`
`
` 1 UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
` 2
` BEFORE THE PATENT TRIAL AND APPEAL BOARD
` 3
` ______________________________________________________
` 4
` APPLE, INC., |
` 5 |
` Petitioner, |
` 6 | Case IPR2018-01281
` vs. |
` 7 | U.S. Patent No.
` QUALCOMM INCORPORATED, | 8,768,865
` 8 |
` Patent Owner. |
` 9 ______________________________________________________
`
`10
`
`11 DEPOSITION OF
`
`12 JAMES F. ALLEN, PH.D.
`
`13
` Thursday, April 25, 2019
`14
` 9:00 a.m.
`15
`
`16 Suite 2100
` 1180 Peachtree Street, NE
`17 Atlanta, Georgia
`
`18
` Tammy G. Mozley, CCR-B-1032, RPR
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Apple Inc. v. Qualcomm Incorporated
`IPR2018-01282
`Qualcomm Ex. 2003
`Page 1 of 189
`
`
`
`1
`
`
` 1 UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
` 2
` BEFORE THE PATENT TRIAL AND APPEAL BOARD
` 3
` ______________________________________________________
` 4
` APPLE, INC., |
` 5 |
` Petitioner, |
` 6 | Case IPR2018-01281
` vs. |
` 7 | U.S. Patent No.
` QUALCOMM INCORPORATED, | 8,768,865
` 8 |
` Patent Owner. |
` 9 ______________________________________________________
`
`10
`
`11 DEPOSITION OF
`
`12 JAMES F. ALLEN, PH.D.
`
`13
` Thursday, April 25, 2019
`14
` 9:00 a.m.
`15
`
`16 Suite 2100
` 1180 Peachtree Street, NE
`17 Atlanta, Georgia
`
`18
` Tammy G. Mozley, CCR-B-1032, RPR
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 1 of 189
`
`
`
`2
`
`
` 1 APPEARANCES OF COUNSEL
`
`
`
` 2
`
` 3 On behalf of the Petitioner:
`
` 4 NOAH G. GRAUBART, Esq.
` Fish & Richardson, P.C.
` 5 Suite 2100
` 1180 Peachtree Street
` 6 Atlanta, Georgia 30309
` 404 892-5005
` 7 graubart@fr.com
`
` 8 BAILE XIE, PH.D., Esq.
` Fish & Richardson, P.C.
` 9 Suite 5000
` 1717 Main Street
`10 Dallas, Texas 75201
` 214 760-6139
`11 xie@fr.com
`
`12
` On behalf of the Patent Owner:
`13
` DANIEL LEVENTHAL, Esq.
`14 Norton Rose Fulbright US, LLP
` Suite 5100
`15 1301 McKenney
` Houston, Texas 77010-3095
`16 713 651-5151
` daniel.leventhal@nortonrosefulbright.com
`17
`
`18 EAGLE ROBINSON, Esq.
` Norton Rose Fulbright US LLP
`19 Suite 1100
` 98 San Jacinto Boulevard
`20 Austin, Texas 78701-4255
` 512 474-5201
`21 eagle.robinson@nortonrosefulbright.com
`
`22
`
`23 - - -
`
`24
`
`25
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 2 of 189
`
`
`
`3
`
`
` 1 TABLE OF CONTENTS
`
`
`
` 2 EXAMINATION PAGE
`
` 3 Examination by Mr. Leventhal 4
` Examination by Mr. Graubart 141
` 4
`
` 5 - - -
`
` 6
`
` 7 EXHIBIT DESCRIPTION PAGE
`
` 8
` Exhibit 1023 Declaration of 137
` 9 Dr. James F. Allen
`
`10 Exhibit 2001 U.S. Patent Office 13
` Confirmation No. 1445
`11
` Exhibit 2002 Declaration of 43
`12 Dr. Earl Sacerdoti
`
`13
`
`14 (Original Exhibits 2001, 2002, and 1023 have been
` attached to the original transcript. Copies of prior
`15 marked Exhibits 1001, 1003, 1005, 1011, and 1021 have
` been attached to the original transcript)
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 3 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`4
`
`
` 1 (Reporter disclosure made pursuant to
`
` 2 Article 10.B of the Rules and Regulations of the
`
` 3 Board of Court Reporting of the Judicial Council
`
` 4 of Georgia.)
