Method and system for personalising machine learning models
AssigneeSamsung Electronics Co., Ltd.
InventorsAlberto Gil C. P. Ramos, Abhinav Mehrotra, Sourav Bhattacharya
Life of this patent
Estimating the term…
Abstract
Broadly speaking, embodiments of the present techniques provide a method and system for personalising machine learning models on resource-constrained devices by using conditional neural networks. In particular, the present techniques allow for resource-efficient use of a conditioning vector by incorporating the conditioning vector into weights learned during training. This reduces the computational resources required at inference time.
Claims13
A computer-implemented method for training, on a server, a machine learning, ML, model having a first neural network and a second neural network, the method comprising:
- obtaining a training dataset comprising a plurality of pairs of data items to be processed by the ML model, each pair of data items comprising a conditioning vector computed from data associated with an individual user and an input data item comprising a plurality of samples, wherein a sample from among the plurality of samples is a text sample or an audio sample;
- and
- training the ML model by jointly:
- training the first neural network of the ML model to alter weights of at least one layer of the second neural network to personalize the weights for individual users using the conditioning vector of the pairs of data items;
- and
- training the second neural network of the ML model to process the input data item of the pairs of data items using the personalized weights for the at least one layer, and to generate output data using the personalized weights.
A server for training a machine learning, ML, model having a first neural network and a second neural network, the server comprising:
- at least one processor coupled to memory and arranged to:
- obtain a training dataset comprising a plurality of pairs of data items to be processed by the ML model, each pair of data items comprising a conditioning vector computed from data associated with an individual user and an input data item comprising a plurality of samples, wherein a sample from among the plurality of samples is a text sample or an audio sample;
- and
- train the ML model by jointly:
- training the first neural network of the ML model to alter weights of at least one layer of the second neural network to personalize the weights for individual users using the conditioning vector of the pairs of data items;
- and
- training the second neural network of the ML model to process the input data item of the pairs of data items using the personalized weights for the at least one layer, and to generate output data using the personalized weights.
A computer-implemented method for training, on a server, a machine learning, ML, model having a first neural network and a second neural network, the method comprising:
- obtaining a training dataset comprising a plurality of pairs of data items to be processed by the ML model, each pair of data items comprising a conditioning vector that represents an individual user and an input data item;
- and
- training the ML model by jointly:
- training the first neural network of the ML model to transform weights of at least one layer of the second neural network into personalized weights for individual users using the conditioning vector of the pairs of data items;
- and
- training the second neural network of the ML model to process the input data item of the pairs of data items using the personalized weights for the at least one layer, and to generate output data,
- wherein the computer-implemented method further comprises receiving at least one constraint, and wherein training the ML model further comprises using the received at least one constraint to train the first and second neural networks to generate output data, and
- wherein using the received at least one constraint to train the ML model to generate output data comprises restricting values of the personalized weights to be within a predefined range.
Dependencies are inferred from the claim text.