Mistakes to Avoid When Filing Computer-Implemented Invention Patents at the EPO

By Gene Quinn
July 25, 2019

If the invention is initially claimed only in broad terms, the examiner may narrowly read the claims to avoid prior art. If such mistakes are made, the subject-matter is claimed too broadly from the beginning and the examiner may unable to identify what and where to search. This can lead to lengthy prosecution.

Mistakes to Avoid with Computer-Implemented Inventions at the EPO. https://www.epo.org/about-us.htmlIn the final installment of my interview with three examiners at the European Patent Office (EPO), we wrap up our conversation about their approaches to examining computer implemented inventions, particularly in the field of artificial intelligence (AI), and how the EPO system compares with the U.S. patent examination system. To start reading this interview from the beginning see: How to Help an EPO Examiner and Improve Your Odds

The examiners, in the order in which they first speak in the interview that follows, are:

Jean-Marc Deltorn

Jean-Marc Deltorn (JMD) holds a PhD in Theoretical physics, a Master of Laws and is currently finishing a PhD in Law on the interplay between AI and privacy. Deltorn has been a patent examiner at the EPO in the Information and Communication Technologies (ICT) sector since 2003, working in a variety of fields, from image analysis to pattern recognition and machine learning and focusing currently on speech recognition. He is also is a member of the EPO Guidelines Working Group on computer-implemented inventions (CIIs), and has co-authored the course material on “patentability of artificial intelligence”.

Abderrahim Moumen

Abderrahim Moumen (AM) holds a master degree and a Ph.D. degree in the field of Telecommunications and Radars. He joined the EPO in 2000 as an examiner in the ICT sector. In 2017, he was appointed team manager leading a team of examiners dealing with patent applications in the field of antennas and microwaves. He specializes in the area of antennas (including smart antennas and antenna optimization), telecommunications, and the Internet of Things (IoT).

Anton Versluis

Anton Versluis (AV) studied Applied Physics with a specialization in Artificial Intelligence. He worked on Smart Robotics for the handicapped at the Netherlands Research Institution TNO and joined the EPO in 2006 as examiner in Pattern Recognition (G06K9), a technical area with a large role for AI, for instance in Self Driving Vehicles.

Below is Part III of my interview with Examiners Deltorn, Versluis and Moumen.

 

GQ:    The U.S. has kind of flatlined in terms of patent applications and at the EPO and in Europe generally, they’re increasing, because it seems to some level the laws in America have reached the point where they are just not patent friendly. It seems to me from an observer’s perspective that the EPO examiners are actually really interested in issuing patents and serving the applicants and doing business with them in a responsible way, and it’s not this adversarial system. Maybe I’m wrong, maybe it is, but if you have any thoughts on that I would be interested, or if you want to not answer that, that’s okay too, because maybe that’s a little bit too charged.

JMD:  The way EPO examiners go about an application is laid down our Guideline . By providing both examiners and applicants alike with a common platform for understanding the patentability requirements, we increase the efficiency of the examination process overall. Coming to a patentable invention is indeed a collegial effort that involves not only the three-member Examining Division, but also the applicant. It is therefore essential that we share a common understanding of our examination process. This communication-based approach is set out in our Guidelines, where we detail our practice as clearly as possible. That is not always evident, of course, in particular when the technical landscape evolves so quickly, as in the case for AI and blockchain. But we do our best to provide a clear overview of our examination practice under the EPC with predictable outcomes for the applicant.

AV:     What patent examiners are here to do is to render a reasoned decision to grant or to refuse a patent application based on publicly available rules and guidelines. Our mission is to implement a system of patenting developed to suit 38 different European states which, at the beginning, had very different ideas on the patent law. Our approach eventually resulted in a transparent and rules-based patent system where everything was laid down in writing. As for the problem-solution-approach, we also have a safeguard, called ex post facto analysis, which basically means that in retrospect, everything is obvious. This safeguard can be invoked by the applicant to state that hindsight reasoning should not be applied to say that once the problem is known, the solution is obvious. This is also an example of our principle that the benefit of the doubt always lies with the applicant. If we have a doubt about inventive step, or about whether something is sufficiently disclosed, the applicant has the benefit of the doubt. Except for one article, 123(2) EPC, which rules out that subject matter may be added to the application as a whole. But for all other requirements, it means that you start looking with the applicant for a way out.

