IPR2023-00033
`Apple Inc.
`v.
`Zentian Limited
`Patent 7,587,319
`
`Patent Owner’s Demonstratives
`
`Presented March 11, 2024
`
`1
`
`
`Apple Inc.
`v.
`Zentian Limited
`Patent 7,587,319
`
`Patent Owner’s Demonstratives
`
`Presented March 11, 2024
`
`1
`
`
`
`Argument Roadmap
`
`v The evidence proves a POSA would not have had a
`reasonable expectation of success
`v Petitioner’s counter-arguments rest on false logic
`v The evidence proves no motivation to combine
`
`2
`
`
`
`Argument Roadmap
`
`v The evidence proves a POSA would not have had a
`reasonable expectation of success
`v Petitioner’s counter-arguments rest on false logic
`v The evidence proves no motivation to combine
`
`3
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`The Petition’s combination of Thelen, Bailey, and Chen has key requirements
`
`Architecture
`
`1.
`
`2.
`
`3.
`
`4.
`
`Group Thelen’s recognizers into groups of recognizers all associated with
`the same context (e.g., sports)
`Port each recognizer group into a cluster of processors, each processor
`cluster connected to a dedicated cluster memory
`Store all recognition models for one context (e.g., sports) in one cluster
`memory (partial lexical data)
`Data stored across all cluster memories constitutes all recognition
`models for all contexts (complete lexical data)
`
`Sur-reply at 1-2; Pet. at 40-43; Ex. 1003 ¶ 83; Ex. 2017, 124:13-126:5, Ex. 2020 ¶¶ 43-44
`
`4
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`The Petition’s combination of Thelen, Bailey, and Chen has key requirements
`
`Architecture
`
`Sur-reply at 1-2; Pet. at 40-43; Ex. 1003 ¶ 83; Ex. 2017, 124:13-126:5, Ex. 2020 ¶¶ 43-44
`
`5
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`The Petition’s combination of Thelen, Bailey, and Chen has key requirements
`
`Operational requirements
`
`1. Every processor in a cluster must simultaneously
`request “the particular subset of vocabulary . . . that’s
`germane to its context” from the single shared cluster
`memory (e.g., golf, tennis, football, basketball). Ex.
`2017, 125:8-126:5, Ex. 2020 ¶¶ 43-44;
`2. Combination’s speech recognition must operate in
`real time, as Thelen did. Ex. 1003 ¶ 84, Ex. 2017,
`154:8-155:17, Ex. 2020 ¶ 42.
`
`Sur-reply at 1-2
`
`6
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`Neither Thelen, Bailey, nor Chen enables Petitioner’s combination
`
`Thelen and Bailey do not teach groups
`of processors each connected to one
`lexical memory
`
`Thelen does not teach how to run its
`techniques (including its recognizers)
`on groups of clustered processors
`
`Sur-reply at 2; Ex. 2017, 120:8-18, 136:7-16; Ex. 2020 ¶¶30-31
`
`7
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`Neither Thelen, Bailey, nor Chen enables Petitioner’s combination
`
`Chen does not teach how to implement any speech
`recognition on its hardware, much less Petitioner’s
`combination with Thelen
`
`Sur-reply at 2; Ex. 2017, 120:19-121:2; Ex. 2020 ¶¶30-31
`
`8
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`Mr. Schmandt’s admissions refute a reasonable expectation of success
`
`Mr. Schmandt admitted he did not:
`
`Mr. Schmandt admitted he did not:
`
`v Explain how a POSA could
`implement parallel operation of
`Thelen’s multiple recognition
`models simultaneously within
`Chen’s clusters, Ex. 2017, 137:21-
`138:24;
`
`v Identify any challenges the POSA
`would face in making the
`combination, Ex. 2017, 141:23-
`142:2;
`
`v Identify the tasks the POSA would
`need to undertake to make the
`combination, Ex. 2017, 144:3-20;
`
`v Analyze “the tasks that would be
`required” of the POSA before he
`reached his conclusion of a
`reasonable expectation of
`success, Ex. 2017, 144:22-145:3;
`
`Sur-reply at 2-3
`
`9
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`Mr. Schmandt’s admissions refute a reasonable expectation of success
`
`Mr. Schmandt admitted he has not:
`
`Mr. Schmandt admitted he has not:
`
`v Supervised anyone involved in
`mapping a speech recognition
`model to a clustered processor
`and memory architecture like
`Chen’s, Ex. 2017, 145:20-24;
`
`v Built the processor to memory
`architecture of any speech
`recognition system identified in his
`background experience, Ex. 2017,
`34:24-35:5, 32:3-9, 106:10-23;
`
`Sur-reply at 2-3
`
`10