`
` 5 JAMES F. ALLEN, PH.D.,
`
` 6 having been first duly sworn, was examined and
`
` 7 testified as follows:
`
` 8 EXAMINATION
`
` 9 BY MR. LEVENTHAL:
`
`10 Q. Good morning, Dr. Allen.
`
`11 A. Good morning.
`
`12 Q. Dr. Allen, in front of you, you have five
`
`13 exhibits. Exhibit 1001 is the '865 patent. Do you
`
`14 see that?
`
`15 A. Yes.
`
`16 Q. Exhibit 1003 is your declaration in
`
`17 IPR 2018-01281. Do you see that?
`
`18 A. 01281, I don't see this on this page.
`
`19 Q. Okay. That may not be on there, but
`
`20 that's fine. Okay.
`
`21 So that is your declaration, correct,
`
`22 that's marked?
`
`23 A. Declaration regarding what?
`
`24 Q. That's your declaration in this case
`
`25 that's been marked Exhibit 1003 in the IPRs we're here
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 4 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`5
`
`
` 1 for.
`
` 2 A. Okay. I don't remember that number. So
`
` 3 if it's -- can you tell me the contents?
`
` 4 Q. Sure. It's at the bottom, bottom right
`
` 5 corner.
`
` 6 A. Great, I see it. We're on board. Okay.
`
` 7 Q. And the second declaration you have is
`
` 8 marked Apple or Exhibit 1021. Do you see that?
`
` 9 A. Correct.
`
`10 Q. Okay. And one of those declarations --
`
`11 and we'll get into which one -- relates primarily to
`
`12 your opinions regarding primary reference called Wang,
`
`13 correct?
`
`14 A. Correct.
`
`15 Q. And you have a copy of Wang in front of
`
`16 you as Exhibit 1005?
`
`17 A. Yes, I do.
`
`18 Q. And the second declaration relates to a
`
`19 primary reference called Louch?
`
`20 A. Correct.
`
`21 Q. And you have a copy of that in front of
`
`22 you marked as Exhibit 1011, correct?
`
`23 A. That is correct.
`
`24 Q. Okay. Now, I just mentioned Wang and
`
`25 Louch. Had you ever heard of either Wang or Louch
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 5 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`6
`
`
` 1 prior to being provided these references as part of
`
` 2 these IPRs?
`
` 3 A. No. I don't think so.
`
` 4 Q. Now, each of the declarations,
`
` 5 Exhibit 1003 and 1021, mention a number of references
`
` 6 in addition to Wang and Louch, correct?
`
` 7 A. Yes. Correct.
`
` 8 Q. Had you heard of any of those references
`
` 9 before beginning your work on these IPRs?
`
`10 A. If I may look at the list.
`
`11 Q. Yes. If it helps, in Exhibit 1003, you
`
`12 may want to look at paragraph 8 starting on page 3.
`
`13 A. Yes. Exactly. No, I didn't know any of
`
`14 these before the proceedings. However, I did find
`
`15 several of these myself.
`
`16 Q. You anticipated my next question.
`
`17 A. Yes.
`
`18 Q. So my next question was going to be: Who
`
`19 provided you these references? How were they located?
`
`20 A. Well, all the patents were provided by the
`
`21 attorneys, in the declarations. I found the Cohn
`
`22 reference, which is reference M; the Ruzzeli, N
`
`23 reference; and the Cilla reference; and the last data,
`
`24 the definition in the IEEE standards.
`
`25 Q. And so just so I did not miss anything,
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 6 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`7
`
`
` 1 going by the sub-letters in the bullet list in
`
` 2 paragraph 8 --
`
` 3 A. Yes.
`
` 4 Q. -- you said the ones you located were M,
`
` 5 N, O, and P?
`
` 6 A. Correct.
`
` 7 Q. Why did you want to cite Cohn, which is
`
` 8 listed as item M in your declarations?
`
` 9 A. What I was attempting to do is to provide
`
`10 background and evidence that there was
`
`11 state-of-the-art in machine learning, various concepts
`
`12 referred to in the declaration.
