Federated learning for anomaly detection
AssigneeNEC Laboratories America, Inc.
InventorsDongjin Song, Yuncong Chen, Cristian Lumezanu, Takehiko Mizoguchi, Haifeng Chen, Wei Zhu
Life of this patent
Estimating the term…
Abstract
Methods and systems for training a neural network include collecting model exemplar information from edge devices, each model exemplar having been trained using information local to the respective edge devices. The collected model exemplar information is aggregated together using federated averaging. Global model exemplars are trained using federated constrained clustering. The trained global exemplars are transmitted to respective edge devices.
Claims20
A method for training a neural network model, comprising:
- collecting model exemplar information from a plurality of edge devices, each model exemplar having been trained using information local to the respective edge devices;
- aggregating the collected model exemplar information together using federated averaging;
- training global model exemplars using federated constrained clustering using a processor;
- and
- transmitting the trained global exemplars to respective edge devices of the plurality of edge devices.
A system for training a neural network model, comprising:
- a hardware processor;
- a network interface in communication with a plurality of edge devices, configured to collect model exemplar information from the plurality of edge devices, each model exemplar having been trained using information local to the respective edge devices, and to transmit trained global exemplars to respective edge devices of the plurality of edge devices;
- and
- a memory that stores a computer program product, which, when executed by the hardware processor, causes the hardware processor to:
- aggregate the collected model exemplar information together using federated averaging;
- and
- train the global model exemplars using federated constrained clustering using a processor.
Dependencies are inferred from the claim text.