Training machine learning models with training data
AssigneesVolkswagen Aktiengesellschaft, Porsche AG, Audi AG
InventorsLiang Kong, Oleg Tolstov
Life of this patent
Estimating the term…
Abstract
In one embodiment, a method is provided. The method includes storing a set of training data. One or more machine learning models are trained based on the set of training data. The method also includes receiving, from a computing device, a request to access the set of training data. The method further includes determining whether the computing device is allowed to access the set of training data. In response to determining that the computing device is allowed to access the set of training data, the method includes transmitting a training token to the computing device. The training token grants a training environment with access to the set of training data for a period of time.
Claims20
A method, comprising:
- storing a set of training data, wherein one or more machine learning models are trained based on the set of training data;
- receiving, from a computing device, a request to access the set of training data;
- determining whether the computing device is allowed to access the set of training data;
- and
- in response to determining that the computing device is allowed to access the set of training data, then:
- transmitting a training token to the computing device, the training token being valid for a period of time;
- the computing device transmitting the training token to a training environment comprising an isolated environment for training the machine learning model;
- the training token granting the training environment with access to the set of training data for the period of time for which the training token is valid;
- and
- the training environment deleting the set of training data after the training environment determines that the period of time has expired in order to maintain confidentiality of the training data.
A method, comprising:
- receiving a training package from a computing device, wherein the training package comprises a first machine learning model that is to be trained based on a set of training data stored in a storage system;
- retrieving the set of training data from the storage system based on a training token provided to the computing device by the storage system;
- training the first machine learning model within a training environment based on the set of training data to generate a trained machine learning model, wherein the set of training data is inaccessible to the first machine learning model outside of the training environment, the training environment comprising an isolated environment for training the machine learning model;
- deleting the set of training data within the training environment after training the first machine learning model in order to maintain confidentiality of the training data;
- and
- transmitting the trained machine learning model to the computing device.
An apparatus, comprising:
- a memory configured to store data;
- and
- a processing device operatively coupled to the memory, the processing device configured to:
- receive a training package from a computing device, wherein the training package comprises a first machine learning model that is to be trained based on a set of training data stored in a storage system;
- receive the set of training data from the storage system based on a training token provided to the computing device by the storage system;
- train the first machine learning model within a training environment based on the set of training data to generate a trained machine learning model, wherein the set of training data is inaccessible to the first machine learning model outside of the training environment, the training environment comprising an isolated environment for training the machine learning model;
- delete the set of training data within the training environment after training the first machine learning model to maintain confidentiality of the training data;
- and
- transmit the trained machine learning model to the computing device.
Dependencies are inferred from the claim text.