`
`13 Q. Now, are you aware that until, I believe,
`
`14 yesterday there was a parallel litigation involving
`
`15 the '865 patent, in which Qualcomm was asserting the
`
`16 '865 patent against Apple?
`
`17 A. A second, no. I don't think so.
`
`18 Q. You weren't aware that there was an
`
`19 ongoing litigation between the parties?
`
`20 A. Oh, you mean a trial court?
`
`21 Q. Yes.
`
`22 A. Yes. I'm aware of that, yes.
`
`23 Q. What materials did you review from the
`
`24 trial court?
`
`25 A. I haven't seen anything from the trial
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 7 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`8
`
`
` 1 court.
`
` 2 Q. And there are places in here where you
`
` 3 cite infringement contentions in the complaint.
`
` 4 A. Uh-huh.
`
` 5 Q. So did you review the complaint in that
`
` 6 case?
`
` 7 A. I have, working with my attorneys on the
`
` 8 document, I had information from that, on that.
`
` 9 Q. Besides the excerpts of the complaint that
`
`10 are specifically cited --
`
`11 A. Yes.
`
`12 Q. -- in your declaration, did you review any
`
`13 other portions of that complaint?
`
`14 A. I haven't reviewed the other, the
`
`15 complaint.
`
`16 Q. Who provided the excerpt of the complaint
`
`17 to you for inclusion in your declaration?
`
`18 A. Well, it would have come from Baile right
`
`19 here.
`
`20 Q. That's fine. And counsel is fine. You
`
`21 don't need to --
`
`22 A. Counsel.
`
`23 Q. -- specifically identify.
`
`24 A. Oh, okay. Yes.
`
`25 Q. I wasn't trying to get that level of
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 8 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`9
`
`
` 1 detail from you.
`
` 2 A. Okay.
`
` 3 Q. Likewise, do you know what infringement
`
` 4 contentions are?
`
` 5 A. In general?
`
` 6 Q. In general. And I'm going to actually --
`
` 7 let me just direct you to item C on page 4 of your
`
` 8 declaration, Exhibit 1003, where it says Exhibit 865
`
` 9 to Qualcomm's patent initial infringement contentions.
`
`10 Do you see that?
`
`11 A. Yes.
`
`12 Q. What portions of Qualcomm's initial
`
`13 infringement contentions did you review for purposes
`
`14 of this declaration?
`
`15 A. I remember looking at the document. I
`
`16 don't -- I did a quick scan of it. I did not read it
`
`17 carefully.
`
`18 Q. Okay. Now, are you aware that Apple
`
`19 provided invalidity contentions in the litigation?
`
`20 A. I am aware there were invalidity
`
`21 contentions, yes.
`
`22 Q. Did you review any portion of those
`
`23 invalidity contentions?
`
`24 A. I didn't.
`
`25 Q. Now, at the top of paragraph 8, on page 3
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 9 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`10
`
`
` 1 of Exhibit 1003, before the bulleted list, it says
`
` 2 you've reviewed the '865 patent. Do you see that?
`
` 3 A. Uh-huh.
`
` 4 Q. Which portions of the '865 patent did you
`
` 5 review?
`
` 6 A. All of it.
`
` 7 Q. Okay. And then it says you reviewed
`
` 8 relevant excerpts of the prosecution history of the
`
` 9 '865 patent. Do you see that?
`
`10 A. Yes.
`
`11 Q. Which -- first of all, what does
`
`12 "relevant" mean when you say that?
`
`13 A. It means excerpts I found relevant. We're
`
`14 talking about the record of the patent examiner.
`
`15 Q. So did you review the entire file history
`
`16 or just excerpts?
`
`17 A. Just excerpts.
`
`18 Q. And who provided you those excerpts?
`
`19 A. My attorneys.
`
`20 Q. Did you do any independent review of the
`
`21 entire file history to determine if you would have
`
`22 found any other portions --
`
`23 A. No.
`
`24 Q. -- relevant. Is there any way I can tell
`
`25 from this -- anywhere in this declaration,
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 10 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`11
`
`
` 1 Exhibit 1003 or Exhibit 1021, which excerpts you
`
` 2 reviewed? That's not a trick question. I just
`
` 3 couldn't find it anywhere.
`
` 4 A. No, but the -- at least the excerpts that
`
` 5 I quote in this document.
