New EPO Guidelines on Mathematical Methods, AI and Machine Learning

08 October 2018

Part G, II-3.3

The updated Guidelines provide an extended section in relation to mathematical methods and, for the first time, a section dedicated to AI and machine learning.

Mathematical methods

As with other computer implemented inventions, the exclusion in relation to mathematical methods is relatively easy to overcome: simply including a technical means such as a computer within the claims will mean that the invention is not excluded from patentability at the EPO per se.

However, examination will then continue with the assessment of inventive step including only those features of the claim which contribute to the technical character of the invention. The new Guidelines lay out, in more detail than ever before, what may contribute to technical character.

The new Guidelines set out that a mathematical method contributes to the technical character if it has a technical purpose. For example, this may be in controlling a machine, determining from measurements how a machine should be operated, analysing or enhancing audio, image or video data, improving data transmission or storage (for example with error correction or compression), encryption techniques, processing load distribution, providing a diagnosis based on physiological measurements or simulating the behaviour of technical items.

The EPO requires that the technical purpose must be specific, not generic, and the claims must be limited- expressly or implicitly- to that technical purpose. Generally, the updated Guidelines suggest that the technical purpose of the mathematical method is to be primarily based on its output rather than its input.

In addition, a mathematical method may contribute to the technical character of the invention if it has a technical implementation, for example being motivated by technical considerations of the internal functioning of a computer.

Artificial Intelligence and Machine Learning

The updated Guidelines helpfully provide examples of situation in which the use of a neural network may be found to be technical. For example, the use of the neural network in heart monitoring apparatus for the purpose of identifying irregular heartbeats, and the use of a neural network in classification of digital images based on low-level features are considered technical applications. By way of contrast, classifying text documents using a neural network but based on their textual content is a linguistic rather than a technical purpose.

It is also set out that, where a classification method serves a technical purpose, the steps of generating a training set and training a classifier may also contribute to the technical character of the invention, if they support achieving a technical purpose.

Simulation, Design or Modelling

Computer implemented simulation of some technical items or specific technical purposes may be considered to contribute to the technical contribution of an invention.

For example, a numerical simulation of the performance of electrical circuits subject to 1/f noise or a specific industrial chemical process may be considered to have technical purpose. In another example, a computer implemented method of designing an optical system using a formula for determining technical parameters such as refractive indices and magnification factors for given input conditions makes a technical contribution. An iterative computer simulation to determine a maximum value for an operating parameter of a nuclear reactor also makes a technical contribution.

However, where the computer-aided determination depends on decisions to be taken by a human user and the technical considerations for making such a decision are missing from the claim, this may not result in a technical effect. Moreover, if the method results in an abstract model of a product system or process, this per se may not be considered to be technical effect even if the modelled product is technical.

Practice points

The updated Guidelines do not suggest any departure from existing EPO practice. However, they do helpfully provide a range of examples which will allow patentees to make a better assessment of how likely an invention is to be patentable.

In addition, they provide a useful guidance as to the features of an invention which should be highlighted in an application. In particular, at least one specific technical purpose or technical implementation should be brought to the forefront when drafting patent applications in this area.

Caroline Day

Caroline Day


Haseltine Lake - Cookie Disclaimer

Our website uses cookies to provide you with a better online experience. If you continue to use our site without changing your browser settings, we'll assume you are happy to receive cookies. Please read our Cookie policy for more information.

Do not show this message again