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`Dr. Anderson’s testimony disproves a reasonable expectation of success
`
`v Petitioner’s combination would
`cause extensive memory collisions
`
`v POSA would have known that the
`combination could slow down speech
`recognition compared to Thelen alone
`
`v every processor in a cluster
`would be required to
`simultaneously request “the
`particular subset of vocabulary
`. . . that’s germane to its
`context” from a single
`memory, Ex. 2017, 125:8-
`126:5, Ex. 2020 ¶¶ 43-44
`
`v Petitioner’s combination would not
`operate in real time, Ex. 2020 ¶ 45
`
`v Mathew I taught that adding
`processors to a parallel processing
`speech recognition system slowed
`down the system, Ex. 2011 at 11,
`Ex. 2020 ¶ 46
`
`v Even Dr. Anderson, a person of
`extraordinary skill, and a team of
`engineers including a DSP expert, could
`not successfully port a far simpler
`signal processing software to a parallel
`processing architecture after six
`months, Ex. 2020 ¶¶ 40-41
`
`Sur-reply at 3-4
`
`11
`
`
`
`An ordinary artisan would not have had
`a reasonable expectation of success
`
`Dr. Anderson’s testimony disproves a reasonable expectation of success
`
`v POSA could not have reasonably
`addressed the numerous
`challenges arising from the
`combination, Ex. 2020 ¶¶ 38-39
`
`v Memory conflicts, controlling task
`sharing, addressing
`communication bandwidth and
`latency problems, and developing
`a messaging strategy across
`clusters
`
`v Person of ordinary skill specialized
`in digital signal processing and
`speech recognition, not parallel
`processing architectures or high
`performance computing, Ex. 2020
`¶ 39
`
`Sur-reply at 3-4
`
`12
`
`
`
`Argument Roadmap
`
`v The evidence proves a POSA would not have had a
`reasonable expectation of success
`v Petitioner’s counter-arguments rest on false logic
`v The evidence proves no motivation to combine
`
`13
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 1: POSITA must be assumed to have requisite knowledge because
`’319 Patent did not address Petitioner’s problems
`
`The problems with Petitioner’s combination arise out of
`the combination, not the invention of the ’319 Patent
`
`The ’319 Patent does not teach Petitioner’s specific
`combination, and its silence cannot support Petitioner
`
`Sur-reply at 5-11
`
`14
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 2: A POSA’s general knowledge of parallel processing architectures
`for speech recognition would have sufficed to achieve the combination
`
`None of the cited prior art taught how to achieve real time
`recognition by porting groups of same-context recognizers
`into a clustered processor/memory architecture
`
`General knowledge of parallel processing architectures is
`not adequate knowledge to achieve the combination
`
`Sur-reply at 11-14
`
`15
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 2: A POSA’s general knowledge of parallel processing architectures
`for speech recognition would have sufficed to achieve the combination
`
`v Petitioner’s combination would
`cause extensive memory collisions
`
`v POSA would have known that the
`combination could slow down speech
`recognition compared to Thelen alone
`
`v every processor in a cluster
`would be required to
`simultaneously request “the
`particular subset of vocabulary
`. . . that’s germane to its
`context” from a single
`memory, Ex. 2017, 125:8-
`126:5, Ex. 2020 ¶¶ 43-44
`
`v Petitioner’s combination would not
`operate in real time, Ex. 2020 ¶ 45
`
`v Mathew I taught that adding
`processors to a parallel processing
`speech recognition system slowed
`down the system, Ex. 2011 at 11,
`Ex. 2020 ¶ 46
`
`v Even Dr. Anderson, a person of
`extraordinary skill, and a team of
`engineers including a DSP expert, could
`not successfully port a far simpler
`signal processing software to a parallel
`processing architecture after six
`months, Ex. 2020 ¶¶ 40-41
`
`Sur-reply at 11-14
`
`16
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 2: A POSA’s general knowledge of parallel processing architectures
`for speech recognition would have sufficed to achieve the combination
`
`Mr. Schmandt admitted he has not:
`
`Mr. Schmandt admitted he has not:
`
`v Supervised anyone involved in
`mapping a speech recognition
`model to a clustered processor
`and memory architecture like