`
` 6 Q. Okay. As you sit here, can you identify
`
` 7 any other excerpts besides the ones you quote?
`
` 8 A. No, I can't.
`
` 9 Q. Can you rule out that there were
`
`10 additional ones?
`
`11 A. That I read?
`
`12 Q. That you read.
`
`13 A. No, I can't. I read many different
`
`14 documents here. So I can't remember everything that I
`
`15 read if I didn't find it relevant.
`
`16 Q. Now, going down to paragraph 9, which is
`
`17 on page 5 of the Exhibit 1003, you state you
`
`18 understand that and -- I'm sorry, let me go -- this is
`
`19 four lines down.
`
`20 I understand that the '865 patent claims
`
`21 priority to US Provisional Application Number
`
`22 61/434,400. Do you see that?
`
`23 A. Yes.
`
`24 Q. Did you review that provisional
`
`25 application?
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 11 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`12
`
`
` 1 A. I did look at the provisional application,
`
` 2 yes.
`
` 3 Q. It's not listed in paragraph 8. That's
`
` 4 why I'm asking.
`
` 5 A. Okay. Well, that is an omission then.
`
` 6 Q. Okay. And Apple did not include it as an
`
` 7 exhibit to this -- to the IPR petitions we're talking
`
` 8 about here today. Are you aware of that?
`
` 9 A. To the IPR, you said?
`
`10 Q. Yes.
`
`11 A. Yes. No, I wasn't aware of that, but --
`
`12 Q. And I cannot find anywhere in this
`
`13 petition, Exhibit 1003 or Exhibit 1021, any citations
`
`14 to anything in the provisional application?
`
`15 A. That's correct.
`
`16 Q. So you didn't cite to it?
`
`17 A. No. I didn't find it very useful.
`
`18 Q. You didn't find it useful, but you did
`
`19 review it?
`
`20 A. Yes.
`
`21 Q. Okay. Why did you not find it useful?
`
`22 A. It is too vague and the claims have almost
`
`23 nothing to do with claims in the final patent.
`
`24 Q. So you reviewed the claims of the
`
`25 provisional application?
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 12 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`13
`
`
` 1 A. I read it.
`
` 2 Q. You read the entirety of it?
`
` 3 A. Yes. I read through it, yes.
`
` 4 Q. Okay. What makes you think that the
`
` 5 claims of the provisional application are relevant?
`
` 6 A. Well, if you provide me a copy of it, I'll
`
` 7 gladly go through it.
`
` 8 MR. LEVENTHAL: Sure. So we're going to
`
` 9 mark this Exhibit 2001.
`
`10 (Exhibit 2001 was marked for
`
`11 identification.)
`
`12 Q. (By Mr. Leventhal) So what I've handed to
`
`13 you, Exhibit 2001, is the provisional application with
`
`14 the number 61/434,400. Do you see that? So if you
`
`15 look in the top left-hand corner of the first page,
`
`16 you'll see the application number.
`
`17 A. I see an attorney docket number.
`
`18 Q. So keep going to the left on that same
`
`19 line. The furthest over to the left is application
`
`20 number.
`
`21 A. No. Oh, there we are, yes. Uh-huh. And
`
`22 you said it was?
`
`23 Q. 61/434,400.
`
`24 A. Correct.
`
`25 Q. And that is the same application that you
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 13 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`14
`
`
` 1 mentioned in paragraph 9 of your declaration?
`
` 2 A. That is correct.
`
` 3 Q. Please do flip through this and then I
`
` 4 want to make -- confirm your prior testimony that
`
` 5 you've seen this before.
`
` 6 A. Can you explain what you want me to do?
`
` 7 Q. Just flip through it to make sure.
`
` 8 Before, you said you looked through all of it
`
` 9 previously.
`
`10 A. Yes.
`
`11 Q. I want you to flip through the document
`
`12 that I've put in front of you, and then I'll re-ask
`
`13 you that question.
`
`14 A. Oh, okay. Yes. Okay.
`
`15 Q. So having reviewed Exhibit 2001, did
`
`16 you -- did you review and consider 2001 in issuing
`
`17 your declarations?
`
`18 A. Yes. I did read this, yes.
