US 11,861,500 B2
Meta-learning system
Martin Kraus, Nuremberg (DE)
Assigned to SIEMENS HEALTHCARE GMBH, Erlangen (DE)
Filed by Siemens Healthcare GmbH, Erlangen (DE)
Filed on Dec. 19, 2018, as Appl. No. 16/225,034.
Claims priority of application No. 17210056 (EP), filed on Dec. 22, 2017.
Prior Publication US 2019/0197360 A1, Jun. 27, 2019
Int. Cl. G06N 3/084 (2023.01); G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06N 3/04 (2023.01); G06N 3/08 (2023.01); G06N 7/00 (2023.01)
CPC G06N 3/084 (2013.01) [G06F 18/214 (2023.01); G06F 18/217 (2023.01); G06N 3/04 (2013.01); G06N 3/08 (2013.01); G06N 7/00 (2013.01)] 28 Claims
OG exemplary drawing
 
1. A meta-learning system, comprising:
processing circuitry configured to,
compute output data from applied input data according to an inner model function, the inner model function depending on model parameters,
compute an error indicating a mismatch between the output data and a target value, and
update the model parameters of the inner model function according to an updated state, the updated state being based on a current state and the error, first training is performed to adjust the model parameters before second training the inner model function, the second training of the inner model function being to minimize an overall sum of a learning strength over time, the second training including minimizing the following function:
Loss=Σi(ei+αdi)
i representing a respective time, ei representing the error at the respective time, di representing the learning strength at the respective time, and α representing a weighting factor.