The UK Court of Appeal has issued its judgement on the UK IPO’s appeal against Emotional Perception AI limited.
Background
The invention in question related to media file recommendation system. The system used an artificial neural network (ANN) to identify and suggest relevant media files. This was achieved by ‘training’ the ANN how to distinguish semantic similarity from the physical properties of a media file. This allows the ANN to make media file recommendations, which are subsequently sent to a user.
During the initial hearing before the UK IPO, the Hearing Officer determined that the ANN was merely a computer program. The sending of a better recommendation file (e.g. a song which the user was likely to enjoy) was deemed to be a beneficial effect of a subjective and cognitive nature.
This therefore did not demonstrate any technical effect beyond the running of the computer program. Consequently, the invention was deemed to fall under excluded subject matter and thus not patentable.
Upon appeal by Emotional Perception AI, this decision was overturned by the UK High Court, who found the excluded subject matter exclusion did not apply on the basis that, even though what made the file recommendation better was not technical criteria (because the semantic similarity was a subjective matter), the ANN had reached that result by “going about its analysis and selection in a technical way”.
The UK IPO comptroller appealed the High Court decision on four grounds:
- the Judge erred in holding that the exclusion from patent protection for “a program for a computer … as such” was not engaged;
- the Judge was wrong to rely on the Appellant’s ‘concession’ that a hardware ANN was a computer but it was a computer with no program, or words to that effect;
- the Judge was wrong to exclude the consideration of the mathematical model exclusion; and
- the Judge was wrong to hold that the claimed invention involves a substantive technical contribution.
Judgement
The UK Court of Appeal upheld the UKIPO comptroller’s initial decision against Emotional Perception AI (EPL) that the ANN is excluded from patentability on the basis that it was a computer program.
In reaching this decision, the court addressed Grounds 1 and 2 together by assessing the meaning of a “program for a computer”, and whether there is a computer program in an ANN. The definition of a computer program adopted by the court was “a set of instructions for a computer to do something”. The court dismissed EPL’s prior arguments (on why an ANN was not a computer program) on the basis they sought to add various limitations into the definition of a computer program (such as the involvement of a human computer programmer & drawing a distinction between instructions created by a computer and those by a human). Under the court’s broad definition, an ANN (implemented exclusively as software or hardware) was a computer program.
Turning to Ground 4, it was considered whether the claimed invention had a technical contribution that would take it outside of excluded subject matter. After some discussion of whether the technical contribution might lie in the training process of the ANN rather than the output it provides, the contribution was determined to be the sending of an improved recommendation message to a user. The court agreed with the Hearing Officer’s view that this contribution was not sufficient to take the claimed invention out of the exclusion as a “technical purpose” beyond merely sending a message was necessary.
The court decided there was no need to assess the mathematical method exclusion (Ground 3), but it may well have been necessary if the ANN was held to not be a computer program.
Refusal of the application was thus upheld on the basis the claimed invention was a computer program, and thus excluded from patentability.
Summary
This decision aligns with EPO case law in Yahoo T0306/10 (which held determining whether song recommendations were “good” or “bad” did not amount to a technical effect). This may suggest the UK Courts are more closely following the EPO’s approach to the assessment of computer programs (which is detailed in G1/19).