Issues

• ODELIA collect csvs for test evaluation

  Priority: MediumType: Enhancement
  KatherLab/MediSwarm#629 · Filter by author oleschwen opened on Oct 7, 2026
  2

• pip Installation in Dockerfile

  Priority: LowType: Enhancement
  KatherLab/MediSwarm#628 · Filter by author oleschwen opened on Oct 6, 2026
  1

• mediswarm-vpn service starts OpenVPN but never gets a tunnel address (CAM and RUMC)

  KatherLab/MediSwarm#620 · Filter by author Ultimate-Storm opened on Sep 25, 2026
  1

• Configure quorum proceeds the instant min is reached: a healthy site that answers 7 s later is left out (tolerant profiles)

  KatherLab/MediSwarm#600 · Filter by author Ultimate-Storm opened on Sep 16, 2026

• Deploy-test harness: registration check reads stale sessions from the shared server log, and per-site DOCKER_OPTIONS break a localhost client

  KatherLab/MediSwarm#599 · Filter by author Ultimate-Storm opened on Sep 16, 2026

• T3.3 phase 3: per-round ε logging, budget-exhausted halt, and the utility cost of calibrated noise

  KatherLab/MediSwarm#593 · Filter by author Ultimate-Storm opened on Sep 14, 2026 · M54 — Jun 2027 (D3.3, MS6)
  1

• D3.2: reduced retraining curve — active vs random selection, 2 arms × 2 budgets

  KatherLab/MediSwarm#592 · Filter by author Ultimate-Storm opened on Sep 14, 2026 · M48 — Dec 2026 (D2.5, D3.2, D3.4)
  1

• October 2026 consortium benchmark on 1.8.1: 20 rounds, all eight sites, per-case return at RUMC_1, UMCU_1, MHA_1

  KatherLab/MediSwarm#591 · Filter by author Ultimate-Storm opened on Sep 14, 2026 · M48 — Dec 2026 (D2.5, D3.2, D3.4)
  1

• Sporadic stall of a simulated client after it aggregates (minimal example, weekly 13 Sep): 'didn't report status for 300 seconds'

  KatherLab/MediSwarm#583 · Filter by author Ultimate-Storm opened on Sep 13, 2026

• Rename or split pr-test.yaml

  Priority: MediumType: Enhancement
  KatherLab/MediSwarm#570 · Filter by author oleschwen opened on Sep 10, 2026
  1

• Repair or drop tests of incompatible startup kits failing to connect

  Priority: LowType: Bug
  KatherLab/MediSwarm#567 · Filter by author oleschwen opened on Sep 10, 2026

• Deploy-test model transfers are DERP-relayed through Frankfurt because the server is off-LAN

  KatherLab/MediSwarm#553 · Filter by author Ultimate-Storm opened on Sep 8, 2026
  1

  

• ODELIA consistently treat challenge and non-challenge models

  Priority: LowType: Enhancement
  KatherLab/MediSwarm#546 · Filter by author oleschwen opened on Sep 8, 2026
  1

• ODELIA config_fed_client.conf and config_fed_server.conf

  Priority: LowType: Enhancement
  KatherLab/MediSwarm#540 · Filter by author oleschwen opened on Sep 7, 2026
  

• ODELIA cleanup methods for saving/loading checkpoints

  Priority: LowType: Enhancement
  KatherLab/MediSwarm#538 · Filter by author oleschwen opened on Sep 7, 2026
  

• ODELIA location of training output

  Priority: MediumType: Enhancement
  KatherLab/MediSwarm#537 · Filter by author oleschwen opened on Sep 7, 2026
  

• ODELIA file names for local models

  Priority: MediumType: Bug
  KatherLab/MediSwarm#536 · Filter by author oleschwen opened on Sep 7, 2026
  

• D3.5 (T3.4): demonstrate the SL prototype in an operational environment (TRL7)

  KatherLab/MediSwarm#532 · Filter by author Ultimate-Storm opened on Sep 3, 2026 · M60 — Dec 2027 (D3.5)
  1

  

• MS6: white-hat attack on the swarm-trained model — needs CAM and RUMC

  KatherLab/MediSwarm#531 · Filter by author Ultimate-Storm opened on Sep 3, 2026 · M54 — Jun 2027 (D3.3, MS6)
  1

  

• D3.4: measures needed to make SL-based AI robust against adversarial attacks

  KatherLab/MediSwarm#529 · Filter by author Ultimate-Storm opened on Sep 3, 2026 · M48 — Dec 2026 (D2.5, D3.2, D3.4)
  2

  

• D3.3 (T3.2): SL software v2.0 — active learning plus regulatory testing node

  KatherLab/MediSwarm#528 · Filter by author Ultimate-Storm opened on Sep 3, 2026 · M54 — Jun 2027 (D3.3, MS6)
  1

  

• D3.2 (T3.2): demonstrate the usefulness of active learning

  KatherLab/MediSwarm#527 · Filter by author Ultimate-Storm opened on Sep 3, 2026 · M48 — Dec 2026 (D2.5, D3.2, D3.4)
  2

  

• D2.5 (T2.4): compare regionally fine-tuned models against the pan-European baseline

  KatherLab/MediSwarm#526 · Filter by author Ultimate-Storm opened on Sep 3, 2026 · M48 — Dec 2026 (D2.5, D3.2, D3.4)
  3

  

• STAMP deploy test: evaluation is skipped because it runs on the orchestrator host, not where the data is

  KatherLab/MediSwarm#476 · Filter by author Ultimate-Storm opened on Jul 20, 2026
  

• Backbone: replace the blanket 100h timeouts with sane per-operation timeouts + FT retry

  KatherLab/MediSwarm#466 · Filter by author Ultimate-Storm opened on Jul 20, 2026
  2