Mary-Anne Hartley (Annie)

As part of her semester project report on the DISCO frontend, @gozgun suggested a list of future work. Copied here for future reference. [GunesOzgun_DISCOFrontend_Sem2022a.pdf](https://github.com/epfml/disco/files/9128404/GunesOzgun_DISCOFrontend_Sem2022a.pdf) - [x] Offering dark mode as...

Clearly not a priority... - Favicon can't really be seen in dark mode  -Redo favicon in contrasting colours (or lighter blue) - `Disco` should...

Related to #100 and #377 [iFedAvg ](https://arxiv.org/abs/2107.06580) detects feature-wise biases in tabular data relative to each user. **Suggestion: use iFedAvg to test for these biases (in a "fake" training round)...

Some users may be apprehensive to start using DISCO without first seeing how/if it works. **A TUTORIAL IS NEEDED** A pre-recorded tutorial video on an example task is a good...

ERROR - COVID-LUS task When attempting to train (either locally or decentrally) using the data provided from DISCO-based link I get the following error 

**AIM** To extend the availability of our (soon to be published) DeepBreath 💨 model to DISCO **CONTEXT** - DeepBreath is a deep learning model that predicts the presence of childhood...

### PROBLEM Currently in DISCO: data owners form **task-specific DISCOllaboratives**  i.e. they join an **existing task**, but cannot just make their data available for...

### PROBLEM Currently in DISCO, tasks comprise a single model with an expected input of a restricted size. - We previously proposed way to modularize training into a series of...

### PROBLEM: Distributed learning has a higher carbon cost than centralized learning (https://arxiv.org/abs/2102.07627)  The problem is worsened in non-IID settings ![Screenshot 2022-10-25 at 21...