UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`Page 186
`
`LIBERTY MUTUAL INSURANCE
`
`COMPANY,
`
`. CBM2012—00002
`
`CBM2012—00004
`
`(JL)
`
`Petitioner,
`
`Patent 6,064,970
`
`CBM2013-0004 {JL)
`
`Patent 8,090,598
`
`PROGRESSIVE CASUALTY
`
`INSURANCE COMPANY,
`
`. CBM2012-0003
`
`Patent Owner.
`
`Patent 8,140,358
`
`CBM2013—0O09
`
`(JL)
`
`VIDEOTAPED DEPOSITION OF SCOTT ANDREWS
`
`Palo Alto, California
`
`Tuesday, September 24, 2013
`
`Volume 2
`
`Reported by:
`
`LESLIE ROCKWOOD, RPR, CSR 3462
`
`Job No. 65807
`
`TSG Reporting — Worldwide
`
`877-702-9580
`Liberty Mutual Exhibit 1048
`
`Liberty Mutual v. Progressive
`CBM2013-00009
`
`Pa;e00001
`
`Liberty Mutual Exhibit 1048
`Liberty Mutual v. Progressive
`CBM2013-00009
`Page 00001
`
`

`
`Page 195
`
`Page 196
`
`
`
`RDL1.--.2OWL‘!-3-1.1.‘N?l—‘
`
`»—A C
`|..I l—‘
`|—I {NJ
`._A La:
`l—‘
`.a
`l—' L’:
`
`I-(DKQCD--.lfl\lJ"-£:§4i|'-,)>-I
`
`n—-
`n.-
`l-‘ l\J
`n—- 1.:-'
`,_. as
`r—-
`<.,1
`,. G‘.
`-4
`.-_-
`
`{Exhibit Liberty Mutual 1004, Japanese
`Unexamined Patent Application Publication,
`H4-132868. llfl9i'90, pages I - 42, having
`been previously marked, was referred to.)
`BY MR. WAMSLEY:
`
`Q. We also have Liberty Mutual Exhibit 102] in
`that matter. which is an excerpt from a book on fuzzy
`logic by Yen and Langari.
`Am I correct‘?
`A. That's correct.
`
`[Exhibit Liberty Mutual I021. Fuzzy Logic,
`Intelligence, Control, and In formation, Yen
`and Langari, pages I — 55, having been
`previously marked, was referred to.)
`BY MR. WAMSLEY:
`
`Q. And then finally we have a paper called "Black
`Magic." which is Liberty Mutual Exhibit 1008 in this
`matter; is that correct?
`A. That's correct.
`
`(Exhibit Liberty Mutual 1008. An Interest
`in Black Magic - Motor Technology, pages 1
`- 2, having been previously marked, was
`referred to.)
`BY MR. WAMSLEY:
`
`Q. Okay.
`
`I'd like to direct you to your rebuttal
`
`MR. MYERS: Objection. 402, 403. And for the
`record, as in previous depositions, I'm simply going to
`cite the number of the Federal Rule of Evidence going
`forward in the deposition rather than make a full
`citation or state a full grounds for my objection, I'll
`simply state the rule number.
`MR. WAMSLI-lY: Well, let me just follow up to
`clarify. You're not intending to reserve the right to
`assert a different objection later to my question, are
`you?
`
`MR. MYERS:
`
`I‘m not going to assert a difierent
`
`rule.
`
`MR. WAMSLEY: Okay. So any objection within
`that rule is what you're saying?
`MR. MYERS: Correct.
`
`MR. WAMSl.F.Y: Okay.
`MR. MYERS:
`I -- it's my --
`MR. WAMSLEY: Now we understand each other.
`
`MR. MYERS: Right. My understanding is the
`Patent Trial and Appeal Board doesn't want speaking
`objections or a full explanation on thcrecord, and as a
`consequence, I'm going to give you the mle number of
`the Federal Rule of Evidence that I'm objecting under.
`And then if that comes up, then I'll have the
`opportunity to explain the basis for that objection in
`
`declaration, Mr. Andrews, Exhibit I019.
`A. Okay.
`Q. And in particular to paragraph 6. And in the
`first sentence of that paragraph, you testify that fuzzy
`logic was weli-established and fairly common by 1996.
`Do you see that‘?
`I see that.
`
`A.
`
`Q. Okay. And is the basis for that opinion the
`existence of the book by Wang called "Adaptive Fuzzy
`System and Control," dated 1994'.’
`A. Well, actually the basis for that is described
`in the subsequent paragraph. Part of it is the book by
`Wang. Let me find it here. Yes. part of it is the book
`by Wang. Part of it is the book by Langari and Yen.
`Part of it is from my own experience leading a group of
`engineers that were doing work with fuzzy logic.
`Q. All as described in this paragraph; is that
`right?
`A. Yes.
`
`Q. Okay. The book that you cite which is
`Exhibit 1021 by Langari and Yen --
`A. That's right.
`Q.
