Skip to content

Latest commit

 

History

History
131 lines (83 loc) · 5.15 KB

RELEASE.md

File metadata and controls

131 lines (83 loc) · 5.15 KB
Raw

Release Notes

Introduction

This document will maintain and continually update the release notes of each version of Rosetta. If you have questions or comments, please contact us via rosetta@latticex.foundation.

Rosetta v1.0.0

  1. Support multi-tasking concurrency, that mean different privacy protocols can be executed simultaneously.

  2. Nodes can be configured with roles, supporting calculation node, data node and result node, and data nodes can be configured in any number according to requirements.

  3. Refactoring code to make the code highly cohesive and low coupled.

  4. Added SecureExp, SecureSqrt, SecureRsqrt, SecureReciprocaldiv secure operations.

  5. Adding audit logs, which are turned off by default and can be turned on according to business scenarios.

  6. Efficient third-party io modules can be integrated into Rosetta by implementing specific interfaces.

  7. Optimize back-end unit tests to provide more friendly test cases.

  8. Some known bugs are fixed.

Rosetta v0.3.0

  1. Added PrivateTextLineDataset, PrivateInput secure operations.

  2. Added SecureLogicalAnd, SecureLogicalOr, SecureLogicalXor, SecureLogicalNot secure operations.

  3. Speedup some backend operations.

  4. Uses related python classes such as PrivateTextLineDataset and iterators to load large data sets, thereby reducing memory usage.

  5. Some known bugs are fixed.

Rosetta v0.2.1

  1. Support 128-bit integer data type, which enables big integer and high precision float-point operations.

  2. Refactor python directories of dynamic libraries and use ROSETTA_MPC_128=ON to enable 128-bit privacy computation.

Rosetta v0.2.0

  1. Refactor the whole architecture so that each layer has a clearer boundary. Especially, an abstract protocol layer is added to decouple cryptographic protocols and TensorFlow Ops, so that TensorFlow does not depend on specific protocols and vise verse.

  2. Support custom data types by mainly utilizing tf.string. This will enable us to have arbitrary length and format of internal ciphertext.

  3. Some APIs, such as activate, are added and refined to support better control on selection and configuration of backend protocol. In this version, you can configure IP ports, host addresses, save mode flag and protocol precision as JSON string by calling activate in your program.

  4. Some known bugs are fixed.

Rosetta v0.1.1

Features

  1. Binary installation of TensorFlow is supported.

Rosetta v0.1.0

Features

  1. Secure Multi-Party Computation (MPC) is supported, the underlying protocol is SecureNN. This is a $3$-party protocol, which is secure in the semi-honest model with honest majority.

  2. Developers could transfer traditional TensorFlow codes into a privacy-preserving manner with minimal changes (import latticex.rosetta).

  3. Static Pass is supported to replace TensorFlow native operations with MPC operations before the execution of the graph.

  4. Dynamic Pass is supported to replace TensorFlow native operations with MPC operations when the graph is executed.

  5. The following MPC operations and related gradients are supported.

    MPC OPs MPC Gradient OPs
    MpcAdd MpcAddGrad
    MpcSub MpcSubGrad
    MpcMul MpcMulGrad
    MpcDiv MpcDivGrad
    MpcTrueDiv MpcTrueDivGrad
    MpcRealDiv MpcRealDivGrad
    MpcMatMul MpcMatMulGrad
    MpcSigmoid MpcSigmoidGrad
    MpcLog MpcLogGrad
    MpcLog1p MpcLog1pGrad
    MpcPow MpcPowGrad
    MpcMax MpcMaxGrad
    MpcMean MpcMeanGrad
    MpcRelu MpcReluGrad
    MpcEqual -
    MpcLess -
    MpcGreater -
    MpcLessEqual -
    MpcGreaterEqual -
    MpcSaveV2 -
    MpcApplyGradientDescentOp -

  6. Support the combination of supported MPC operations to implement arbitrary models, such as: linear regression model, logistic regression model, etc.

  7. Support for specifying the type (plaintext or ciphertext) of model to save;

  8. Support for specifying where to save the model (Party0\Party1\Party2\all Parties);

Supported Platforms

Rosetta has only been extensively tested on Intel X64 machines running Ubuntu 18.04.

Known Problems

This section contains all known problems with the Rosetta system, listed by component. As new problems are discovered, they will be added to these sections.

Additional Information

More details of Rosetta could be found in the documentation fold. If you have any questions or comments about Rosetta, please feel free to contact us via rosetta@latticex.foundation.