Unverified50% confidenceFactExact time
The Expectation-Maximization (EM) algorithm was formally proposed by Dempster, Laird, and Rubin in 1977 for parameter estimation of probabilistic models with latent variables.
1
Sources
50%
Confidence
Long-term
Relevance
9/4/2026
First Seen
Sources
Related Entities
Related Claims
Unverified卡尔曼滤波是线性高斯假设下的贝叶斯最优估计器，其预测-更新两步迭代对应先验与后验概率的递推计算65% similarUnverified贝叶斯优化基于概率代理模型（通常为高斯过程）建模历史实验结果，在探索与利用之间智能权衡，通常能以更少实验次数找到接近最优的参数配置64% similarUnverifiedProbability calibration refers to a post-processing technique that converts model output prediction scores into true probability estimates, with commonly used methods including Platt Scaling and Isotonic Regression.64% similarUnverified生成式模型建模联合概率 P(x,y) 或数据分布 P(x)，判别式模型直接建模条件概率 P(y|x)，该区分最初由 Andrew Ng 和 Michael Jordan 在 2001 年的论文中系统比较63% similarUnverifiedBayesian optimization constructs a probabilistic surrogate model of the objective function (typically a Gaussian process) and uses an acquisition function (such as Expected Improvement or Upper Confidence Bound) to determine the next sampling point.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/852486API
curl https://kongchang.com/api/v1/knowledge/claims/852486MCP
get_claim(id=852486)