[00:00:02] Speaker 03:
We have a busy morning in front of us today.

[00:00:05] Speaker 03:
We have four cases scheduled for argument Our first case is Palo Alto networks versus Finjen 17-2059 Mr.. Armand you ready?

[00:00:23] Speaker 03:
Yes

[00:00:31] Speaker 02:
May it please the court.

[00:00:34] Speaker 02:
The court should vacate the final written decision in this IPR because the board failed to perform reasoned decision making as required by the Administrative Procedure Act.

[00:00:44] Speaker 02:
Palo Alto Networks obtained cross-examination admissions from Finjin's two experts, Dr. Medvedevich and Dr. Binz, so impactful to patentability that Finjin moved to exclude the testimony of its own experts.

[00:01:01] Speaker 02:
calling it highly prejudicial.

[00:01:04] Speaker 02:
That's Appendix 497.

[00:01:06] Speaker 02:
The board denied the motion to exclude, but then in its final written decision, without explanation, did not comment on any of Dr. Medvedevich's admissions and expressly stated that it did not rely on the BIMS cross-examination testimony.

[00:01:23] Speaker 03:
Wasn't that portion of Dr. BIMS testimony confined to

[00:01:29] Speaker 03:
secondary considerations of non-obviousness?

[00:01:33] Speaker 02:
It was directed to secondary considerations.

[00:01:35] Speaker 03:
And the court didn't answer that, did it?

[00:01:38] Speaker 03:
No, Your Honor, it did not.

[00:01:41] Speaker 03:
Why are you concerned that it did not comment on Dr. Bims' testimony?

[00:01:45] Speaker 03:
If it didn't reach those issues, why should we hold the court to comment on those?

[00:01:51] Speaker 02:
Because Dr. Bims had prepared infringement claim charts.

[00:01:55] Speaker 02:
That was the substance of his opinion testimony on secondary considerations.

[00:02:00] Speaker 02:
And in that infringement analysis, Dr. Binns was opining upon the scope of the claims.

[00:02:08] Speaker 02:
The issue is that patent claims can't, like a nose of wax, be treated one way for infringement and a different way for validity.

[00:02:19] Speaker 02:
And the reason that Binns' testimony was relevant here

[00:02:24] Speaker 02:
was that he was interpreting the patent claims for purposes of his secondary considerations infringement analysis in a way that was flatly inconsistent with the way that Dr. Medvedevich was interpreting the claims when he was reporting.

[00:02:39] Speaker 03:
Are you arguing that the board's required to comment on every piece of evidence that's submitted?

[00:02:45] Speaker 02:
Your Honor, we're arguing that, as in the Princeton-Vanguard versus Frito-Lay case,

[00:02:51] Speaker 02:
that reasoned decision making requires the board to do more than cherry pick evidence that supports the conclusion that a substantial evidence review requires the examination of the record as a whole, taking into account both the evidence that justifies and detracts from the agency's decision.

[00:03:11] Speaker 02:
The court in the Princeton Vanguard case said that the board can't shortcut its consideration of the factual record before it.

[00:03:20] Speaker 02:
More recently,

[00:03:21] Speaker 02:
This court in the last month decided the Altair Pharmaceuticals versus Paragon Biotech case.

[00:03:27] Speaker 02:
That was number 17-1487.

[00:03:30] Speaker 02:
In that case, the CEO of Altair, Mr. Al Sawaya, prepared two declarations that related to validity of the claim challenge there.

[00:03:42] Speaker 02:
The board refused to consider the second declaration at all and gave no weight to the first declaration.

[00:03:50] Speaker 02:
This court held that the PTAS decision to assign no weight to Mr. Alsalliah's trial testimony was an abuse of discretion.

[00:03:57] Speaker 02:
And importantly, and I think relevantly to this case, held that an agency's refusal to consider evidence bearing on the issue before it is by definition arbitrary and capricious, and that the agency must take account of all the evidence of record, including that which detracts from the conclusion the agency ultimately reaches.

[00:04:17] Speaker 02:
The board did not do that in this case.

[00:04:20] Speaker 02:
I'd like to draw the court's attention to one of the most clarion examples where the board not only failed to consider the evidence that detracted from its decision, but also where the board's finding wasn't supported by any evidence at all.

[00:04:34] Speaker 02:
The entire issue in this case turned on a claim limitation that requires dynamically building a parse tree.

[00:04:43] Speaker 02:
This is about malware detection.

[00:04:45] Speaker 00:
Can I interrupt you just to ask you one question?

[00:04:48] Speaker 00:
Do you see anything in the patent specification that talks about why it is that you want to dynamically build while receiving?

[00:04:58] Speaker 00:
Like what the advantages of doing that or anything that touts that as being important to the invention or anything?

[00:05:07] Speaker 02:
Your honor, I can't cite to any discussion in the patent disclosure on that.

[00:05:12] Speaker 02:
The dynamically limitations, and there are two, are not discussed.

[00:05:17] Speaker 02:
at any length in the patent disclosure or the claims.

[00:05:20] Speaker 02:
In fact, there's only one or two references at all to that issue.

[00:05:25] Speaker 02:
However, the cross-examination testimony of Dr. Medvedevich specifically addressed that question.

[00:05:31] Speaker 02:
And as we've explained in our briefing, Dr. Medvedevich agreed that going back decades now, in the field of malware detection, particularly when you're dealing with firewalls that are examining streams of data, it's of the utmost importance

[00:05:46] Speaker 02:
to attempt to analyze data in real time.

[00:05:49] Speaker 02:
And that goes directly to the question of dynamically building.

[00:05:54] Speaker 00:
What about the language that's referring to, for example, that the data stream is lexically analyzed again?

[00:06:02] Speaker 00:
And this reference to again several times in the prior, how do you respond to that?

[00:06:06] Speaker 00:
And why doesn't that provide substantial evidence to support the board's findings?

[00:06:11] Speaker 02:
The board's finding regarding lexical analysis is based upon the, and by the way, the board's finding cites to appendix 3091, paragraph 73 and 74.

[00:06:25] Speaker 02:
That's the only evidence that the board cites in support of its finding that the entire data stream representing a file must be analyzed to determine the language the code is in.

[00:06:37] Speaker 02:
Appendix 3091, paragraph 73 to 74, is Dr. Medvedevich's declaration.

[00:06:43] Speaker 02:
And the only thing that he says in that declaration is that the word again means that the stream is fully analyzed twice.

[00:06:53] Speaker 02:
And there is a citation to a dictionary definition.

[00:06:56] Speaker 02:
There is no technical analysis.

[00:06:58] Speaker 02:
There is no explanation.

[00:06:59] Speaker 02:
And he didn't speak in paragraph 73 or 74.

[00:07:03] Speaker 02:
to the issue of lexical analysis in terms of language detection at all, in terms of what's required.

[00:07:09] Speaker 02:
However, on cross-examination, because his opinions in paragraphs 73 and 74 were totally untethered to the reality of the underlying technology, we asked questions of Dr. Medvedevich about the context of this invention, what was known in the art.

[00:07:27] Speaker 02:
And he agreed, and he provided testimony, and this is covered at

[00:07:31] Speaker 02:
Appendix 438 and 439.

[00:07:33] Speaker 02:
We spent pages in our reply brief on this.

[00:07:36] Speaker 02:
He agreed that in the context of streaming data, the state of the art and understanding of the skill in the art was that streaming data travels over networks as packets and that all you need is a packet or a handful of packets to identify the programming language that a data stream is written in because

[00:08:00] Speaker 02:
We're talking about internet-based technologies.

[00:08:03] Speaker 02:
And for example, there's an HTML tag or a script tag that will be contained in a single packet or a series of the first few packets in a data stream.

[00:08:13] Speaker 02:
And so the opinion that the board cites to, or I'm sorry, the testimony of paragraph 73 and 74 that the board cites to not only doesn't reach the issue, the only basis for Dr. Medvedevich's opinion there is pointing to the word again,

[00:08:29] Speaker 02:
in the disclosure with no explanation why that would require complete analysis of an entire data stream in order to identify the language.

[00:08:39] Speaker 02:
Now, again, the cross-examination testimony is consequential here because, as explained in appendix 438, 439, and thereafter, also addressed in our blue brief at page 47,

[00:08:55] Speaker 02:
A single packet with an HTML tag is all that's necessary to identify the language so that you can apply the proper language rules to create your parse tree.

[00:09:08] Speaker 02:
Also, at appendix 439, there's discussion and citation to the evidence where Dr. Medvedevich testified on cross-examination that a person with skill in the art would have been motivated to speed up malware detection.

[00:09:23] Speaker 02:
that the goal is to completely process the data stream as it passed.

