`
`UNITED STATES DISTRICT COURT
`SOUTHERN DISTRICT OF NEW YORK
`
`
`
`
`NETWORK-1 TECHNOLOGIES, INC.,
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`v.
`
`
`
`Plaintiff,
`
`GOOGLE LLC and YOUTUBE, LLC,
`
`
`
`
`
`
`Defendants.
`
`
`
`
`
`
`14 Civ. 2396 (PGG)
`
`14 Civ. 9558 (PGG)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`PLAINTIFF NETWORK-1 TECHNOLOGIES, INC.’S
`OPENING CLAIM CONSTRUCTION BRIEF
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 2 of 28
`
`
`TABLE OF CONTENTS
`
`
`
`I.
`II.
`III.
`
`
`
`INTRODUCTION .............................................................................................................. 1
`BACKGROUND ON THE PATENTS-IN-SUIT .............................................................. 2
`CLAIM CONSTRUCTION PRINCIPLES ........................................................................ 6
`A. Claim Interpretation Focuses on the Meaning of Terms to Persons of Ordinary Skill in the
`Art ....................................................................................................................................... 6
`B. Claims Are Definite Unless They Fail to Inform Those of Skill in the Art of the Scope of
`the Invention with Reasonable Certainty ............................................................................ 8
`IV.
`AGREED CONSTRUCTIONS .......................................................................................... 9
`DISPUTED CONSTRUCTIONS ..................................................................................... 10
`V.
`A. “non-exhaustive [. . .] search” ........................................................................................... 10
`1. The Intrinsic Evidence Confirms Network-1’s Construction of “Non-Exhaustive Search”
`....................................................................................................................................... 11
`2. Network-1’s Construction Also Conforms to the Understanding of Those Skilled in the Art
`and the Extrinsic Evidence ................................................................................................ 13
`3. Network-1’s Proposed Construction Under Phillips Is Consistent with, Albeit Narrower
`Standard ............................................................................................................................ 15
`4. Defendants’ Assertion of Indefiniteness Cannot Be Supported ....................................... 17
`B. “correlation information” .................................................................................................. 18
`1. The Term “Correlation Information” Does Not Require Construction ............................ 19
`2. Alternatively, the Term “Correlation Information” Should Be Defined Using the Claim
`Language ........................................................................................................................... 20
`3. “Correlation Information” Is Not Indefinite ..................................................................... 21
`C. “extracted features” and “extracting features” .................................................................. 21
`CONCLUSION ................................................................................................................. 24
`VI.
`
`than, the Federal Circuit’s Prior Construction Under the Broadest Reasonable Construction
`
`
`
`i
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 3 of 28
`
`
`TABLE OF AUTHORITIES
`
`CASES
`
`Dow Chem. Co. v. Nova Chems. Corp. (Canada),
`809 F.3d 1223 (Fed. Cir. 2015) .................................................................................................. 9
`
`
`Eidos Display, LLC v. AU Optronics Corp.,
`779 F.3d 1360 (Fed. Cir. 2015)................................................................................................. 10
`
`
`Eli Lilly & Co. v. Teva Parenteral Meds., Inc.,
`845 F.3d 1357 (Fed. Cir. 2017)................................................................................................... 8
`
`
`Google Inc. v. Network-1 Techs., Inc.,
`No. CBM2015-00113, Paper No. 2 (Petition) (PTAB Apr. 13, 2015) ..................................... 18
`
`
`Google Inc. v. Network-1 Techs., Inc.,
`No. IPR2015-00345, Paper No. 30 (Final Written Decision) (PTAB June 20, 2016) .............. 15
`
`
`Google Inc. v. Network-1 Techs., Inc.,
`No. IPR2015-00347, Paper No. 30 (Final Written Decision) (PTAB June 20, 2016) .............. 15
`
`
`Google LLC v. Network-1 Techs., Inc.,
` 726 F. App’x 779, 786 (Fed. Cir. 2018) ............................................................................. 16, 17
`
`MeadWestVaco Corp. v. Rexam Beauty & Closures, Inc.,
`731 F.3d 1258 (Fed. Cir. 2013)................................................................................................... 8
`
`
`Nautilus, Inc. v. Biosig Instruments, Inc.,
