Tom
The link to the Alan Turing institute's research group on privacy preserving data analysis was stale.
## What? Remove the syft dataset/dataloader dependency from the VerticalDataset/loaders. Inherit instead from torch dataset/loader ## Why? Those classes are not present in syft `>= 0.3.0`
## What? Train a single-headed neural network on MIMIC data. Look to previous VFL papers for reference. ## How long? Once MIMIC data access has been granted, ~2 weeks. ##...
## Description Use [`papermill`](https://pypi.org/project/papermill/) in CI pipeline to test that each notebook example can execute ## Type of Test - **Integration test** (e.g. checking if a certain group or set...
## Feature Description - Create functions which split PyTorch datasets into separate datasets - Should work for image and non-image datasets - Functions should apply random IDs to datapoints -...
## Feature Description A class which looks at a non-split PyTorch model an estimates flops at each layer. This user can use this class to evaluate where to split the...
## Description Coverage tests were removed in #23 as there was some issue with the bazel build. This was a temporary solution to get the PR merged. We should re-add...
## Feature Description Make PyVertical pip-installable. This should happen after the integration of pyvertical with pysyft, when we have a clearer idea of pyvertical-specific functionality ## Is your feature request...
## Description There is a hardcoded SplitNN in the source code. This should be removed. ## Are you interested in working on this improvement yourself? - Yes, I am, but...
## Feature Description Implement a dual-headed splitNN. Each head takes some data as input and computes some representation of the data. The two intermediate vectors are combined and the rest...