e3844e186e6eb8736e9f53c0c5889527-Paper.pdf
–Neural Information Processing Systems
Inference networks oftraditional Variational Autoencoders (VAEs) aretypically amortized, resulting in relatively inaccurate posterior approximation compared to instance-wise variational optimization. Recent semi-amortized approaches were proposedtoaddress thisdrawback; however,theiriterativegradient update procedures can be computationally demanding.
Neural Information Processing Systems
Feb-10-2026, 20:35:53 GMT