`572 U.S. 898 (2014) .............................................................................................................. 8, 11
`
`
`O2 Micro Int’l Ltd. v. Beyond Innovation Tech. Co.,
`521 F.3d 1351 (Fed. Cir. 2008)........................................................................................... 19, 22
`
`
`Phillips v. AWH Corp.,
`415 F.3d 1303 (Fed. Cir. 2005) (en banc) ....................................................................... 7, 15, 16
`
`
`Rembrandt Data Techs., LP v. AOL, LLC,
`641 F.3d 1331 (Fed. Cir. 2011)................................................................................................... 9
`
`
`Sonix Tech. Co. v. Publ’ns Int’l, Ltd.,
`844 F.3d 1370 (Fed. Cir. 2017)................................................................................................. 17
`
`
`Sys. Mgmt. Arts Inc. v. Avesta Techs., Inc.,
`137 F. Supp. 2d 282 (S.D.N.Y. 2001) ........................................................................................ 9
`
`
`
`
`
`ii
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 4 of 28
`
`
`TABLE OF AUTHORITIES (continued)
`
`
`
`Telemac Cellular Corp. v. Topp Telecom, Inc.,
`247 F.3d 1316 (Fed. Cir. 2001)................................................................................................... 7
`
`
`Teva Pharms. USA, Inc. v. Sandoz, Inc., ___U.S. ___,
`135 S. Ct. 831 (2015) .............................................................................................................. 6, 8
`
`
`U.S. Surgical Corp. v. Ethicon, Inc.,
`103 F.3d 1554 (Fed. Cir. 1997)................................................................................................. 19
`VaporStream, Inc. v. Snap Inc., No. 17-cv-220,
`2018 U.S. Dist. LEXIS 34606 (C.D. Cal. Feb. 27, 2018) ......................................................... 21
`
`
`Wellman, Inc. v. Eastman Chem. Co.,
`642 F.3d 1355 (Fed. Cir. 2011)............................................................................................. 8, 10
`
`
`STATUTE
`
`35 U.S.C. § 112 ........................................................................................................................... 1, 8
`
`RULE
`
`37 C.F.R. § 42.100(b) ................................................................................................................... 15
`
`
`
`
`
`
`iii
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 5 of 28
`
`
`I.
`
`INTRODUCTION
`
`The parties ask the Court to construe four terms from the claims of the patents-in-suit: (1)
`
`“non-exhaustive search”; (2) “correlation information”; (3) “extracted features”; and (4)
`
`“extracting features.” Claim construction requires that the Court determine the ordinary and
`
`customary meaning of these terms to a person of ordinary skill in the relevant art at the time of the
`
`effective filing date of the patents-in-suit. This understanding is informed by intrinsic evidence
`
`such as the context of the particular claims in which a given term appears and the patent
`
`specification, as well as extrinsic evidence such as contemporaneous publications and expert
`testimony.
`
`Defendants Google LLC and YouTube, LLC assert that the terms “non-exhaustive search”
`
`and “correlation information” are indefinite. Indefiniteness is a species of patent invalidity under
`
`35 U.S.C. § 112, paragraph 2. Defendants bear a heavy burden in proving indefiniteness—they
`
`must prove by clear and convincing evidence that the claims fail to inform persons skilled in the art
`
`of the scope of the invention with reasonable certainty. Defendants cannot meet that burden as to
`
`either term they challenge here. Although Google previously proposed a construction for
`
`“non-exhaustive search” in a related proceeding, Defendants do not provide any proposed claim
`
`construction for either “non-exhaustive search” or “correlation information” here. On the other
`
`hand, Plaintiff Network-1 Technologies, Inc. proposes constructions for these two terms that are
`
`supported by the intrinsic and extrinsic evidence, and in the case of “non-exhaustive search,” a
`
`construction that is consistent with the prior constructions from related proceedings, including that
`previously proposed by Google.
`
`For “extracted features” and “extracting features,” Network-1 offers constructions that
`
`provide context to the fact-finder on what “extracted features” are and what types of actions
`
`
`
`1
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 6 of 28
`
`
`“extracting features” entails, with that context rooted in the patent specification. Defendants’
`
`construction lacks this context and should be rejected in favor of Network-1’s clearer construction.
`
`Without the clarification Network-1 seeks, it is likely the Court will need to address the scope of
`these terms again later in the case.
