Research map / Label-efficient and robust learning / Semi-supervised learning
Few-Shot Non-Parametric Learning with Deep Latent Variable Model
Zhiying Jiang, Yiqin Dai, Ji Xin, Ming Li, Jimmy Lin
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 |
|---|---|---|---|
| use 8 instead of 2 latent variables in BB-ANS (ablation) | BB-ANS · CIFAR-10 | accuracy: 0.369 -> 0.356 | Hurt |
| use 8 instead of 1 latent variable in Bit-Swap (ablation) | Bit-Swap · CIFAR-10 | accuracy: 0.226 -> 0.348 | Improved |
| use 2 instead of 1 latent variable in BB-ANS (ablation) | BB-ANS · CIFAR-10 | accuracy: 0.226 -> 0.369 | Improved |
| greyscale+average instead of average aggregation (ablation) | BB-ANS · CIFAR-10 | accuracy: 0.369 -> 0.408 | Improved |
| minimum instead of average aggregation (ablation) | BB-ANS · CIFAR-10 | accuracy: 0.369 -> 0.284 | Hurt |
| concatenation instead of average aggregation (ablation) | BB-ANS · CIFAR-10 | accuracy: 0.369 -> 0.235 | Hurt |
| greyscale+average instead of average aggregation (ablation) | Bit-Swap · CIFAR-10 | accuracy: 0.339 -> 0.369 | Improved |
| BB-ANS compressor instead of gzip | NPC-LV · CIFAR-10 | accuracy: 31.3 -> 40.2 | Improved |
| BB-ANS instead of Bit-Swap as compressor with NCD | NPC-LV · MNIST | accuracy: 93.2 -> 93.6 | Improved |
| NPC-LV BB-ANS instead of supervised CNN with 5 shots | NPC-LV · CIFAR-10 5-shot | test accuracy: 23.4 -> 35.3 | Improved |
| NPC-LV BB-ANS instead of supervised VGG with 10 shots | NPC-LV · CIFAR-10 10-shot | test accuracy: 29.7 -> 36.0 | Improved |
| NPC-LV BB-ANS instead of VAT with 5 shots | NPC-LV · MNIST 5-shot | test accuracy: 97.0 -> 77.6 | Hurt |
Closest papers by meaning
- Practical lossless compression with latent variables using bits back coding (ICLR 2019 · similarity 0.85)
- Supervised learning through the lens of compression (NeurIPS 2016 · similarity 0.84)
- Unified Multivariate Gaussian Mixture for Efficient Neural Image Compression (CVPR 2022 · similarity 0.84)
- On Error and Compression Rates for Prototype Rules (AAAI 2023 · similarity 0.84)
- Linear Attention Modeling for Learned Image Compression (CVPR 2025 · similarity 0.83)
- Language Models as Zero-shot Lossless Gradient Compressors: Towards General Neural Parameter Prior Models (NeurIPS 2024 · similarity 0.83)
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.
