Research map / Probabilistic and scientific modeling / Variational and generative sampling
Inference from Quantized Data via Normal Variance-Mean Mixtures
Chenyu Gao, Zhexian Yang, Ziping Zhao
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace Gaussian with symmetric GH distribution in multi-bit quantized matrix completion | multi-bit ECM · Quantized matrix completion, MovieLens 1M (40% train) | RMSE: 0.9151 -> 0.8908 | Improved |
| replace Gaussian with t distribution in multi-bit quantized matrix completion | multi-bit ECM · Quantized matrix completion, MovieLens 1M (40% train) | RMSE: 0.9151 -> 0.9091 | Improved |
| replace Gaussian with symmetric GH distribution | multi-bit ECM · Quantized matrix completion, MovieLens 1M (40% train) | accuracy: 0.4453 -> 0.4503 | Improved |
| use multi-bit instead of 1-bit quantization model with Gaussian | Gaussian ECM · Quantized matrix completion, MovieLens 1M (40% train) | RMSE: 0.9659 -> 0.9151 | Improved |
| use multi-bit GH instead of SVD baseline | matrix completion · Quantized matrix completion, MovieLens 1M (40% train) | RMSE: 0.9148 -> 0.8908 | Improved |
| use multi-bit GH instead of Gaussian: running time | multi-bit ECM · Quantized matrix completion, MovieLens 1M (40% train) | time (s): 90.84 -> 272.18 | Hurt |
| replace Gaussian with GH distribution in 1-bit compressive sensing | 1-bit ECM · 1-bit compressive sensing, SNR = -5 dB | cosine similarity: 0.5487 -> 0.7705 | Improved |
| replace Gaussian with GH distribution in 1-bit compressive sensing | 1-bit ECM · 1-bit compressive sensing, SNR = -10 dB | cosine similarity: 0.2896 -> 0.5386 | Improved |
| replace Gaussian with GH distribution in 1-bit compressive sensing | 1-bit ECM · 1-bit compressive sensing, SNR = 0 dB | cosine similarity: 0.8220 -> 0.8906 | Improved |
| replace Gaussian with t distribution in 1-bit compressive sensing | 1-bit ECM · 1-bit compressive sensing, SNR = -10 dB | cosine similarity: 0.2896 -> 0.3031 | No clear effect |
| use heavy-tailed or skewed NVMM models instead of OLS and Gaussian EM | quantized linear regression · synthetic, t and log-normal noise | MSE | Improved |
Closest papers by meaning
- Toward Global Convergence of Gradient EM for Over-Paramterized Gaussian Mixture Models (NeurIPS 2024 · similarity 0.86)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · similarity 0.86)
- Efficient Mixture Learning in Black-Box Variational Inference (ICML 2024 · similarity 0.85)
- GSQ: Highly-Accurate Low-Precision Scalar Quantization for LLMs via Gumbel-Softmax Sampling (NeurIPS 2026 · similarity 0.85)
- Conformal Inference under High-Dimensional Covariate Shifts via Likelihood-Ratio Regularization (NeurIPS 2025 · similarity 0.85)
- Is $\sqrt{d}$ Separation Necessary for Gradient EM to Learn Gaussian Mixtures in High Dimensions? (NeurIPS 2026 · similarity 0.85)
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.