[[Advertisement]]

Assessing Self-Driving Vehicles

GQ:    Yeah, and that is supposed to be the way it is here in America, but I think most patent attorneys would say that the benefit of the doubt is given to the rejection stamp in America. This is unfortunate a lot of times.

I have a few other specific questions I want to hit on here if we can, and they are probably relatively quick. When we are talking about autonomous driving or self-driving vehicles, is the approach any different or should it be any different, or should applicants just treat that like any other CII application?

JMD:  I would say that there’s no particular difference between the manner we approach AI applications and CII in general. What matters is that we offer a clear procedure for the applicant to follow, which goes back to the problem-solution approach. We’re trying to be as transparent as possible in the way we approach a file. With respect to the problem-solution approach in the context of mixed type inventions (inventions which include both technical and non-technical, e.g. mathematical, features), the examiner has to evaluate whether or not those abstract features contribute to solving a technical problem. We follow here the well-established procedure set out in the COMVIK (T0641/00) decision by the EPO’s Boards of Appeal. Another important point concerns obviousness: Examiners have to evaluate the causal relationship between the closest prior art and the solution claimed by the applicant. In other words, examiners need to consider whether a person skilled in the art, prompted by the objective technical problem, and aware of the closest prior art would indeed have arrived at the claimed invention.

AM:    I agree with Jean-Marc. It does not matter if the invention relates to blockchain or AI, or if the field of application is the field of self-driving vehicles, medical technology or telecommunications. The bottom line is that the invention is about an algorithm, a mathematical algorithm that runs on a device and therefore falls under CII. As you can imagine, CII inventions used to be more common in the ICT sector, and over the past years, we have developed a lot of experience in how to deal with them. However, because of the advent of new players and the shift to other technical fields, other sectors at the EPO are receiving CII applications as well. Medical technology for example is being examined at the EPO’s Healthcare, Biotechnology and Chemistry sector, and is dealing with growing numbers of AI applications. AI applications related to self-driving cars – another active field – are being handled by our Mobility and Mechatronics sector. We put a lot of effort into harmonization and cooperation between the sectors to ensure that CII applications are treated in the same way, irrespective of the technical area they belong to.

The Importance of Training and Collaboration

JMD:  This also includes internal training, which is quite an important component in the context of fast-evolving technologies such as AI or blockchain. This is essential since applications in AI that initially focused on functional developments are now extending to a variety of applicative fields. To illustrate this point in the context of deep learning, inventive efforts may initially be concentrated, for example, on the specific implementation of convolutional neural network adapted to dedicated hardware. But once this new functional platform becomes established, it can be applied to many different, technical fields, in a more, generic manner. So, this deep learning engine will be implemented in a variety of contexts, for instance an industrial process, a self-driving vehicle, but also a medical device, a washing machine or other household appliances. Because the AI engine is applied to different technical fields, the patent application may be treated by an examiner specialized in this applicative domain but who is not necessarily trained on AI. In this sense, continuous learning helps bringing all examiners on board, so that the examination of mixed-type inventions that include algorithmic features follows a systematic and predictable process.

AV:     One of the first applications of AI was optical character recognition, so image interpretation, and that quickly evolved into medical applications, and those techniques were then fed back yet into other applications. When we have medical applications, we work with our colleagues in the medical field. An example of this is for AI in a pacemaker.

JMD:  At the EPO, we can bring together in an Examining Division an interdisciplinary team that includes specialists in AI as well as experts of other fields. We can associate in the same mixed division a specialist in biology or chemistry, for example, with an expert in the field of machine learning. Such interdisciplinary divisions bring greater consistency to the examination of inventions that cover multiple technical domains. This approach is of particular importance in the context of AI inventions that are by now found across all fields of technology.

Mistakes to Avoid

GQ:    One question I really want to see if we can get some insight into is dealing with mistakes to avoid. Obviously, without getting into anything confidential or outing anybody, are there any general things that you would tell applicants about what they may think works but that you just don’t find useful or helpful, or not as helpful as they may think it is?

AV:     I would cite here independent claims where various features set out in the description are claimed in combination in order to find new combinations that were not set out in the prior art before. We don’t find such a mish- and- mash really helpful because you are likely to end up missing essential features, which means that the claim may have some elements that you didn’t find in the prior art, but it is also not complete. It is neither properly defined nor is it disclosed, and if you word it like that in the description as well, there is no way you can combine these half-finished claims. It is more useful to just describe the elements as they are and how they work together with the other elements, and how they can be exchanged or replaced or omitted. This will allow you to formulate any claim that you want later on.

GQ:    One of the things I’ve heard applicants are doing here in the U.S., which I think is smart and I’ve been preaching a version of for a long time, is get away from these egregiously broad first claims and give the examiner something that is reasonable and that will really put the case on a good footing. I think if you ask for the moon, the stars and the universe with claim one, which just has to make the examiner think from the get-go, well, this person is just going to be unreasonable.

AV:     Indeed. the examiner may then get the wrong idea about where you are going, and you may end up with a search report that is not really helpful, that cites, for instance, just a computer, which doesn’t tell you anything about the patentability of what you really invented.

JMD. If the invention is initially claimed only in broad terms, the examiner may strive to narrow it in view of the available prior art, following his or her understanding of the invention. If the subject-matter is claimed too broadly at the beginning, the examiner may have a hard time identifying what and where to search. This may lead to lengthy iterative processes between the applicant and Office to arrive at a common understanding. Clarity is also important from the onset in that a lack of clear definition of the invention may narrow the extent of strategies that the applicant may resort to eventually.

AM:  As we discussed initially, it is very important for applicants to find the right balance between the aspects of the invention they would like to disclose in the document and which ones they prefer to keep secret to gain a competitive advantage. However, they need to bear in mind that an objection of lack of disclosure may become relevant during the examination procedure, especially when claims are amended. Aspects related to how a neural network is trained are often kept secret. I believe it is helpful when patent attorneys are aware of the mathematics behind AI/ML. They obviously do not need to be experts, but this could enable them to better advise their clients and effectively interact with examiners.

JMD:  Of course, we are ready to correct typographic errors or any mistake that would be obvious to the skilled person. However, it is only if these amendments are obvious that we may allow them, so it is of the utmost importance to be very careful in drafting the application, especially when drafting mathematical equations or pseudo-code that relates to the essential features of the invention.

GQ:    That’s why having the flow charts is important, the schematics, the pseudo-code and actual code if it exists, all of those things are important because then if you do make that kind of typographical error, then it becomes clear that it was a typographical error.

Comparing USPTO and EPO Guidelines

GQ:    So, I guess the last thing I wanted to talk about is the EPO guidelines. The latest revisions have come out several months ago now. We had USPTO guidelines come out at the beginning of 2019 and I’m wondering what the similarities and differences are if you’ve had an opportunity to look at the USPTO guidelines?

JMD:  The impression I get as an examiner is that certain aspects of practices of the two offices are coming closer. For example, there is a similarity between step 2A of the USPTO practice and our process of examining mixed-type inventions at the EPO. But it’s clearly still very early days. We will need more time to reach an in-depth interpretation of similarities and differences between the USPTO and the EPO guidelines.

GQ:    For years, I have encouraged people to try to hit the European standard. Because for so long we didn’t know what the standard was in the U.S. and the European standards sort of made sense—you’ve got a technical solution to a technical problem and that means, in lay terms, explain your invention really. One of the guys I went to law school with became for a period of time chief patent counsel for Microsoft, and he said he thinks that eventually America is going to realize that the European standard is the right one and if not, adopt something that is pretty close to the technical solution to a technical problem approach, so he wrote his patent applications for Europe and then filed them everywhere. I don’t know whether you guys have any comment on that, but I think that’s a best practice.

