Unverified50% confidenceFactExact time
K-means can be viewed as the hard-limit case of the EM algorithm on Gaussian mixture models, where the EM algorithm uses soft assignments while K-means uses hard assignments
1
Sources
50%
Confidence
Long-term
Relevance
9/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe K-means clustering algorithm is essentially a specific instance of coordinate descent59% similarUnverified命名中的K_M后缀表示采用分组量化策略，在相同比特数下通常保留更好的模型精度55% similarUnverified在特定垂直领域语料上微调的Embedding模型往往能提升同义查询相似度得分，降低阈值误判率55% similarUnverifiedMLA通过低秩分解将KV Cache显存开销从O(n·d_model)压缩到O(n·d_latent)，其中d_latent约为d_model的1/854% similarUnverifiedKL散度在变分推断（VAE的ELBO目标）、知识蒸馏和模型压缩中扮演核心角色54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/851289API
curl https://kongchang.com/api/v1/knowledge/claims/851289MCP
get_claim(id=851289)