`
`II.
`
`BACKGROUND ON THE PATENTS-IN-SUIT
`
`Professor Ingemar J. Cox, a Professor in the Department of Computer Science at the
`
`University College London and at the University of Copenhagen, is the inventor of the three
`
`patents-in-suit. Ex. 1,1 U.S. Patent No. 8,010,988 (“the ’988 patent”); Ex. 2, 8,250,237 (“the ’237
`
`patent”); Ex. 3, U.S. Patent No. 8,904,464 (“the ’464 patent”). Dr. Cox is also the author of nearly
`
`300 academic papers in the fields of information retrieval, digital watermarking, and video search
`
`identification techniques. He has served as the Head of the Future Media Group and Director of
`
`Research in the Computer Science Department at University College London, a visiting professor
`
`at the Danish Technical University, a distinguished visiting scholar at Pennsylvania State
`
`University, a fellow of Computer Science at the NEC Research Institute, and a principal
`
`investigator at the Robotics Principles Department of AT&T Bell Laboratories. Dr. Cox holds a
`
`Ph.D. from Oxford University in engineering science and a Bachelor’s of Science in electronics
`
`and computer science from University College London.
`
`In September 2000, Dr. Cox filed a provisional patent application that ultimately led to the
`
`three patents-in-suit. The patents describe systems and methods for identifying digital media
`
`content. ’988 patent at 5:37-55 (§ 4.0).2 More specifically, the patents describe systems in which
`
`
`1 All references to “Ex.” in this brief are exhibits to the appended Declaration of Amy E. Hayden in
`Support of Plaintiff Network-1 Technologies, Inc.’s Opening Claim Construction Brief.
`2 All three patents-in-suit share the same specification, with minor modifications. For ease, unless
`otherwise noted, citations to the specification throughout this brief will be to the ’988 patent
`specification, though the same material is present in all three of the patents. The citations
`
`
`
`
`2
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 7 of 28
`
`
`a database of reference works (such as known recordings of popular songs, or known videos of
`
`television programs or movies) can be maintained. Id. at 8:4-59 (§ 4.2.1.1.2). Each of these
`
`reference works can be represented by a set of extracted features—an electronic “sketch” or
`“fingerprint.” Id. at 7:11-8:2 (§ 4.2.1.1.1); see also Mitzenmacher Decl. ¶¶ 23-25.
`
`The patents teach that in addition to these sketches or fingerprints, the database could also
`
`store information about actions to be performed that are connected or associated with each
`
`reference work. ’988 patent at 8:4-59 (§ 4.2.1.1.2); id. at 23:36-51; see also Mitzenmacher Decl.
`
`¶¶ 26-29. For example, each time the video of a particular television program is shown to a
`
`viewer, additional information, like an advertisement or related weblink, could be displayed to the
`viewer. See, e.g., ’988 patent at 8:15-20, 23:42-47; see also Mitzenmacher Decl. ¶¶ 20-21.
`
`The patents contemplate a system where unknown content needs to be compared to the
`
`reference works to see if it matches any of the reference works. See, e.g., ’988 patent at 6:31-7:10
`
`(§ 4.2.1.1); see also Mitzenmacher Decl. ¶ 30. For example, in an internet video system, the
`
`operator might wish to compare each new video uploaded by a user to the library of known videos
`
`to see if it matches any of those known reference videos. Such comparisons could help to identify
`
`duplicative videos, to limit copyright infringement by users of the system, and/or to facilitate
`better identification and description of videos on the system.
`
`The patents teach that the new work (the unknown video in this example), like each of the
`
`reference works, can be processed to create a set of extracted features—a “sketch” or a
`
`
`reference the section numbers used in the specification to facilitate ease of cross-reference, where
`applicable. There are, however, slight differences in the specifications. The “Summary of the
`Invention” section of the specifications varies somewhat, and there is some additional discussion
`in the later ’237 and ’464 patents of certain references that are incorporated by reference in all of
`the specifications. See also Declaration of Professor Michael D. Mitzenmacher in Support of
`Plaintiff Network-1 Technologies, Inc.’s Opening Claim Construction Brief (“Mitzenmacher
`Decl.”) ¶ 19.
