Learning together, without pooling the data.
Research is a process of testing the boundaries.
The research direction
The work investigates how participants can collaborate on drug–target interaction modeling while retaining their local datasets. The initial architecture uses centralized federated coordination; a later direction explores peer discovery and distributed coordination.
Two distinct architectures
Centralized FedAvg and a peer-to-peer research direction have different failure and coordination assumptions. Shared terminology does not make their implementations interchangeable. Repository mechanics and scientific claims need separate evidence.
Research status
Two manuscripts connected to the internship were described as under submission. This overview does not imply acceptance, publication or a public preprint. Unpublished paper contents and private datasets are not reproduced here.