`
`19 Q. But you found none of it to be -- I don't
`
`20 want to put words in your mouth, but what did you say
`
`21 before? What was your view of it?
`
`22 A. You'll have to look back on the record to
`
`23 what I said before, but I did not find this
`
`24 informative beyond the patent.
`
`25 Q. Okay. Okay. I'd like to turn to a few
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 14 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`15
`
`
` 1 pages forward in your declaration to paragraph 16 of
`
` 2 Exhibit 1003.
`
` 3 A. Okay.
`
` 4 Q. Okay. So in paragraph 16 -- in
`
` 5 paragraph 16, you say there's an initial phase where
`
` 6 you convert raw signals into more abstract
`
` 7 representations called features. Do you see that?
`
` 8 A. Yes.
`
` 9 Q. In terms of the terminology that the
`
`10 patent uses, is there any other synonyms for features?
`
`11 I mean, is that similar to what the patent uses as
`
`12 variables or varying parameters or parameters?
`
`13 MR. GRAUBART: Object to form.
`
`14 THE WITNESS: Can you repeat the question?
`
`15 Q. (By Mr. Leventhal) Sure. I'm just
`
`16 listing some of the terminology that the patent -- and
`
`17 I'll list them one by one. Counsel had a good
`
`18 objection there.
`
`19 So when you talk about that there's a
`
`20 conversion of raw signals into abstract
`
`21 representations often called features, do you see
`
`22 that?
`
`23 A. Yes. I do.
`
`24 Q. The word that you use in your declaration,
`
`25 "features," is that similar to how the '865 patent
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 15 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`16
`
`
` 1 uses the word "variable and "varying parameter"?
`
` 2 A. Yes. It could be.
`
` 3 Q. Okay. Paragraph 17, you say the next
`
` 4 typical phase of activity recognition is mapping these
`
` 5 features into possible activities and that's often
`
` 6 called classification. Do you see that?
`
` 7 A. Yes.
`
` 8 Q. Now, further down in paragraph 17, you say
`
` 9 Wang discloses classification, correct?
`
`10 A. Correct.
`
`11 Q. Now, Wang does not disclose performing
`
`12 classification using any machine learning at all, does
`
`13 it? In Wang, the classification is manually done by a
`
`14 human typing in an XML file, correct?
`
`15 A. No.
`
`16 Q. Why is that not correct?
`
`17 A. Well, if we want to look at Wang, I think
`
`18 you are confusing what the word "classification"
`
`19 means. There is a sensing of what classification
`
`20 could be a human putting a label on it, but
`
`21 classification and machine learning is actually taking
`
`22 a learned model and identifying an output that's
`
`23 similar to the learned model.
`
`24 And we see actually in Section 6.3 of
`
`25 Wang, where he goes through and describes the process
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 16 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`17
`
`
` 1 by which they learn the features, where actually they
`
` 2 collect accelerometer data. They have people mark
`
` 3 what they're doing, and then they learn what features
`
` 4 correspond to those activities.
`
` 5 Q. Okay. And please do stick on 6.3.
`
` 6 A. Uh-huh.
`
` 7 Q. 6.3 repeatedly says "we," wasn't this
`
` 8 just --
`
` 9 A. Sorry.
`
`10 Q. 6.3 uses the word "we."
`
`11 A. We.
`
`12 Q. Isn't 6.3 describing the machine gathering
`
`13 data and the researchers reading it and the
`
`14 researchers manually making conclusions based on that?
`
`15 A. Not at all.
`
`16 Q. Where do you see anything where it says
`
`17 the machine makes conclusions?
`
`18 A. Okay. It computes the standard deviation
`
`19 of different activities within the empirical interval
`
`20 that's already doing computation on the signal in
`
`21 order to learn, essentially to determine a parameter.
`
`22 Q. But that's a standard deviation --
`
`23 A. No.
`
`24 Q. -- that's not the conclusion that walking
`
`25 equals moving slowly, correct?
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 17 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`18
`
`
` 1 A. No. It's not, no, but it is a parameter.
`
` 2 Q. So let's go to -- while we have Wang out,
`
` 3 let's go to Table 1 which is on page 6 if you don't
`
` 4 mind.
`
` 5 A. Uh-huh. Sorry. Table 1 on 6?
`
` 6 Q. Yes.