`
`
`
`3
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 8 of 28
`
`
`“fingerprint”3—of the new unknown video. ’988 patent at 6:61-65. These extracted features can
`
`then be compared to the database of extracted features of the known references. Id. at 6:66-7:3.
`
`To make such comparisons, the patents note that comparing the fingerprint for the unknown video
`
`to the fingerprints from all of the known references could be time consuming. Id. at 8:60-9:55 (§
`
`4.2.1.1.3). This difficulty stems from two causes: first, the database of known references could be
`
`very large; and second, each comparison of two “fingerprints” can be itself computationally
`
`intensive. The comparisons can be computationally intensive for two reasons: 1) the need to find
`
`close matches that are not necessarily identical, and 2) the need to compare potentially “high
`dimensional” representations. See id.; see also Mitzenmacher Decl. ¶¶ 31-36.
`
`The patents teach that comparisons of the fingerprints of the unknown work and the
`
`reference fingerprints need to be looking not for exact matches, but for close matches, because one
`
`will often want the system to identify two pieces of content as matching even though they have
`
`small or subtle differences between them. ’988 patent at 8:60-9:55 (§ 4.2.1.1.3). An unknown
`
`work may be non-identical to a reference work only because it has noise or distortions in it, not
`
`because the underlying content is in fact entirely different. Id. at 9:1-3. For example, an incoming
`
`video that contains a portion of a television program might have been recorded from a broadcast
`
`television signal that contains some static or “noise” that creates distortions, or it might have been
`
`recorded by someone using a video camera pointed at a television on which the program was being
`
`displayed, or it might contain a clip that was only a portion of the television program. In each of
`
`these cases, since the video contains the reference television program (or a part of it), designers
`
`would want the system to identify the uploaded video as a match to the reference, even though it
`
`
`3 Although either the term “sketch” or “fingerprint” can be used to describe the set of features
`extracted from an electronic work (see Mitzenmacher Decl. ¶ 24), for simplicity, this brief refers to
`them as “fingerprints.”
`
`
`
`4
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 9 of 28
`
`
`was not perfectly identical. To determine whether the uploaded video is a match to the reference,
`
`the comparisons would need to determine not just that the two fingerprints are different, but also
`
`some measure of how different they are. Designers can decide how different two things have to be
`before they are not identified as a match. Id. at 9:39-55; see also Mitzenmacher Decl. ¶¶ 31-36.
`
`The individual comparisons also could be complicated by the “high dimensionality” of the
`
`“fingerprints” (sets of features extracted from the works). The patent makes reference to this issue
`
`of a “high dimensional feature space.” ’988 patent at 9:23-24. Fingerprints need to contain
`
`enough data to be sufficiently complex so that each fingerprint represents the underlying content
`
`(the primary work) with a low likelihood of two different references having the same fingerprint.
`
`See Mitzenmacher Decl. ¶ 29. To achieve sufficient complexity that the fingerprint actually
`
`represents an audio or video work, it is often necessary to use many data points. This might be
`
`achieved by capturing different types of information about the work at a variety of points in time.
`
`In the context of a song, for example, one could capture snapshots of or “sample” particular values
`
`(for example the pitch and intensity values) at multiple times during the song. Id. ¶ 33; see also
`
`’988 patent at 7:11-42 (describing various feature extraction methodologies). This data could then
`be used as the fingerprint for comparison to the database of fingerprints for the reference works.
`
`If the uploaded work matches one of the reference works, the system can then determine
`
`what action or actions are associated with the reference work and apply the same action or actions
`
`to the uploaded work. ’988 patent at 9:57-10:40 (§ 4.2.1.1.4). For example, if the system provided
`
`that an advertisement or a weblink was to be shown with works that match the reference work, the
`
`advertisement or weblink can also be shown with the now-identified uploaded work. The patents
`
`describe an example in which the incoming work is an advertisement for Ford Motor Company,
`
`
`
`5
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 10 of 28
`
`
`and the associated action may be to direct the viewer to the website of a local Ford dealership. Id.
`at 24:23-27.
`The specifications contain a flow diagram showing the processes detailed above:
`
`
`
`Professor Mitzenmacher discusses this flow diagram in further detail in his accompanying
`declaration. Mitzenmacher Decl. ¶¶ 22-30.
