readme: rm papers

README.md

@@ -34,21 +34,6 @@ ___
 
 ___
 
-:test_tube: **RESEARCH-BASED DESIGN** 
-
-DISCO leverages latest research advances, enabling open-access and easy-use distributed training which is
-- :tornado: efficient ([R1](https://github.com/epfml/powergossip), [R2](https://github.com/epfml/ChocoSGD)) 
-- :lock: privacy-preserving ([R3](https://eprint.iacr.org/2017/281.pdf), [R4](https://arxiv.org/abs/2006.04747))
-- :hammer_and_wrench: fault-tolerant and dynamic over time ([R5](https://arxiv.org/abs/2106.06639), [R6](https://arxiv.org/abs/2206.08307))
-- :ninja: robust to malicious actors and data poisoning ([R7](https://arxiv.org/abs/2012.10333), [R8](https://arxiv.org/abs/2006.09365))
-- :apple: :banana: interpretable in imperfectly interoperable data distributions ([R9](https://arxiv.org/abs/2107.06580))
-- :mirror: personalizable  ([R10](https://arxiv.org/abs/2103.00710))
-- :carrot: fairly incentivizing participation
-
-
-___
-
-
 :checkered_flag: **HOW TO USE DISCO**
 - Start by exploring our examples tasks in the [`DISCOllaboratives` page](https://discolab.ai/#/list). 
 - The example DISCOllaboratives are based on popular ML tasks such as [GPT2](https://d4mucfpksywv.cloudfront.net/better-language-models/language-models.pdf), [Titanic](https://www.kaggle.com/c/titanic), [MNIST](https://www.kaggle.com/c/digit-recognizer) or [CIFAR-10](https://www.kaggle.com/pankrzysiu/cifar10-python)