[TBD]Handling encrypted data in machine learning

Introduction

Laymen explanation

Privacy in the Internet of Things is a fundamental challenge for the Ubiquitous healthcare systems that depend on the data aggregated and collaborative deep learning among different parties.

In general, Distributed learning (sometimes known as federated learning) allows a group of independent data owners to collaboratively learn a model over their data sets without exposing their private data.

If you love to know how it is possible to do this, then this document helps.

Technical explanation

Privacy-preserving machine learning (PPML) is a class of techniques that let machine learning models compute directly on encrypted data, returning encrypted results. Only the person who encrypted the input data can decrypt the result.

Why machine can work on encrypted data?

ML learning models work on [Verify it]

• Similarity measure. Similarity measure depends on distance metric. In other words, it is distance between data points.

  • Relationship between words/features

In case, above criteria are met, then ML model can be applied on encrypted data itself (without decrying the data).

Criteria on encryption method

Order-preserving encryption (OPE)

It allows data to be encrypted in such a way that the ciphertexts — the encrypted versions of the data — preserve the order for the plaintexts — the unencrypted versions. That is, for any plaintexts a and b, a > b if and only if the ciphertext of a is greater than that of b (Enc(a) > Enc(b)), and vice versa.

Properties

• Machine learning models compute directly on encrypted data, returning encrypted results.

• Only the person who encrypted the input data can decrypt the result.

• It adds computational overhead, which often makes the resulting models too slow to be practical.

Cryptographic tools used

Order-preserving encryption (OPE)

It is used to encrypt the values those features are being tested against.

Pseudo-random functions (PRFs)

It is used to generate pseudorandom “features names” for the values that the tree nodes are testing

Additively homomorphic encryption (AHE)

AHE allows to combine the outputs of the regression trees to obtain the final encrypted result.

privacy-preserving version of a machine learning algorithm

XGBoost

Reference

https://www.amazon.science/blog/machine-learning-models-that-act-on-encrypted-data

https://images.app.goo.gl/oNnSSY3rWwCUs68P8

https://oblivc.org/ppml/

https://youtu.be/rwyWiDyVmjE

https://www.semanticscholar.org/paper/MSCryptoNet%3A-Multi-Scheme-Privacy-Preserving-Deep-Kwabena-Qin/999be5b34a7243447ec1e1936f54b3c5e1012b01

https://youtu.be/Jy7ozgwovgg

https://ieeexplore.ieee.org/document/9039294