Unverified50% confidenceFactExact time
Amortized inference originates from the design paradigm of variational autoencoders (VAEs), using a shared inference network to reduce the parameter count from O(n) to the network's fixed parameter count, while enabling one-step inference generalization to new observation points.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
Unverified变分自编码器（VAE）和扩散模型的训练目标建立在变分推断和对数似然最大化的概率框架之上75% similarUnverified贝叶斯定理在朴素贝叶斯分类器、贝叶斯神经网络、变分自编码器（VAE）和扩散模型中都有直接应用71% 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70% similarUnverifiedThe optimization refactors the execution path of the existing VAE at the inference level, rather than retraining the model.70% similarUnverifiedThe derivation of the training objective ELBO (Evidence Lower Bound) for the Variational Autoencoder (VAE) is built upon operations on KL divergence and expectations.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002320API
curl https://kongchang.com/api/v1/knowledge/claims/1002320MCP
get_claim(id=1002320)