# US 2020/0394470 A1 · Efficient verification of maching learning applications

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0394470 A1
- Document ID: US20200394470A1
- Status: Published application
- Kind: A1
- Application number: 16/438,476
- Filed: 2019-06-12
- Published: 2020-12-17
- Earliest claimed: 2019-06-12
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Venkata Sitaramagiridharganesh Ganapavarapu; Kanthi Sarpatwar; Karthikeyan Shanmugam; Roman Vaculin
- Web page: https://patentgrep.com/patent/US20200394470A1

## CPC

- G06N 20/00: Machine learning
- G06F 18/211: Selection of the most significant subset of features
- G06F 18/2148: {characterised by the process organisation or structure, e.g. boosting cascade}
- G06F 18/2185: {the supervisor being an automated module, e.g. intelligent oracle}
- G06Q 20/08: Payment architectures
- G06Q 20/22: Payment schemes or models
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

An example operation may include one or more of generating, by a training participant client, a plurality of transaction proposals, each of the plurality of transaction proposals corresponding to a training iteration for machine learning model training related to stochastic gradient descent, the machine learning model training comprising a plurality of training iterations, the transaction proposals comprising a gradient calculation performed by the training participant client, transferring the plurality of transaction proposals to one or more endorser nodes or peers each comprising a verify gradient smart contract, executing, by each of the endorser nodes or peers, the verify gradient smart contract; and providing endorsements corresponding to the plurality of transaction proposals to the training participation client.

## Claims

### Claim 1

A system, comprising:
a training participant client, comprising a training dataset and configured to:
generate a plurality of transaction proposals that each correspond to a training iteration for machine learning model training related to stochastic gradient descent, the machine learning model training comprising a plurality of training iterations, the transaction proposals comprising a gradient calculation performed by the training participant client; and
a blockchain network, comprising:
one or more endorser nodes or peers, each comprising a verify gradient smart contract and configured to:
receive the plurality of transaction proposals;
execute the verify gradient smart contract; and
provide endorsements that correspond to the plurality of transaction proposals to the training participation client.

### Claim 2 (depends on 1)

The system of claim 1, wherein the verify gradient smart contract does not perform gradient calculation, wherein the verify gradient smart contract verifies results of gradient calculations from the transaction proposals.

### Claim 3 (depends on 2)

The system of claim 2, wherein the transaction proposals specify a batch from the training dataset, a loss function, current parameter weights, and a gradient calculation.

### Claim 4 (depends on 3)

The system of claim 3, wherein the batch comprises a subset of samples from the training dataset for a training iteration.

### Claim 5 (depends on 2)

The system of claim 2, wherein data and model parameters that correspond to the training dataset are stored outside the blockchain network, wherein verification-related data is stored in a blockchain for the blockchain network.

### Claim 6 (depends on 2)

The system of claim 2, wherein in response to the training participation client receives the endorsements, the training participation client includes the endorsements in a transaction and submits the transaction to the blockchain network.

### Claim 7 (depends on 2)

The system of claim 2, wherein the one or more endorser nodes or peers not comprising the verify gradient smart contract, wherein the one or more endorser nodes forward the transaction proposals to an independent auditor outside the blockchain network, wherein the independent auditor comprises the verify gradient smart contract and is configured to:
receive the plurality of transaction proposals;
execute the verify gradient smart contract; and
provide verification results that correspond to the plurality of transaction proposals to the endorser nodes or peers.

### Claim 8

A method, comprising:
generating, by a training participant client, a plurality of transaction proposals, each of the plurality of transaction proposals corresponding to a training iteration for machine learning model training related to stochastic gradient descent, the machine learning model training comprising a plurality of training iterations, the transaction proposals comprising a gradient calculation performed by the training participant client;
transferring the plurality of transaction proposals to one or more endorser nodes or peers each comprising a verify gradient smart contract;
executing, by each of the endorser nodes or peers, the verify gradient smart contract; and
providing endorsements corresponding to the plurality of transaction proposals to the training participation client.

### Claim 9 (depends on 8)

The method of claim 8, wherein the verify gradient smart contract does not perform gradient calculation, wherein the verify gradient smart contract verifies results of gradient calculations from the transaction proposals.

### Claim 10 (depends on 9)

The method of claim 9, wherein the training participant client comprising a training dataset, wherein each of the transaction proposals specify a batch from the training dataset, a loss function, current parameter weights, and a gradient calculation.

### Claim 11 (depends on 10)

The method of claim 10, wherein the batch comprises a subset of samples from the training dataset for a training iteration.

### Claim 12 (depends on 10)

The method of claim 10, wherein data and model parameters corresponding to the training dataset are stored outside the blockchain network, wherein verification-related data is stored in a blockchain for the blockchain network.

### Claim 13 (depends on 9)

The method of claim 9, wherein in response to the training participation client receives the endorsements, the method further comprising:
including, by the training participation client, the endorsements in a transaction; and
submitting the transaction to the blockchain network.

### Claim 14 (depends on 9)

The method of claim 9, wherein the one or more endorser nodes or peers not comprising the verify gradient smart contract, wherein the method further comprising:
forwarding, by the one or more endorser nodes, the transaction proposals to an independent auditor outside the blockchain network, wherein the independent auditor comprising the verify gradient smart contract, wherein the method further comprising:
receiving, by the independent auditor, the plurality of transaction proposals;
executing the verify gradient smart contract; and
providing, by the independent auditor, verification results corresponding to the plurality of transaction proposals to the endorser nodes or peers.

### Claim 15

A non-transitory computer readable medium comprising instructions, that when read by a processor, cause the processor to perform:
generating, by a training participant client, a plurality of transaction proposals, each of the plurality of transaction proposals corresponding to a training iteration for machine learning model training related to stochastic gradient descent, the machine learning model training comprising a plurality of training iterations, the transaction proposals comprising a gradient calculation performed by the training participant client;
transferring the plurality of transaction proposals to one or more endorser nodes or peers each comprising a verify gradient smart contract;
executing, by each of the endorser nodes or peers, the verify gradient smart contract; and
providing endorsements corresponding to the plurality of transaction proposals to the training participation client.

### Claim 16 (depends on 15)

The non-transitory computer readable medium of claim 15, wherein the verify gradient smart contract does not perform gradient calculation, wherein the verify gradient smart contract verifies results of gradient calculations from the transaction proposals.

### Claim 17 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein the training participant client comprising a training dataset, wherein each of the plurality of transaction proposals specify a batch from the training dataset, a loss function, current parameter weights, and a gradient calculation, wherein the batch comprises a subset of samples from the training dataset for a training iteration.

### Claim 18 (depends on 17)

The non-transitory computer readable medium of claim 17, wherein data and model parameters corresponding to the training dataset are stored outside the blockchain network, wherein verification-related data is stored in a blockchain for the blockchain network.

### Claim 19 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein in response to the training participation client receives the endorsements, the instructions cause the processor to further perform:
including, by the training participation client, the endorsements in a transaction; and
submitting the transaction to the blockchain network.

### Claim 20 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein the one or more endorser nodes or peers not comprising the verify gradient smart contract, wherein the instructions cause the processor to further perform:
forwarding, by the one or more endorser nodes, the transaction proposals to an independent auditor outside the blockchain network, wherein the independent auditor comprising the verify gradient smart contract, wherein the instructions cause the processor to further perform:
receiving, by the independent auditor, the plurality of transaction proposals;
executing the verify gradient smart contract; and
providing, by the independent auditor, verification results corresponding to the plurality of transaction proposals to the endorser nodes or peers.
