Résumé de EPC2000 Art 056 pour la décision T0702/20 du 07.11.2022

In T 702/20 the board held that a neural network defines a class of mathematical functions which, as such, was excluded matter. As for other "non-technical" matter, it could therefore only be considered for the assessment of inventive step when used to solve a technical problem, e.g. when trained with specific data for a specific technical task. According to the board, the claim as a whole specified abstract computer-implemented mathematical operations on unspecified data, namely that of defining a class of approximating functions (the network with its structure), solving a (complex) system of (non-linear) equations to obtain the parameters of the functions (the learning of the weights). According to the claim, the neural network had a new structure because the hierarchical neural network was formed by loose couplings between the nodes in accordance with a sparse parity-check matrix of a low-density parity-check code. The appellant argued that the proposed modification in the neural network structure, in comparison with standard fully-connected networks, would reduce the amount of resources required, in particular storage, and that this should be recognized as a technical effect, following G 1/19. The board noted that, while the storage and computational requirements were indeed reduced in comparison with the fully-connected network, this did not in and by itself translate to a technical effect, for the simple reason that the modified network was different and would not learn in the same way. So it required less storage, but it did not do the same thing. For instance, a one-neuron neural network required the least storage, but it would not be able to learn any complex data relationship. The proposed comparison was therefore deemed incomplete, as it only focused on the computational requirements, and insufficient to establish a technical effect. The claimed invention thus lacked inventive step. As a further remark, the board stressed that there could be no reasonable doubt that neural networks can provide technical tools useful for automating human tasks or solving technical problems. In most cases, however, this required them to be sufficiently specified, in particular as regards the training data and the technical task addressed. In this particular case, the board could not see, considering the content of the application, for which type of learning tasks the proposed structure may be of benefit, and to what extent.