`
` 7 A. Okay. Sorry.
`
` 8 Q. So you have a number of state new names
`
` 9 here and you have state features, correct?
`
`10 A. Correct.
`
`11 Q. A human is who decided to define walking
`
`12 as location equals keep on changing and motion equals
`
`13 moving slowly, correct?
`
`14 A. That is correct.
`
`15 Q. What would you call that? I appreciate I
`
`16 used the word "classification" wrong. What would you
`
`17 call that?
`
`18 A. I would call that building pattern search
`
`19 to be stored in the system or database which are then
`
`20 used to match the activity.
`
`21 Q. I love this industry. I'm disappointed
`
`22 there is not a shorter phrase for it, a shorter label.
`
`23 Is there a shorthand we can use for that
`
`24 as we talk about it?
`
`25 A. Well, you could say they're defining
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 18 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`19
`
`
` 1 patterns.
`
` 2 Q. Okay. And while we're here, the
`
` 3 right-hand column of Table 1 says sensors monitored.
`
` 4 Do you see that?
`
` 5 A. Yes.
`
` 6 Q. It was human programming to determine
`
` 7 which sensor should be monitored when a pattern is
`
` 8 active, correct?
`
` 9 A. I think that is not clear from the paper.
`
`10 They do -- they talk about learning combinations of
`
`11 sensors if we go back to 6.3. Well, yeah, 6.3 only
`
`12 talks about accelerometers. So I think the paper does
`
`13 not say one way or another on where those come from.
`
`14 Q. So just a couple more things while we're
`
`15 on page 6, and we'll come back to this later. So if
`
`16 you look at Figure 3 --
`
`17 A. Yes.
`
`18 Q. There is a box at the top left that says
`
`19 XML state descriptor. Do you see that?
`
`20 A. Yes.
`
`21 Q. You'd agree with me that that XML file in
`
`22 Wang is entirely generated by a human manually,
`
`23 correct?
`
`24 A. That is generated by a human, yes.
`
`25 Q. And the last question I have on this, and
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 19 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`20
`
`
` 1 we'll come back to it later, is in Table 1. It says
`
` 2 sensors monitored. It was unclear to me whether those
`
` 3 are the sensors that need to be monitored to detect
`
` 4 this -- the particular state in the row or whether
`
` 5 those are the sensors that should be monitored after
`
` 6 that particular state has been detected and while
`
` 7 you're looking for the next state. Do you have any
`
` 8 view on what is being shown in that last column
`
` 9 labelled sensors monitored?
`
`10 A. Those are definitely -- they include the
`
`11 sensors that are needed to identify the state.
`
`12 Q. Okay.
`
`13 A. It is -- what was the second part of your
`
`14 question?
`
`15 Q. So the second question is -- and let me
`
`16 take a step back. The purpose of Wang generally is to
`
`17 identify state transitions, correct?
`
`18 MR. GRAUBART: Object to form.
`
`19 THE WITNESS: No.
`
`20 Q. (By Mr. Leventhal) Okay. A large part of
`
`21 Wang in identifying a second state is by detecting a
`
`22 transition from a first state. Is that fair?
`
`23 A. It uses the context of the first state to
`
`24 efficiently attempt to recognize a second state.
`
`25 Q. Based on a transition from the first
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 20 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`21
`
`
` 1 state?
`
` 2 A. I'm not sure what that question means,
`
` 3 but -- what I just said is true, whether you call that
`
` 4 a transition or not.
`
` 5 Q. Well, I'm trying to use the terminology of
`
` 6 the patent and if you look on page -- excuse me, of
`
` 7 the paper and if you look on page 5 in the label of
`
` 8 Figure 2.
`
` 9 A. Of figure -- sorry. 5?
`
`10 Q. See, it says Figure 2, the sequential
`
`11 sensor management rules used to detect state
`
`12 transitions.
`
`13 A. Yes.
`
`14 Q. Do you see that?
`
`15 A. Uh-huh.
`
`16 Q. So the disclosure, at least a portion of
`
`17 the disclosure of Wang relates to being in a first
`
`18 state and detecting a transition to a second state?
`
`19 A. Which is a -- which means exactly the same
`
`20 thing as recognizing a first state and then
`
`21 recognizing a second state based on that first state.
