Tracker: docs/FORMULAB_V1_TASK_TRACKER.md#fvl-09--active-learning--next-best-experiment
Uncertainty-aware experiment recommendation loop over the existing DOE + Laboratory architecture. Does NOT replace the DOE engine. 10 tasks (FVL-09.001..010).
Acceptance criteria: safety/regulatory filtering hard-excludes candidates; defensible acquisition criterion (documented); human selection required, no automatic lab execution; completed experiments feed retraining; an unperformed recommendation is never labeled validated.
Dependencies: FVL-07, FVL-02.
Scope freeze rule: see docs/FORMULAB_V1_FINAL_SCOPE.md — no new DOE system.
Tracker: docs/FORMULAB_V1_TASK_TRACKER.md#fvl-09--active-learning--next-best-experiment
Uncertainty-aware experiment recommendation loop over the existing DOE + Laboratory architecture. Does NOT replace the DOE engine. 10 tasks (FVL-09.001..010).
Acceptance criteria: safety/regulatory filtering hard-excludes candidates; defensible acquisition criterion (documented); human selection required, no automatic lab execution; completed experiments feed retraining; an unperformed recommendation is never labeled validated.
Dependencies: FVL-07, FVL-02.
Scope freeze rule: see docs/FORMULAB_V1_FINAL_SCOPE.md — no new DOE system.