`
`III. CLAIM CONSTRUCTION PRINCIPLES
`
`A.
`
`Claim Interpretation Focuses on the Meaning of Terms to Persons of
`Ordinary Skill in the Art
`Patent claim interpretation is a question of law, often with the need for subsidiary
`
`factfinding. See Teva Pharms. USA, Inc. v. Sandoz, Inc., ___ U.S. ___, 135 S. Ct. 831, 838 (2015)
`
`(citing Markman v. Westview Instruments, Inc., 517 U.S. 370, 378, 388, 390 (1996)). The Federal
`
`
`
`6
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 11 of 28
`
`
`Circuit has explained that “[t]he inquiry into how a person of ordinary skill in the art understands a
`
`claim term provides an objective baseline from which to begin claim interpretation.” Phillips v.
`AWH Corp., 415 F.3d 1303, 1313 (Fed. Cir. 2005) (en banc) . The Phillips Court further noted that
`
`“the person of ordinary skill in the art is deemed to read the claim term not only in the context of
`
`the particular claim in which the disputed term appears, but in the context of the entire patent,
`
`including the specification.” Id. Thus, the Court’s task in claim construction is to determine how
`
`a person of skill in the art at the time of the patent (i.e., the priority date of the application—in this
`
`case September 2000) would interpret a disputed term, having considered the patent specification
`and prosecution history:
`
`It is the person of ordinary skill in the field of the invention through whose eyes the
`claims are construed. Such person is deemed to read the words used in the patent
`documents with an understanding of their meaning in the field, and to have
`knowledge of any special meaning and usage in the field. The inventor’s words
`that are used to describe the invention—the inventor’s lexicography—must be
`understood and interpreted by the court as they would be understood and
`interpreted by a person in that field of technology. Thus the court starts the
`decision-making process by reviewing the same resources as would that person,
`viz., the patent specification and the prosecution history.
`Id. (quoting Multiform Desiccants, Inc. v. Medzam, Ltd., 133 F.3d 1473, 1477 (Fed. Cir. 1998)).
`
`The patent specification (including the patent claims) and prosecution history, collectively
`
`considered the “intrinsic evidence,” are the first source for claim construction and the importance
`
`of intrinsic evidence has been emphasized by the Federal Circuit. Id. at 1316-17.4 Courts may
`
`also, when necessary, consider extrinsic evidence, which “consists of all evidence external to the
`
`patent and prosecution history, including expert and inventor testimony, dictionaries, and learned
`
`
`4 Where, as here, the specification incorporates by reference additional materials, such as articles,
`the materials incorporated by reference are also treated as if those materials were set forth in the
`specification, and thus is also intrinsic evidence. See, e.g., Telemac Cellular Corp. v. Topp
`Telecom, Inc., 247 F.3d 1316, 1329 (Fed. Cir. 2001) (“When a document is ‘incorporated by
`reference’ into a host document, such as a patent, the referenced document becomes effectively
`part of the host document as if it were explicitly contained therein.”).
`
`
`
`7
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 12 of 28
`
`
`treatises.” Id. at 1317. However, “the ordinary meaning of claim language as understood by a
`
`person of skill in the art may be readily apparent even to lay judges, and claim construction in such
`
`cases involves little more than the application of the widely accepted meaning of commonly
`understood words.” Id. at 1314.
`
`B.
`
`Claims Are Definite Unless They Fail to Inform Those of Skill in the Art of the
`Scope of the Invention with Reasonable Certainty
`Defendants assert that some of the claims are invalid for indefiniteness. The Supreme
`
`Court has held that the claims of a patent are indefinite under 35 U.S.C. § 112 only if the “claims,
`
`read in light of the specification delineating the patent, and the prosecution history, fail to inform,
`
`with reasonable certainty, those skilled in the art about the scope of the invention.” Nautilus, Inc.
`
`v. Biosig Instruments, Inc., 572 U.S. 898, 901 (2014). The definiteness requirement “mandates
`
`clarity, while recognizing that absolute precision is unattainable.” Id. at 910. “[D]efiniteness is to
`
`be evaluated from the perspective of someone skilled in the relevant art,” and “in assessing
`
`definiteness, claims are to be read in light of the patent’s specification and prosecution history.”
