Mechanism · Training-transcript verification (proof-of-learning)No provably robust verification without better optimisation theoryOverviewEvidence & limitsSourcesOn this pageFindingSources← All known flawsSignificantTheoretical argumentOpenFang et al. conclude that "one cannot develop a provably robust PoL verification mechanism without further understanding of optimization in deep learning" 4. Shavit notes that the PoL literature "has been heuristic-based" 5. Sources: [4] · [5]