[00:09:26] Speaker 02:
Those cross-examination admissions were completely contrary and contradictory to the board's finding that the entire data stream, the entire file, has to be analyzed in order to determine the language that the code is written in.

[00:09:42] Speaker 02:
Now, I'd like to turn back briefly to Dr. Bims's cross-examination admissions.

[00:09:49] Speaker 02:
Again, he was addressing the

[00:09:50] Speaker 02:
he had been offered to address secondary considerations.

[00:09:53] Speaker 02:
But in the context of providing an infringement opinion with supporting claim charts where he was comparing the language of the claims that were petitioned against a product offered by a company called WebSense.

[00:10:07] Speaker 02:
So in the context of those infringement positions, Dr. Bims was cross-examined.

[00:10:12] Speaker 02:
And he testified, and this is at appendix 2761, that a person of ordinary skill would understand

[00:10:20] Speaker 02:
that techniques such as detecting, as well as analyzing, parse tree nodes would take place as the program code is being downloaded.

[00:10:27] Speaker 02:
So he provided an opinion that a person of skill in the art would have understood that you would dynamically analyze and build a parse tree.

[00:10:35] Speaker 02:
This is the testimony that the board indicated in its final decision it was not relying upon.

[00:10:43] Speaker 02:
Again, although we recognize that the issue of secondary considerations wasn't reached by the board,

[00:10:50] Speaker 02:
because of its finding, we still should not have a situation where the patent owners allowed to offer two different experts who offer completely contradictory testimony.

[00:11:01] Speaker 02:
And then the board decides not to address the contradictory testimony of the second expert solely because it was on secondary considerations.

[00:11:10] Speaker 02:
This is, as mentioned in the Amazon versus Barnes and Noble case, the nose of wax problem.

[00:11:16] Speaker 03:
The board cited to Shandani and said,

[00:11:20] Speaker 03:
to the part, and this is at 103, column 7, the file to be scanned, which may be stored.

[00:11:28] Speaker 03:
And it goes on and says, it does the storing before it's converted into a data stream.

[00:11:36] Speaker 03:
And why is that not substantial evidence that supports the board's finding?

[00:11:41] Speaker 02:
Because as addressed in our briefing, Ciannani has two different embodiments.

[00:11:46] Speaker 02:
One is a file-based embodiment.

[00:11:48] Speaker 02:
One is a network-based embodiment.

[00:11:50] Speaker 02:
I believe you're referring to Appendix 1466, where Ciannani explicitly discloses two embodiments, one virus detection on a data stream and another virus detection on a file.

[00:12:04] Speaker 02:
Even, Your Honor, in the case of the file, Ciannani expressly states that its virus detection methodologies.

[00:12:12] Speaker 03:
In that case, isn't the file first stored before it's converted?

[00:12:16] Speaker 02:
There are two embodiments, one embodiment,

[00:12:19] Speaker 02:
the file-based embodiment does in some circumstances involve taking a file and turning it into a data stream.

[00:12:27] Speaker 02:
Yes, but we were relying on the network-based embodiment, not on the file-based embodiment.

[00:12:34] Speaker 03:
But the citation I gave you speaks of a network.

[00:12:40] Speaker 03:
Information received via a network such as the internet or a wireless transmission medium

[00:12:46] Speaker 03:
such as telephone landlines or RF airways is converted to a data stream.

[00:12:51] Speaker 03:
So the file is scanned and then stored before conversion to a data stream.

[00:12:58] Speaker 02:
Your honor, I don't have the citation in front of me, but even Dr. Medvedevich agreed, and it is cited in our briefs, that the network-based embodiment was a separate embodiment from the file-based embodiment discussed in Chai Nani.

[00:13:11] Speaker 02:
In the network-based embodiment, the conversion of the file would happen at a server somewhere, not on the local computer.

[00:13:19] Speaker 02:
I'll reserve the rest of my time unless the court has any other questions.

[00:13:23] Speaker 03:
OK, thank you.

[00:13:29] Speaker 03:
Mr. Andre.

[00:13:32] Speaker 03:
May it please the court.

[00:13:39] Speaker 01:
The board made factual findings

[00:13:41] Speaker 01:
as to the scope and content for the prior art.

[00:13:44] Speaker 01:
They made factual finding as to the difference between the art and the invention.

[00:13:49] Speaker 01:
They made factual findings as to the level of one skill in the art, how that one skill in the art would view the art.