`
`22 So there's no additional information from the fact
`
`23 it's called a transition.
`
`24 Q. A transition doesn't mean the first state
`
`25 ended?
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 21 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`22
`
`
` 1 A. No.
`
` 2 Q. You can have a transition from a first
`
` 3 state to a second state and the first state can still
`
` 4 be ongoing?
`
` 5 A. It still could be true, yes.
`
` 6 Q. Is that the case in Wang?
`
` 7 A. Yes. I believe that is the case in Wang.
`
` 8 Q. That was a bad question. Let's go
`
` 9 specifically back to Table 1.
`
`10 A. Uh-huh. Table 1.
`
`11 Q. Table 1, which is on page 6.
`
`12 A. Yes.
`
`13 Q. The bottom two states are walking and
`
`14 vehicle. Do you see that?
`
`15 A. Yes, I do.
`
`16 Q. Those two states cannot overlap in time,
`
`17 can they?
`
`18 A. They cannot.
`
`19 Q. Okay. All right. And we will come back
`
`20 to Wang, but I just had that one question while we
`
`21 were on paragraph 17. We can go back to the
`
`22 declaration now.
`
`23 A. Okay.
`
`24 Q. So we're going to go back to page 10 of
`
`25 the declaration.
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 22 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`23
`
`
` 1 A. Uh-huh.
`
` 2 Q. And paragraph 18 refers to something
`
` 3 called training.
`
` 4 A. Yes.
`
` 5 Q. Can you please, just at a high level, kind
`
` 6 of the tutorial you've done here, explain what
`
` 7 training is?
`
` 8 A. Training is -- as explained in more detail
`
` 9 in later paragraphs, but training is a step of all
`
`10 machine learning algorithms, where essentially you
`
`11 start with data.
`
`12 And let's take the easy case to
`
`13 understand. You start with data that has been
`
`14 labelled in some way. So you know this set of data
`
`15 corresponds to this label. So in this context, you
`
`16 might have a set of parameter values and walking or
`
`17 something is a label.
`
`18 And so the training phase is essentially
`
`19 running algorithms that essentially try and build a
`
`20 statistical model off the data that best shows how
`
`21 the -- that best shows, given the label, how
`
`22 accurately you could predict the data.
`
`23 Q. Okay. That's helpful. And the reason I
`
`24 ask that is because I'm having a bit of confusion of
`
`25 we started with the classifications -- sorry. We
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 23 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`24
`
`
` 1 started with the defining pattern step, and then we
`
` 2 had classification. Is training then a discrete third
`
` 3 step after those first two?
`
` 4 A. Actually, I think your question -- you're
`
` 5 saying -- no, talking about machine learning. Where
`
` 6 does it talk about patterns first?
`
` 7 Q. So this is in paragraph 18, and perhaps
`
` 8 that's my problem. I couldn't tell if training was a
`
` 9 third step since paragraph 16 was a first step, 17 was
`
`10 a second step. I couldn't tell if paragraph 18 was a
`
`11 third step.
`
`12 A. No. The training is a way, as it says, to
`
`13 identify the patterns.
`
`14 Q. Okay.
`
`15 A. So there's not separate things there. The
`
`16 patterns are kind of loosely, very loosely defined at
`
`17 this level here. They may be -- they might not look
`
`18 like any pattern you could -- we could, even as
`
`19 humans, we could understand. It's just statistical
`
`20 properties of data.
`
`21 Q. So the two steps that we discussed
`
`22 earlier, building patterns and classification, in a
`
`23 machine learning system, those can be done at least in
`
`24 part through training?
`
`25 A. In machine learning systems, they're done
`
`www.LexitasLegal.com/Premier Lexitas 888-267-1200
`
`Page 24 of 189
`
`
`
`James Allen , PH.D. - April 25, 2019
`
`25
`
`
` 1 almost always by training.
`
` 2 Q. Okay. And we discussed in Wang the
`
` 3 machine learning portion is kind of after those first
`
` 4 two steps but maybe improving, maybe improving the
`
` 5 patterns --
`
` 6 A. No. Again, you're confusing
`
` 7 classification.
`
` 8 Q. Okay.
`
` 9 A. There is labelling.
`
`10 Q. Labelling?