JMD:  I think that both approaches contribute to a more consistent and predictable examination practice. The effort of providing regularly updated guidelines both at the EPO and at the USPTO contributes to bringing greater clarity and predictability for applicants.

Furthering the Dialogue

GQ:    Yeah, I would agree with that.

I’d like to just wrap up by giving each of you if you want a minute to give your final thoughts. What would be the one thing that you would want practitioners to remember from this interview?

JMD:  What I would say mainly is that the evaluation of an invention is part of a dialogue between the applicant and the patent office. This process starts with a clear statement of the invention and, as Abder mentioned, at the very beginning, by clearly specifying the features in the claimed subject-matter that are essential to the definition of the invention. The process continues on our side by being open to the arguments from the applicant. This constructive dialogue can lead to an effective examination process.

AV:     I would say, if there is any doubt, then please ask us. We are here to make sure the procedure and the process is followed as laid down by 38 countries, and you can also read the Guidelines. The part about the problem-solution approach and then maybe the part about nontechnical features, that are in the Guidelines and the examples that are most applicable to you. There are parts about user interfaces, about computer programs, about AI, and we work hard to be transparent. Hopefully that will give some guidance on how to write your application.

GQ:    Abder, the final word goes to you.

AM:    Gene, thank you very much for this great opportunity. First, when it comes to AI/ML applications always make sure as applicant to explain on which part of the AI solution the invention lies. Second, ensure to have a fallback position in the description by disclosing some concrete examples. It is a continuous learning process and sharing our experience and working together on this matter with other patent offices will definitely be beneficial to all users of the patent system worldwide.

GQ:    Thank you all, this was, this was truly fascinating. I really appreciate your time.

The Author

Gene Quinn

Gene Quinn is a Patent Attorney and Editor and President & CEO ofIPWatchdog, Inc.. Gene founded IPWatchdog.com in 1999. Gene is also a principal lecturer in the PLI Patent Bar Review Course and Of Counsel to the law firm of Berenato & White, LLC. Gene’s specialty is in the area of strategic patent consulting, patent application drafting and patent prosecution. He consults with attorneys facing peculiar procedural issues at the Patent Office, advises investors and executives on patent law changes and pending litigation matters, and works with start-up businesses throughout the United States and around the world, primarily dealing with software and computer related innovations. is admitted to practice law in New Hampshire, is a Registered Patent Attorney and is also admitted to practice before the United States Court of Appeals for the Federal Circuit. CLICK HERE to send Gene a message.

Warning & Disclaimer: The pages, articles and comments on IPWatchdog.com do not constitute legal advice, nor do they create any attorney-client relationship. The articles published express the personal opinion and views of the author and should not be attributed to the author’s employer, clients or the sponsors of IPWatchdog.com. Read more.

Discuss this

There are currently 4 Comments comments. Join the discussion.

  1. Night Writer July 26, 2019 8:50 am

    One thing that is weird is that the “computer-implemented” invention at the EPO has to be claimed and presented in terms of solving a technical problem that has to do with resources. But once you are in at the EPO as in yes that is a technical problem then all is well for all applications that recite that technical problem. Not so at the PTO. And I’ve had the EPO recognize applications as solutions to recognized technical problem whereas the PTO rejected the application under 101.

  2. Anon July 26, 2019 12:47 pm

    Night Writer,

    Perhaps that is due to the notion (US) of “merely using” a computer — no matter if that use really is more than “merely using” — falls to the overarching failure to recognize that software is merely a design choice of “wares” for computing arts.

    In other words, the US has become “pre-conditioned” to view the (mythical) general computer as not being a real machine at all.

    IF the US would treat the words of Congress with their full merit, then claims to a machine, or more correctly, claims to improvements to a machine — would NOT engender the automatic “NO” if that machine being improved happened to be that (again, mythical) general computer.

    It is as if somehow that “general computer” has been turned into a special ghost OF a machine, and is not a machine at all, to which then, improvements to that machine are caused to have the words of Congress disappear.

  3. blockchain course in hyderabad August 7, 2019 7:17 am

    nice information on blockchain has given thank you very much.

Post a Comment

Respectfully add to the discussion.

Name *
Email *
Website