[00:13:57] Speaker 01:
That evidence that they cite to, the factual findings that they concluded, is substantial evidence.

[00:14:05] Speaker 01:
And that's what we're looking at.

[00:14:06] Speaker 01:
Is there substantial evidence to support the board's finding?

[00:14:08] Speaker 01:
That's what this appeal is about.

[00:14:12] Speaker 01:
There's nothing in Chandani, the primary reference that the appellant is relying on here, that would lead one to believe that they do dynamic building of a parse tree.

[00:14:25] Speaker 01:
In fact, the board made factual findings that Chandani does do actually file this, restore the file before they do any type of scanning.

[00:14:36] Speaker 01:
That goes with the network treatment.

[00:14:37] Speaker 03:
Under both embodiments?

[00:14:38] Speaker 01:
Excuse me?

[00:14:38] Speaker 01:
Under both embodiments?

[00:14:39] Speaker 01:
There's actually only one single embodiment.

[00:14:41] Speaker 01:
What Chandani does in column seven of the patent is it will take a file and create a data stream from the file after it's already stored.

[00:14:50] Speaker 01:
There's a single embodiment.

[00:14:51] Speaker 01:
There's not two embodiments.

[00:14:52] Speaker 01:
We disagree with my friend's characterization of that.

[00:14:56] Speaker 03:
And the reference that I cited, are you familiar with that?

[00:14:59] Speaker 03:
Yes.

[00:14:59] Speaker 03:
And does that go to a single embodiment?

[00:15:02] Speaker 03:
It does.

[00:15:04] Speaker 01:
And in fact, Judge Stoll talked about how the Chandani talks about how it scans it

[00:15:11] Speaker 01:
again, and that's key.

[00:15:14] Speaker 01:
You can't scan it again if it's not already in storage.

[00:15:17] Speaker 01:
So there's nothing in Chandani that would lead one to believe that you could actually dynamically build the parse tree as the stream was coming in.

[00:15:32] Speaker 01:
My friend talked about the expert depositions in cross-examinations.

[00:15:38] Speaker 01:
I want to address that very quickly.

[00:15:40] Speaker 01:
Dr. Bems was put up for the secondary considerations issue.

[00:15:44] Speaker 01:
And Dr. Bems discussed a product from WebSense, an infringing product, and WebSense has since taken a license to these patents.

[00:15:55] Speaker 01:
And he was talking about one skill in the art at the time of infringement.

[00:15:59] Speaker 01:
He was not talking about one skill in the art at the time of the invention, because he was looking at those skill in the infringement at the time, which was very recently,

[00:16:09] Speaker 01:
would look at this after the teaching of the patents, he was not put in to talk about one skill in the art earlier, because at secondary considerations, that wouldn't be relevant.

[00:16:19] Speaker 01:
As far as Dr. Medvedevich, he talked about a, once again, a web-sense product, talking about the speed of it, and then he talked about some HTML tags, which had nothing to do with this at all.

[00:16:33] Speaker 01:
In the briefs and in the oral argument, just to be clear,

[00:16:37] Speaker 03:
You're not arguing that objective indicia that occurs or relates to after the patent has been filed or after it's been issued, that it's not relevant?

[00:16:53] Speaker 01:
We're saying, because the board didn't reach the issue of secondary consideration, it's not relevant here.

[00:17:00] Speaker 01:
If they would have reached secondary considerations, obviously, it would have been.

[00:17:02] Speaker 01:
It's moot at this point.

[00:17:04] Speaker 01:
What they showed Dr. Benz,

[00:17:07] Speaker 01:
was declarations regarding a product that was released years after the filing date or the priority date of these patents.

[00:17:15] Speaker 01:
And they're asking that one skill in the art, how would they look at it at that time period?

[00:17:19] Speaker 01:
At what time period?

[00:17:20] Speaker 01:
At the time period of infringement, not at filing.

[00:17:22] Speaker 01:
And he was talking about how one skill in the art would look at it at the time of infringement, which is a very different equation because you do that for secondary considerations.

[00:17:29] Speaker 01:
The board didn't consider the BIMs because they didn't get to the secondary considerations.

[00:17:35] Speaker 01:
which is completely appropriate in this circumstance.

[00:17:40] Speaker 01:
And if you look at the actual testimony, both from the sighted in the breeze, they don't support the appellant's position that Chandani teaches dynamic building and parse tree.