`-- what you have quoted there indicates that
`the book takes the view that fuzzy logic is an emerging
`technology; correct‘?
`
`Page 198
`
`either in front of the board or in a paper that's tiled
`with the board if it becomes necessary.
`MR. WAMSLEY: We understand each other. then,
`
`Jim. Thank you for the clarification.
`Could I ask you to read the question back,
`please.
`(The record was read by the reporter
`as follows:
`
`"QUESTION: What you have quoted there
`indicates that the book takes the view that
`
`fuzzy logic is an emerging technology;
`e0rrect‘?")
`‘II [E WITNIESS:
`
`l wouldn't characterize it that
`
`way. Actually, it says that it's been accepted as an
`emerging technology since the late 1980s.
`BY MR. WAMSLIEY:
`
`Q. And this is as of 1999, when this book was
`published; correct?
`A. That's correct, I think, yes.
`Q. Now you say in the next sentence that by 1996,
`you had studied several fuzzy logic systems and
`supervised many engineers with similar fuzzy logic
`experience.
`Do you see that?
`I see that.
`
`A.
`
`TSG Reporting — Worldwide
`
`877-702-9580
`
`4
`
`(Pages 195 to 198)
`
`Page 00002
`
`Page 00002
`
`

`
`Page 219
`
`Page 220
`
`identify these particular parameter values associated
`with Kosaka's membership functions.
`Do I have that right?
`I think almost. Ithink I said that last time
`A.
`we did this.
`
`I think the way I've stated it here in the
`declaration is not that a person of skill in insurance
`would have that ability.
`I've stated that in order to
`determine these values, you would need someone who was a
`person of ordinary skill in the insurance aspects of
`this kind of system.
`Again, it's not just any old person who knows
`something about insurance; it's somebody who is actually
`knowledgeable about, for example, understanding the
`risks associated with following distances and swcrving
`and the other parameters that Kosaka identifies here.
`So you'd need a person who was knowledgeable
`about risks associated with that so that they would then
`be able to actually determine what these values are.
`And that's what I mean by a person skilled in the
`insurance aspects of the ‘W0 patent.
`Q. But in fact, you have no expertise that would
`allow you to testify whether that person knowledgeable
`about those risks that you just referred to would be an
`expert or instead someone with lesser skill, do you‘?
`
`MR. MYERS: Objection. 402, 403.
`'l'I'lE WI'lNI~lSS: Are you asking me ifl would be
`able to determine whether a given person was an expert
`versus a person of ordinary skill in those aspects‘?
`BY MR. WAMSLEY:
`
`Q. That's a different question than the one I
`asked.
`
`A. Okay.
`MR. WAMSLEY: Let me try having it read back,
`and if it's still not working, we'll rephrase.
`(The record was read by the reporter
`as follows:
`
`"QUESTION: But in fact, you have no expertise
`that would allow you to testify whether that
`person knowledgeable about those risks that you
`just referred to would be an expert or instead
`someone with lesser skill, do you‘?"}
`MR. MYERS: Objection. 402, 403.
`THE WITNESS:
`I guess probably not because the
`delineation of a person of ordinary skill versus
`expertise in insurance isn't really my field.
`BY MR. WAMSLEY:
`
`Q. Okay. Let's move on to paragraph 9 of your
`rebuttal declaration. And here, among other things, you
`testify as to the risk evaluation value in Kosaka;
`
`Page 221
`
`Page 222
`
`correct’?
`
`A. That's right.
`Q. And in your opinion, you say a person of
`ordinary skill would understand that risk evaluation
`value to be a single crisp value; concct?
`A. That's whatl said.
`
`Q. And that's because of what you describe in the
`next sentence there of the process called
`defuzzifioation.
`
`Am lright?
`MR. MYERS: Objection. 402, 403.
`THE WITNESS: The process called
`defuzzification is the process that would take the
`membership -- the output membership values, membership
`function values, and convert them into a single crisp
`value.
`BY MR. WAMSLEZY:
`
`Q. And with that understanding, am I correct that
`it's because of that, the existence of that
`defuzzilieation process. that you are of the opinion
`that I(osaka's risk evaluation value would be a single
`crisp value‘?
`MR. MYERS: Objection. 402. 403.
`THE WITNESS:
`I'm not sure that I would
`characterize it that way.
`It‘s not because of the
`
`dcfuzzification process. The issue is that there would
`be no usable output until you defuzzified it.
`BY MR. WAMSLEY:
`
`In your testimony in your declaration about
`Q.
`defuzzification, you cite to the Langari book; correct?
`A. That's right.
`Q. So let's look at that. That's Exhibit [02].
`Arid you particularly cite a couple of pages
`there. Let's look at the first such citation at page
`38. Tell me when you're there.
`A. I'm there.
`
`Q. And you see the reference to defuzzification in
`the middle of the page there; correct‘?
`A. Yes.
`
`Q. And according to this text, this is an optional
`step in fuzzy logic; correct‘?
`A.
`'l'ha1's what it says here.