`Chen’s, Ex. 2017, 145:20-24;
`
`v Built the processor to memory
`architecture of any speech
`recognition system identified in his
`background experience, Ex. 2017,
`34:24-35:5, 32:3-9, 106:10-23;
`
`Mr. Schmandt’s conclusion of reasonable success is not credible
`
`Sur-reply at 11-14
`
`17
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 3: Skilled persons knew of various speech recognition hardware
`architectures
`
`Prior art is only presumptively enabled for what it teaches
`
`Chen, Thelen, and Bailey do not enable the combination
`
`Chen cannot be assumed to be enabled for every speech
`recognition architecture imaginable
`
`Sur-reply at 14-15 (gathering cases); Ex. 2020 ¶¶ 30-48; Ex. 2017, 120:8-121:2, 136:7-16
`
`18
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 4: “Zentian does not argue that the combination is inoperable, [or]
`could not be done”
`
`Patent Owner need only show Petitioner’s failure of proof
`
`Patent Owner need not show impossibility
`
`Petitioner conflates ordinary skill with extraordinary skill
`
`Sur-reply at 15-16 (gathering case law)
`
`19
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 5: Zentian’s alleged complications would arise with any clustered
`processor architecture
`
`Petitioner has no evidence to support its allegation
`
`Petitioner again attempts to improperly shift its burden
`
`The combination requires real time recognition but the
`combination’s design prevents real time recognition
`
`Sur-reply at 16-19, Ex. 2020 ¶¶ 42-48, Ex. 1003 ¶¶ 83-84, Ex. 2017 at 124:13-126:5, 137:21-138:24, 141:23-142:2, 144:22-145:3, 154:8-155:17
`
`20
`
`
`
`Petitioner’s arguments rest on false logic
`
`Pet. Arg. 6: Zentian’s argument regarding memory collisions and slower
`processing does not account for multiple processors
`
`The fact of multiple processors as utilized within the
`combination causes slower processing, as in Mathew I
`
`Petitioner has no evidence that Chen’s multiple
`processors would “more than account for any memory
`contention issues” in the combination
`
`Petitioner’s expert admitted he could not say the
`combination would have more power than Thelen
`
`Sur-reply at 19-22, Ex. 2020 ¶¶ 44-46, 53-55, 59-60; Ex. 2021 at 11; Ex. 2017, 120:19-121:2, 130:23-131:4, 136:3-6
`
`21
`
`
`
`Argument Roadmap
`
`v The evidence proves a POSA would not have had a
`reasonable expectation of success
`v Petitioner’s counter-arguments rest on false logic
`v The evidence proves no motivation to combine
`
`22
`
`
`
`The evidence proves no motivation to combine
`
`Dr. Anderson’s testimony disproves a motivation to combine
`
`v The combination would cause
`extensive memory collisions that
`would cause it to be slower than
`Thelen, not faster
`
`v POSA would have known from
`Mathew I that more processors
`would not necessarily make parallel
`processing speech recognition
`faster
`
`v POSA would not have been
`motivated by a combination that
`would be slower than Thelen, and
`that would fail to achieve real time
`recognition
`
`v Thelen’s memory architecture
`would have been a poor fit for
`Chen’s multiple private memories
`
`v Petitioner’s combination requires a
`distributed memory model because
`it gives memory access across
`adjacent clusters
`
`v Petitioner’s combination would
`have required significant
`reprogramming effort that would
`have defeated a motivation to
`combine
`
`Sur-reply 22-23, 25-26, Ex. 2021 at 11, Ex. 2020 ¶¶ 51-53, 56-58, 59-60, 62, Ex. 2017 at 155:7-17, Pet. 38-39, Ex. 1005 at 9:24-28, Paper 21 at 21-22
`
`23
`
`
`
`The evidence proves no motivation to combine
`
`Pet. false motivation 1: increased processing power
`
`Mr. Schmandt admitted he could not say the combination
`would have more power than Thelen
`
`Sur-reply at 22, 24, Ex. 2017 at 130:23-131:4, 136:3-6
`
`24
`
`
`
`The evidence proves no motivation to combine
`
`Pet. false motivation 2: increased processing speed
`
`v POSA would have known that the
`combination could slow down
`speech recognition compared to
`Thelen alone
`
`v The combination would cause
`extensive memory collisions that
`would cause it to be slower than
`Thelen, not faster
`
`v Mathew I taught that adding
`processors to a parallel
`processing speech recognition
`system slowed down the
`system, Ex. 2021 at 11, Ex.