[00:17:55] Speaker 01:
It just doesn't do it.

[00:17:57] Speaker 01:
It's completely unrelated to that issue.

[00:18:01] Speaker 01:
So even if there was something to be said about the board not looking at the cross-examination, which they did.

[00:18:06] Speaker 01:
And if you look at what the board actually cited to, this goes back and starts on page 18 of the appendix and going through page 26.

[00:18:18] Speaker 01:
They talk about the two different experts, appellants expert and our expert.

[00:18:25] Speaker 01:
And they weigh those experts' testimony and what they have been said provided.

[00:18:30] Speaker 01:
And they concluded that Dr. Medvedevich was more credible.

[00:18:35] Speaker 01:
His was more likely reading.

[00:18:37] Speaker 01:
The board is entitled, as this court has found in the Belden case, the board is entitled to rely on its own reading of prior art to find how skilled artisans would understand it.

[00:18:50] Speaker 01:
And in this case, that's exactly what the board did.

[00:18:52] Speaker 00:
I had asked a question earlier about the 408 patent

[00:19:00] Speaker 00:
and whether there was any disclosure in that patent of the dynamically building, you know, the importance of that or what, is there anything that you can point to in the specification that talks about why it's desirable, for example, to have the dynamic building step occurring while the receiving step is occurring?

[00:19:20] Speaker 01:
Yeah, in the column one in the background of the invention, it talks about how efficiency is important.

[00:19:26] Speaker 01:
And when I talk about identifying the problems

[00:19:28] Speaker 01:
And then I think in column two and again in column five, the patentee talks about how you can look at the day stream as it comes in.

[00:19:38] Speaker 01:
That's an important concept because in theory what you don't want to do is use unnecessary bandwidth.

[00:19:45] Speaker 01:
So if you can actually build the parse tree as the stream is coming in and be doing analysis at the same time,

[00:19:51] Speaker 01:
Once you detect an exploit or a malware, then you can cut it off right there.

[00:19:56] Speaker 01:
You can stop the process and stop it from coming into your system.

[00:20:00] Speaker 00:
I thought that the patent talked about efficiency generally.

[00:20:03] Speaker 00:
But do you have a specific site, so where it ties those concepts together?

[00:20:09] Speaker 01:
I don't have it at my fingertips.

[00:20:10] Speaker 01:
I'm sorry to say it wasn't raised in the briefing, so we didn't focus on that.

[00:20:15] Speaker 01:
But it does talk about generally the efficiency.

[00:20:18] Speaker 01:
And I think that's consistent with the briefing that was in the case.

[00:20:21] Speaker 01:
The dynamic building and then the next claim element, the dynamic detecting is very important concepts for malware because you obviously want to do it efficiently, quickly, and before malware can get into your computer system.

[00:20:35] Speaker 01:
If you do that while it's dynamically building, you can cut off before the whole file is brought into the system.

[00:20:48] Speaker 01:
If you have no further questions,

[00:20:51] Speaker 01:
I think I've said all I need to say.

[00:20:53] Speaker 01:
Thank you.

[00:20:58] Speaker 03:
Firmly, you've got three minutes.

[00:21:05] Speaker 02:
In the court's review, I'd ask the court to pass its aye over the board's findings in Appendix 20 and 21, in which it states that Ciannani requires multiple passes through the file.

[00:21:19] Speaker 02:
first to determine the appropriate script language, and then to lexically analyze the data stream to generate the stream of tokens.

[00:21:27] Speaker 02:
And it's finding at appendix 20 that it's not possible for the data stream to still be incoming when the data stream's lexically analyzed again.

[00:21:37] Speaker 02:
Both of those key findings in this case are based only upon Dr. Medvedevich's declaration, Testimony of Paragraphs 73 and 74.

[00:21:48] Speaker 02:
There's no evidence in either one of those paragraphs other than his observation that the word again means the file has to be reviewed fully twice.

[00:21:58] Speaker 02:
That's not substantial evidence to support the board's findings.

[00:22:02] Speaker 02:
And the board's failure to weigh the contrary evidence violated the Administrative Procedure Act.

[00:22:07] Speaker 02:
I have nothing else unless the court has other questions.

[00:22:09] Speaker 02:
Any questions?

[00:22:11] Speaker 02:
No, sir.

[00:22:11] Speaker 03:
Thank you.

[00:22:12] Speaker 03:
We thank the party for the arguments.