Unverified50% confidenceFactExact time
Under matched corpus size and compute, randomized SAFE's true yield is 2.8x that of randomized SMILES.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe two levels of stochasticity in doubly stochastic optimization are mini-batch random sampling over the data and Monte Carlo estimation of the variational expectation.74% similarUnverifiedOpportunity-normalized accuracy is computed as (model accuracy − random-guess accuracy) / (1 − random-guess accuracy).70% similarUnverified研究表明即使架构和超参数完全相同，不同随机种子训练出的模型可能在准确率上产生 0.5% 到 2% 的自然波动68% similarUnverifiedRandomizing only the fragment order of SAFE achieves just one-seventh of the genuinely novel molecule yield of simultaneous dual-axis randomization.67% similarUnverifiedFor a randomly ranking classifier, the expected Average Precision is approximately equal to the positive-class prevalence; at a prevalence of 1%, the random baseline AP is about 0.01.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114335API
curl https://kongchang.com/api/v1/knowledge/claims/1114335MCP
get_claim(id=1114335)