Back to results
US 2019/0295073 A1Published application
Secure data processing transactions
AssigneeVia Science, Inc.
InventorsJesús Alejandro Cárdenes Cabré, Jeremy Taylor, Colin Gounden
Life of this patent
Estimating the term…
Abstract
A request to train a neural-network model is received, as are a first offer to supply encrypted training data and a second offer to train the model. After determining that the offers satisfy the request, a model provider told to train the model. If they model performs in accordance with a quality metric, it is sent to a model user.
Claims20
A computer-implemented method comprising:
- receiving, from a device associated with a model user, a first request to create a neural-network model;
- receiving, from a device associated with a data source, a first offer to provide first encrypted training data;
- receiving, from a device associated with a model provider, a second offer to train the neural-network model;
- determining that the first offer and the second offer satisfy the first request;
- sending, to the device associated with the model provider, a second request to create the neural-network model using the first encrypted training data;
- receiving, from the device associated with the model provider, the neural-network model;
- determining that the neural-network model satisfies a quality metric;
- and
- based at least in part on determining that the neural-network model satisfies the quality metric, sending, to the device associated with the model user, the neural-network model.
6 more elements
A system comprising:
- at least one processor;
- and
- at least one memory including instructions that, when executed by the at least one processor, cause the system to:
- receive, from a device associated with a model user, a first request to create a neural-network model;
- receive, from a device associated with a data source, a first offer to provide first encrypted training data;
- receive, from a device associated with a model provider, a second offer to train the neural-network model;
- determine that the first offer and the second offer satisfy the first request;
- send, to the device associated with the model provider, a second request to create the neural-network model using the first encrypted training data;
- receive, from the device associated with the model provider, the neural-network model;
- determine that the neural-network model satisfies a quality metric;
- and
- based at least in part on determining that the neural-network model satisfies the quality metric, send, to the device associated with the model user, the neural-network model.
9 more elements
A computer-implemented method comprising:
- receiving, from a device associated with a model user, a first request to create a neural-network model;
- receiving, from a device associated with a model provider, a first offer to train the neural-network model;
- determining that the first offer satisfies the first request;
- determining an amount of computing resources associated with creating the neural-network model;
- sending, to the model user, an indication of a payment associated with the amount of computing resources;
- sending, to the device associated with the model provider, a second request to create the neural-network model using first encrypted training data;
- receiving, from the device associated with the model provider, the neural-network model;
- determining that the neural-network model satisfies a quality metric;
- and
- based at least in part on determining that the neural-network model satisfies the quality metric, sending, to the device associated with the model user, the neural-network model.
7 more elements
Dependencies are inferred from the claim text.