`
`Id. at 908. Defendants bear the burden of proving their indefiniteness defense by clear and
`
`convincing evidence. Wellman, Inc. v. Eastman Chem. Co., 642 F.3d 1355, 1366 (Fed. Cir. 2011).
`
`The “general principles of claim construction apply” to the definiteness inquiry, which is a
`
`question of law that may be underpinned by subsidiary factual determinations concerning any
`
`extrinsic evidence presented by the parties. Eli Lilly & Co. v. Teva Parenteral Meds., Inc., 845
`
`F.3d 1357, 1370 (Fed. Cir. 2017); see also MeadWestVaco Corp. v. Rexam Beauty & Closures,
`
`Inc., 731 F.3d 1258, 1270 (Fed. Cir. 2013) (“constru[ing] the court’s denial of [defendant’s]
`
`motion for summary judgment of indefiniteness as a denial for the failure to meet the summary
`
`judgment standard”). As such, invalidation of patent claims as indefinite is precluded when “the
`
`extrinsic evidence is sufficient to give rise to a genuine issue of material fact as to indefiniteness.”
`
`
`
`8
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 13 of 28
`
`
`Sys. Mgmt. Arts Inc. v. Avesta Techs., Inc., 137 F. Supp. 2d 282, 400-02 (S.D.N.Y. 2001); see also
`
`Rembrandt Data Techs., LP v. AOL, LLC, 641 F.3d 1331, 1343 (Fed. Cir. 2011) (reversing the
`
`district court’s grant of summary judgment of indefiniteness because expert testimony created a
`
`genuine factual dispute concerning how a skilled artisan would understand certain disclosures in
`
`the patent specification); Dow Chem. Co. v. Nova Chems. Corp. (Canada), 809 F.3d 1223, 1226
`
`(Fed. Cir. 2015) (“[D]efiniteness . . . is amenable to resolution by the jury where the issues are
`
`factual in nature.”) (quoting BJ Servs. Co. v. Halliburton Energy Servs., Inc., 338 F.3d 1368, 1372
`(Fed. Cir. 2003)).
`
`IV. AGREED CONSTRUCTIONS
`
`As reflected in the Amended Joint Claim Construction Chart (Dkt. No. 1465 at 2), the
`parties have reached agreement on the construction of the following claim terms:
`
`Claim Term
`“sublinear” [search]
`
`“neighbor”
`“near neighbor”
`
`
`“near neighbor
`search”
`
`
`“approximate nearest
`neighbor search”
`
`Agreed Construction
`“A search whose execution time scales with a less than linear
`relationship to the size of the data set to be searched, assuming
`computing power is held constant.”
`“A close, but not necessarily exact or the closest, match of a feature
`vector, compact electronic representation, or set of extracted features to
`another, wherein the distance or difference between the two feature
`vectors, compact electronic representations, or sets of extracted features
`falls within a defined threshold.”
`“A search using an algorithm designed to identify a close, but not
`necessarily exact or the closest, match of a feature vector, compact
`electronic representation, or set of extracted features to another, wherein
`the distance or difference between the two feature vectors, compact
`electronic representations, or sets of extracted features falls within a
`defined threshold.”
`“A search using an algorithm designed to identify a close, but not
`necessarily exact or the closest, match of a feature vector, compact
`electronic representation, or set of extracted features to another, wherein
`the distance or difference between the two feature vectors, compact
`electronic representations, or sets of extracted features falls within a
`defined threshold.”
`
`
`5 Citations to the docket correspond to documents filed in Case No. 14 Civ. 2396 (PGG).
`
`
`
`9
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 14 of 28
`
`
`Claim Term
`“machine-readable
`instructions”
`
`Agreed Construction
`“code or pseudocode that is executed using a computer processor, i.e.,
`that is discernable by a computer processor and dictates steps to be
`carried out by one or more computer processors”
`
`
`DISPUTED CONSTRUCTIONS
`
`V.
`
`The following are the claim terms as to which the parties offer differing constructions,
`requiring the Court’s interpretation.
`
`A.
`
`“non-exhaustive [. . .] search”
`
`Network-1’s Proposed Construction
`“A search designed to locate a [near] neighbor
`without comparing to all possible matches (i.e.,
`all records in the reference data set), even if the
`search does not locate a [near] neighbor.”
