Bogdan Cebere

FR: Implement min/max for encrypted values

1

## Feature Description For implementing MaxPooling, we need to have a way to do min/max operations in encrypted space. Some potential solutionsa/algorithms: - https://eprint.iacr.org/2019/417.pdf - https://eprint.iacr.org/2019/1234.pdf In order to be...

- Improve the documentation for the project. - Generate sphinx docs. - Upload the docs to readthedocs.

CKKSVector: use multithreading for sum_inplace

2

## Description The sum_inplace operation of the CKKSVector can be done using multithreading. More precisely, we can parallelize [this loop](https://github.com/OpenMined/TenSEAL/blob/master/tenseal/cpp/tensors/ckksvector.cpp#L247). You can see an example of how to use the...

FR: Investigate Microsoft EVA support

4

## Feature Description We need to investigate the support for HE compilers. Microsoft EVA is a compiler for SEAL. https://github.com/microsoft/EVA Paper: https://arxiv.org/pdf/1912.11951.pdf Update this task with any findings related to...

FR: BatchNorm support for training

1

## Feature Description BatchNorm support in encrypted space for training. For BatchNorm, we need support for computing: - the mean - this implies a summation and a multiplication with plain...

## Feature Description Implement MaxPooling operation. Depends on https://github.com/OpenMined/TenSEAL/issues/277. In order to be approved, the PR must include Python/C++ tests for the MaxPooling operation.

## Feature Description For computing the standard deviation in encrypted space, we need to be able to compute the square root of a tensor. One iterative algorithm for square root...

## Description Some TenSEAL bindings are not used anymore in the Python API, and we should remove them. You can find the current Python API in: - https://github.com/OpenMined/TenSEAL/tree/master/tenseal/tensors - https://github.com/OpenMined/TenSEAL/blob/master/tenseal/enc_context.py...

## Description Improve logging: - make the logging configurable in the library. We can use dedicated bindings for the logging logic, with settings like output stream/log level/etc. - make the...

## Description Implement the sub_plain_inplace operation for BFVVector [here](https://github.com/OpenMined/TenSEAL/blob/master/tenseal/cpp/tensors/bfvvector.cpp#L268). You can see a similar implementation in the CKKSVector class [here](https://github.com/OpenMined/TenSEAL/blob/master/tenseal/cpp/tensors/ckksvector.cpp#L290). In order to be accepted, the PR must include: -...