`Q. So in that respect, a designer would be free to
`not use defuzzification as part of the fuzzy logic
`system.
`Am I right?
`MR. MYERS: Objection. 402, 403.
`TI IE WIINESS:
`I don't think it really says
`I mean, he says for applications that need at
`that.
`crisp output, for example, in control systems. So any
`
`TSG Reporting - Worldwide
`
`(Pages 219 to 222)
`10
`877-702-9580
`
`Page 00003
`
`Page 00003
`
`

`
`Page 223
`
`Page 224
`
`time you are going to ultimately try to make use ofthc
`output in some specific way. you need a value that you
`can use. You don't need percentages of membership in a
`membership function.
`I don't know what you would -- in ‘970, I don't
`know how you would determine an insurance premium based
`on the notion that someone was 20 percent low risk and
`50 percent medium risk and 70 or 30 percent high risk.
`You would ultimately have to calculate what is the
`aggregate risk from that, which is ultimately getting a
`crisp value out of the fuzzy system.
`I think the fact that he says these are
`optional is more if you were having cascaded Fuzzy logic
`fixnctions, you don't necessarily have to defuzzify and
`refuzzify and defuzzify and refuzzify every single time.
`But at the end ofthe day, having an output of
`a luazy system that isn't a value that you can use isn't
`very useful.
`BY MR. WAMSLEY:
`
`Q. You made a reference to control systems in your
`last answer. You'd agree with me that the way fuzzy
`logic is used in Kosaka is it's not controlling
`anything, is it‘?
`MR. MYERS: Objection. 402, 403.
`'l‘IIE WI‘l'Nl:lSS:
`1 think it‘s ultimately
`
`controlling the insurance premium.
`BY MR. WAMSLEY:
`
`Q. You -- so, in your opinion, coming up with risk
`evaluation values that are then used in insurance
`
`premium calculation is an example of a control system?
`A.
`I mean, I could take you through my logic on
`that, but it's not a control system as in a -- you know,
`a stability control for an airplane or something like
`that or a cruise control system, but in fact, it is --
`Q. Or an elevator control system‘?
`A. Right. But it is in fact something ofa
`feedback system. if you consider that you are going to
`measure risk and the associated potential for loss
`associated with that and then decide what factors in
`
`driving contribute to that, you are actually ultimately
`building a system that is a control system. Because if
`you base your premiums on the -- on these factors in the
`right way, then eventually new drivers are going to
`drive in that way, and you'll be able to assess their
`risk accurately.
`So at the end of the day, you have to have a
`crisp value to assign some level of risk. Youjust
`can't think that a system that has real-world
`application is going to end up with a membership set
`function and you're going to use that. So somewhere you .
`
`Page 225
`
`Page 226
`
`have to defuzzify this.
`Q. And in your rebuttal declaration in
`paragraph 9, you say: "Kosaka explicitly describes
`using defuzzification."
`. Do you see that?
`' A. Yes, I do.
`Q. So you've still got Kosaka in fiont ofyou;
`right, Mr. Andrews?
`A. Uh-h uh.
`
`Q. Would you agree with me that the mention of
`defuzzification that you've cited to at page 8 of Kosaka
`is with respect to Kosaka's first fuzzy logic part 62 as
`shown in Figure 9'?
`A. That's correct.
`
`Q. Would you also agree with me that nowhere else
`does Kosaka mention using defuzzification processes with
`respect to any other output?
`A. Well. he says the logical output level says
`the -- this is in the right-hand paragraph of page 8,
`second paragraph down. So the risk evaluation value
`resulting from a comprehensive determination carried out
`at this third l‘tt7.7.y logic part 65 is then output to the
`output controller, 66, where the logical output level
`and the output in accordance with hold time level are
`sent to the warning device and the monetary amount lile.
`
`So he's talking about an output level. He's
`not talking about a series of membership function
`values. I'm not sure what a warning device or what the
`controller would do with a series ofmembership values.
`And you asked about a control system earlier in
`relation to Langari, and he's actually saying you output
`it to an output controller. So maybe it is a control
`system.
`He doesn't say explicitly here that the output
`of the -- or the resulting membership function from
`fuzzy logic unit 3 is defuzzified, but I don't think he
`needs to say that.
`Q. You understand in looking at Figure 9 -- and
`feel free to consult the accompanying text -- you agree
`with me that the inputs to Box 65, which is fuzzy logic
`unit 3, are themselves fitzzy values?
`A. He talks about that in the top of page 8. So
`you'll see these are also input as fuzzy input values.
`Q. And that's what you would expect; right?
`Because l'u?J.y values are -- being used in fuzzy logic
`unit I and 2: right?
`A. That is what you would expect. You could have
`them be completely freestanding. So you could implement
`fuzzy logic I as a standalone unit that takes analog
`inputs or even digital representations of analog input
`
`TSG Reporting — Worldwide
`
`(Pages 223 to 226)
`11
`877-702-9580
`
`Page 00004
`
`Page 00004