`
`
`Defendants’ Proposed Construction
`Indefinite.
`
`Defendants assert this claim term renders the asserted claims containing it (claim 17 of the
`
`’988 patent and claims 1, 8, 10, 16, 18, 25, 27, and 33 of the ’464 patent) invalid for indefiniteness.
`
`To prove indefiniteness, Defendants must establish by clear and convincing evidence that one of
`
`ordinary skill in the art would not find the term “non-exhaustive search” understandable with
`
`reasonable certainty. See Wellman, 642 F.3d at 1366; see also, e.g., Eidos Display, LLC v. AU
`
`Optronics Corp., 779 F.3d 1360, 1365 (Fed. Cir. 2015) (reversing district court’s finding of
`
`indefiniteness because a person of ordinary skill would have understood the claim limitation at
`
`issue).
`
`As discussed in detail below, the specification explains this term in much detail. See, e.g.,
`
`’988 patent at 21:6-22:37. “Non-exhaustive search” is also a term of art understood by skilled
`
`artisans in the field. See Mitzenmacher Decl. ¶¶ 40-41, 44-55. Indeed, both the Patent Trial and
`
`Appeal Board (“PTAB”) of the United States Patent and Trademark Office (“USPTO”) and the
`
`Federal Circuit were able to construe this term in proceedings involving the patents-in-suit. The
`
`act of performing a “non-exhaustive search” would be readily understandable to skilled artisans,
`
`
`
`10
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 15 of 28
`
`
`and the patents-in-suit provide “reasonable certainty[ to] those skilled in the art about the scope of
`the invention.” See Nautilus, 572 U.S. at 901.
`
`1.
`
`The Intrinsic Evidence Confirms Network-1’s Construction of
`“Non-Exhaustive Search”
`The patents explain the concept of “non-exhaustive search” in the specification. ’988
`
`patent at 8:60-9:55 (§ 4.2.1.1.3), 21:6-22:37. The specification explains that an exhaustive search
`
`potentially requires comparing the query to every record in a data set to be searched, describing
`
`such searches as “a linear search of all N entries.” Id. at 9:24-26 (emphasis added). On average,
`
`where a search could be halted after a match is found, this would require N/2 comparisons (where
`
`N is the number of records in the data set). Id. at 9:26-27. A non-exhaustive search, therefore, is
`
`one where the search query is compared to fewer records in the data set, and will not be compared
`
`to every record in the data set being searched. See Mitzenmacher Decl. ¶ 42.
`
`The specification provides several examples of non-exhaustive search methodologies.
`
`These methodologies include: binary search, clustering, kd-trees, vantage point trees, excluded
`middle vantage point forests, and searches for neighbors:
`
`Exemplary Non-Exhaustive
`Search Method from the
`Specification
`binary search (see id. at 9:19-22,
`21:23-39)
`
`clustering (see id. at 9:29-38,
`21:13-17)
`
`Explanation of the Exemplary Search Method
`
`“If binary search was possible, then a database containing
`N vectors would require at most log(N) comparisons.” Id.
`at 9:19-20. “A binary search of a 90,000,000-item database
`requires less than 20 comparisons. In contrast, a linear
`search will require an average of 45,000,000!” Id. at
`21:37-39.
`Clustering does not require comparing every record in a
`data set.6
`
`
`6 The patents-in-suit incorporate by reference Duda & Hart for its extraction and matching
`techniques, and Fukunaga for its extraction technique. Both incorporated references provide
`further information on clustering. See Ex. 4, Richard O. Duda & Peter E. Hart, Pattern
`Classification and Scene Analysis at 189-260 (Chapter 6, entitled “Unsupervised Learning and
`
`
`
`
`11
`
`
`
`
`
`Case 1:14-cv-02396-PGG-MHD Document 148 Filed 05/30/19 Page 16 of 28
`
`
`Explanation of the Exemplary Search Method
`
`Exemplary Non-Exhaustive
`Search Method from the
`Specification
`kd-trees (see id. at 9:29-38, 22:1-9) A kd-tree is a data structure that enables searching less than
`the full data set before finding, or not finding, a match.7
`A traditional vantage point tree is a data structure that
`returns the nearest neighbor without searching the entire
`data set.8
`An excluded middle vantage point forest, like a traditional
`vantage point forest, does