`2020 ¶ 46
`
`v Schmandt admitted that Thelen in
`unmodified form had no problems
`achieving real time recognition
`
`v POSA would not have been
`motivated by a combination that
`would be slower than Thelen, and
`would fail to achieve real time
`recognition
`
`Sur-reply at 22-24, Ex. 2021 at 11, Ex. 2020 ¶¶ 51-53, 59-60, 62, Ex. 2017 at 155:7-17
`
`25
`
`
`
`The evidence proves no motivation to combine
`
`Pet. false motivation 3: cost
`
`v Mr. Schmandt admitted he did not
`perform a quantitative or
`qualitative cost-benefit analysis
`
`v He further admitted he did not
`compare the cost of Chen’s
`architecture to any other clustered
`processing architectures
`
`v Petitioner cannot state what the
`combination would have been
`cheaper than, or what costs would
`have been “reduced”
`
`v Cost alone would not suffice as a
`motivation where combination
`system would be slower than
`Thelen alone, would fail to meet
`Thelen’s real time recognition
`requirement, would require
`significant reprogramming effort,
`and would not have more power
`than Thelen alone
`
`Sur-reply at 23, 25, Ex. 2017 at 42:14-19, 63:8-15
`
`26
`
`
`
`The evidence proves no motivation to combine
`
`Pet. false motivation 4: flexibility and scalability
`
`v Mr. Schmandt admitted that MARS
`was flexible because its pipeline
`was programmable, and scalable
`because more processors could be
`added to it
`
`v Mr. Schmandt further admitted that
`MARS was being compared to SRI-
`Berkeley, which did not have those
`features
`
`v Petitioner has no showing that
`Thelen was not already flexible
`(programmable pipeline) or
`scalable (more processors could be
`added)
`
`Sur-reply at 23, 25; Ex. 2017 at 105:8-106:4, Ex. 2020 ¶ 61
`
`27
`
`
`
`The evidence proves no motivation to combine
`
`Pet. false motivation 5: increased benefits over a single processor
`
`Thelen is already multi-processor, thus not relevant
`
`Sur-reply at 26
`
`28
`
`
`
`The evidence proves no motivation to combine
`
`Pet. false motivation 6: reduce duplication of lexical data
`
`Petitioner’s Reply states combination would require
`duplication of lexical data. Paper 21 at 21-22.
`
`Other problems with the combination would defeat a
`motivation to combine regardless
`
`Sur-reply at 26-27
`
`29
`
`
Accessing this document will incur an additional charge of $.
After purchase, you can access this document again without charge.
Accept $ Charge
Still Working On It
This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.
Give it another minute or two to complete, and then try the refresh button.
A few More Minutes ... Still Working
It can take up to 5 minutes for us to download a document if the court servers are running slowly.
Thank you for your continued patience.
This document could not be displayed.
We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.
Your account does not support viewing this document.
You need a Paid Account to view this document. Click here to change your account type.
Your account does not support viewing this document.
Set your membership
status to view this document.
With a Docket Alarm membership, you'll
get a whole lot more, including:
- Up-to-date information for this case.
- Email alerts whenever there is an update.
- Full text search for other cases.
- Get email alerts whenever a new case matches your search.
One Moment Please
The filing “” is large (MB) and is being downloaded.
Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.
Your document is on its way!
If you do not receive the document in five minutes, contact support at support@docketalarm.com.
Sealed Document
We are unable to display this document, it may be under a court ordered seal.
If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.
Access Government Site
