AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Differentiable Random Partition Models

Thomas M. Sutter, Alain Ryser, Joram Liebeskind, Julia E Vogt

NeurIPS 2023 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Variational methods for sequence models · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
model dependent cluster assignments with DRPM instead of i.i.d. categorical (VaDE)DRPM-VC · MNISTACC: 0.82 -> 0.94Improved
model dependent cluster assignments with DRPM instead of i.i.d. categorical (VaDE)DRPM-VC · FMNISTACC: 0.56 -> 0.65Improved
DRPM-VC versus GMM in latent space on MNISTDRPM-VC · MNISTNMI: 0.86 -> 0.89Improved
use pretrained encoder embeddings instead of raw data space for GMMGMM · MNISTACC: 0.41 -> 0.88Improved
DRPM-VAE learns assignment of latent dimensions to shared factors instead of HG-VAE heuristicDRPM-VAE · mpi3d, n_s=1shared-factor balanced accuracy: 0.22 -> 0.39Improved
DRPM-VAE versus HG-VAEDRPM-VAE · mpi3d, n_s=5shared-factor balanced accuracy: 0.28 -> 0.42Improved
DRPM-VAE versus HG-VAE on independent factors at n_s=5DRPM-VAE · mpi3d, n_s=5independent-factor balanced accuracy: 0.01 -> 0.01No clear effect
DRPM-VAE versus HG-VAE on estimating number of shared factorsDRPM-VAE · mpi3dMSE of estimated n_sNo clear effect
DRPM-MTL partitions last shared layer by task instead of unitary loss scalingDRPM-MTL · noisyMultiMNISTtask accuracyMixed
increase noise ratio alpha on the right task from 0.6 to 0.8DRPM-MTL · noisyMultiMNISTdimensions assigned to right taskImproved
maximum noise ratio alpha=0.9DRPM-MTL · noisyMultiMNISTaverage accuracyHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy