro' s adaptive algorithm learns changes to a
user's interests. Id., Abstract. Like the recited method, Widyantoro's
learning process also relies on the user's feedback. Id. at 407, § 3.3.1. This
feedback represents whether the user likes or dislikes a document's content.
Id. With this feedback, Widyantoro's learning process updates the
descriptor feature vectors. Id. This updating uses equation 9 in section
3.3.2:
d(new) = d(old) * (1 - a) + fvd * a
5
Appeal2018-000247
Application 13/320,265
We agree with the Examiner that Widyantoro adapts this logic in the
case of the user's positive or negative feedback, as claim 21 recites. See
Final Act. 4 (citing Widyantoro 408, equation 9); Ans. 25. Specifically, in
equation 9 above, dcis d~ for positive feedback, d~ for negative feedback,
and dft for both positive and negative feedback. Widyantoro 408, § 3.3.3.2.
So contrary to Appellants' argument (see App. Br. 24--25), Widyantoro uses
different logic-i.e., an equation with a different coefficient-after receiving
different feedback. Ans. 25.
The Examiner concludes that it would have been obvious to
incorporate Widyantoro' s adapting in Van Laerhoven to arrive at the recited
classifier. Final Act. 4--5. For the reasons discussed below, Appellants have
not persuaded us that the Examiner erred in this regard.
Appellants argue that the Examiner's proposed combination lacks any
rational underpinning, and thus, is pure hindsight. App. Br. 21-22;
Reply Br. 1-3. In Appellants' view, impermissible hindsight is "the only
way" to modify Van Laerhoven with Widyantoro. App. Br. 21. Appellants
argue the Examiner's proposed combination would render Van Laerhoven
unsuitable for its intended purpose. Id. at 24; Reply Br. 2-3. According to
Appellants, introducing Widyantoro' s feedback into the Self Organizing
Map ("SOM") and Kohonen Self Organizing Map ("KSOM") neural
networks disclosed by Van Laerhoven would "break the Van Laerhoven
system, such that it no longer conforms to the SOM or KSOM neural
networks." App. Br. 24; see also Reply Br. 4. Appellants argue that the
SOM algorithm's stabilization would be lost because external feedback
would lead to violations of Van Laerhoven's inbound winner-takes-all
principle. Reply Br. 2. Appellants argue that, because the winner-takes-all
6
Appeal2018-000247
Application 13/320,265
principle would be violated, Van Laerhoven would have to be completely
redesigned. Id.
Appellants, however, have not shown that redesigning Van Laerhoven
in this way would have been uniquely challenging or otherwise beyond the
level of ordinarily skilled artisans. See Leapfrog Enters., Inc. v. Fisher-
Price, Inc., 485 F.3d 1157, 1162 (Fed. Cir. 2007). Even if Van Laerhoven
would need to be changed or redesigned, it is well settled that "a
determination of obviousness based on teachings from multiple references
does not require an actual, physical substitution of elements." In re Mouttet,
686 F.3d 1322, 1332 (Fed. Cir. 2012) (citations omitted). Here, Appellants
have at most shown that Van Laerhoven would operate in a different way in
the Examiner's proposed combination.
Appellants provide no persuasive evidence to substantiate the claim
that the modification would render Van Laerhoven unsuitable for its
intended purpose. Instead, Appellants only offer conclusory assertions that
have little probative value. See In re Geisler, 116 F.3d 1465, 1470 (Fed. Cir.
1997), cited in Ans. 23.
Specifically, Appellants assert the proposed combination "would
break the SOM and KSOM neural networks." App. Br. 24. The Examiner,
however, finds that the Van Laerhoven-Widyantoro combination would
classify in a different way than Van Laerhoven alone-i.e., the combination
would use the user's feedback. See Ans. 24--25. In its asserted rebuttal of
this finding, Appellants have not provided persuasive evidence that Van
Laerhoven would be unsuitable for classification. Rather, Appellants have
only shown that Van Laerhoven would no longer use a particular feature for
classification-i.e., the winner-takes-all aspect. See App. Br. 25;
7
Appeal2018-000247
Application 13/320,265
Reply Br. 2. In this regard, Appellants take an overly narrow view of Van
Laerhoven's intended purpose. Here, Van Laerhoven's intended purpose is
classifying sensor data. See Van Laerhoven, Abstract. And considering the
weight of the evidence before us, we are unconvinced that the Examiner's
modification to Van Laerhoven would render it unsuitable for classifying
sensor data.
Appellants also argue that Van Laerhoven and Widyantoro are
non-analogous art. App. Br. 22-23. In Appellants' view, Van Laerhoven
teaches a neural network and Widyantoro has "nothing to do with neural
networks." Id. at 22; Reply Br. 2. Appellants argue that "a skilled worker
would not be motivated to combine Van Laerhoven with Widyantoro,
because the technologies are so different." App. Br. 23.
Prior art is analogous if it is (1) from the same field of endeavor as
that of Appellants' invention regardless of the problem addressed or
(2) reasonably pertinent to the particular problem with which the inventor is
involved. In re Bigio, 381 F.3d 1320, 1325 (Fed. Cir. 2004) (emphasis
added). That is, the analogous-art test does not ask whether the prior art
references are analogous to each other, as Appellants' argument suggests.
See App. Br. 22-23; Reply Br. 2. Rather, the proper inquiry is whether the
references are analogous to Appellants' invention or the problem with which
the inventor is concerned. See Bigio, 381 F.3d at 1325. For at least the
reason that Appellants' arguments do not squarely address the relevant
considerations, these analogous-art arguments are unpersuasive.
Furthermore, Widyantoro's user-feedback processing is at least reasonably
pertinent to the inventor's problem because Widyantoro teaches a way to use
feedback to reduce undesired outcomes. See Widyantoro 405, 407--408.
8
Appeal2018-000247
Application 13/320,265
To the extent Appellants argue that the differences between Van
Laerhoven and Widyantoro weigh against the Examiner's proposed
combination, we find that reasoning unpersuasive for the reasons discussed
above. We also note that the Examiner finds that Van Laerhoven and
Widyantoro teach adaptive algorithms. Ans. 25. The proposed combination
merely substitutes one basis for adapting with another. See Final Act. 4--5;
Ans. 25.
Accordingly, we sustain the rejection of claim 21, which is
representative of claims 22, 23, 31, and 38--40. See supra note 3.
THE REMAINING OBVIOUSNESS REJECTIONS
Claims 25-30 and 32-37 depend from claims 21, 38, and 39. In
arguing for the patentability of claims 25-30 and 32-37, Appellants refer to
the arguments presented for the independent claims. See App. Br. 26-31;
Reply Br. 4. Thus, for the reasons discussed in connection with claims 21,
38, and 39, we also sustain the rejections of claims 25-30 and 32-37.
DECISION
We affirm the Examiner's decision to reject claims 21-23 and 25--40.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l )(iv). See
37 C.F.R. § 4I.50(f).
AFFIRMED
9