Private transfer learning

Patent metadata

Field	Value
Publication number	US-2021125051-A1
Application number	US-201916662087-A
Country	US
Kind code	A1
Filing date	Oct 24, 2019
Priority date	Oct 24, 2019
Publication date	Apr 29, 2021
Grant date	—

Abstract

Embodiments are disclosed for a method for private transfer learning. The method includes generating a machine learning model comprising a training application programming interface (API) and an inferencing API. The method further includes encrypting the machine learning model using a predetermined encryption mechanism. The method additionally includes copying the encrypted machine learning model to a trusted execution environment. The method also includes executing the machine learning model in the trusted execution environment using the inferencing API.

First claim

What is claimed is:

1 . A computer-implemented method for private transfer learning, comprising:
generating a machine learning model comprising a training application programming interface (API) and an inferencing API; encrypting the machine learning model using a predetermined encryption mechanism; copying the encrypted machine learning model to a trusted execution environment; and executing the machine learning model in the trusted execution environment using the inferencing API.

2 . The method of claim 1 , wherein the machine learning model comprises a deep neural network (DNN) model.

3 . The method of claim 2 , wherein the DNN model is trained to perform a generic task.

4 . The method of claim 3 , wherein the training API trains the DNN model to perform a task that refines the generic task to a more specific task than the generic task.

5 . The method of claim 1 , further comprising generating a combination of a body of the machine learning model with a head of an additional machine learning model in the trusted execution environment.

6 . The method of claim 5 , further comprising performing the inferencing API for the combination.

7 . The method of claim 5 , further comprising performing the training API for the combination.

8 . A computer program product comprising program instructions stored on a computer readable storage medium, wherein the computer readable storage medium is not a transitory signal per se, the program instructions executable by a processor to cause the processor to perform a method comprising:
generating a machine learning model comprising a training application programming interface (API) and an inferencing API; encrypting the machine learning model using a predetermined encryption mechanism; copying the encrypted machine learning model to a trusted execution environment; and executing the machine learning model in the trusted execution environment using the training API.

9 . The computer program product of claim 8 , wherein the machine learning model comprises a deep neural network (DNN) model.

10 . The computer program product of claim 9 , wherein the DNN model is trained to perform a generic task.

11 . The computer program product of claim 10 , wherein the training API trains the DNN model to perform a task that refines the generic task to a more specific task than the generic task.

12 . The computer program product of claim 8 , the method further comprising generating a combination of a body of the machine learning model with a head of an additional machine learning model in the trusted execution environment.

13 . The computer program product of claim 12 , the method further comprising performing the inferencing API for the combination.

14 . The computer program product of claim 12 , the method further comprising performing the training API for the combination.

15 . A system comprising:
a computer processing circuit; and a computer-readable storage medium storing instructions, which, when executed by the computer processing circuit, are configured to cause the computer processing circuit to perform a method comprising: generating a machine learning model comprising a training application programming interface (API) and an inferencing API; encrypting the machine learning model using a predetermined encryption mechanism; copying the encrypted machine learning model to a trusted execution environment; executing the machine learning model in the trusted execution environment using the training API; and executing the machine learning model trained by the training API by using the inferencing API.

16 . The system of claim 15 , wherein the machine learning model comprises a deep neural network (DNN) model.

17 . The system of claim 16 , wherein the DNN model is trained to perform a generic task.

18 . The system of claim 17 , wherein the training API trains the DNN model to perform a task that refines the generic task to a more specific task than the generic task.

19 . The system of claim 15 , the method further comprising generating a combination of a body of the machine learning model with a head of an additional machine learning model in the trusted execution environment.

20 . The system of claim 19 , the method further comprising performing the inferencing API for the combination.

21 . The system of claim 19 , the method further comprising performing the training API for the combination.

22 . A system comprising:
a computer processing circuit; a graphical processing circuit (GPU); and a computer-readable storage medium storing instructions, which, when executed by the computer processing circuit, are configured to cause the computer processing circuit to perform a method comprising: generating a machine learning model comprising a training application programming interface (API) and an inferencing API; encrypting the machine learning model using a predetermined encryption mechanism; copying the encrypted machine learning model to a trusted execution environment; executing the machine learning model in the trusted execution environment by executing the training API on the GPU; and executing the machine learning model trained by the training API by executing the inferencing API on the GPU.

23 . The system of claim 22 , wherein:
the machine learning model comprises a deep neural network (DNN) model; the DNN model is trained to perform a generic task; and the training API trains the DNN model to perform a task that refines the generic task to a more specific task than the generic task.

24 . A computer-implemented method for private transfer learning, comprising:
generating a machine learning model comprising a training application programming interface (API) and an inferencing API; encrypting the machine learning model using a predetermined encryption mechanism; copying the encrypted machine learning model to a trusted execution environment; and executing the machine learning model in the trusted execution environment using the training API.

25 . The method of claim 24 , wherein:
the machine learning model comprises a deep neural network (DNN) model; the DNN model is trained to perform a generic task; and the training API trains the DNN model to perform a task that refines the generic task to a more specific task than the generic task.

Assignees

• IBM

Inventors

• Linton Jeb R
• Behnken John
• Melchionne John
• Amisano Michael
• Wright David K

Classifications

• G06N3/047

  Probabilistic or stochastic networks · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/0985

  Hyperparameter optimisation; Meta-learning; Learning-to-learn · CPC title

• G06N3/0475

  Generative networks · CPC title

• G06N3/09

  Supervised learning · CPC title