Research map / Video / Video segmentation and tracking
Recurrent Ladder Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| add next-step prediction as an auxiliary task (ablation) | RLadder · Occluded Moving MNIST | classification error (%): 1.51 ± 0.21 -> 0.74 ± 0.09 | Helped |
| add decoder-to-encoder connections (ablation) | RLadder · Occluded Moving MNIST | classification error (%): 1.24 ± 0.05 -> 0.74 ± 0.09 | Helped |
| add a decoder (ablation) | hierarchical RNN encoder · Occluded Moving MNIST | classification error (%): 1.60 ± 0.05 -> 0.74 ± 0.09 | Helped |
| add the classification task (ablation) | RLadder · Occluded Moving MNIST | prediction error (×10^-5): 155.2 ± 2.5 -> 150.1 ± 0.1 | Helped |
| add Mean Teacher consistency training using unlabeled data | RLadder · Occluded Moving MNIST, 1k labeled + 59k unlabeled | classification error (%): 5.20 ± 0.77 -> 1.69 ± 0.14 | Helped |
| replace Tagger's Ladder engine with the recurrent Ladder engine | Tagger · Brodatz-textured MNIST, 50k labeled | classification error (%): 12.15 ± 0.1 -> 5.9 ± 0.2 | Helped |
| iterate recurrent inference from iteration 1 to iteration 3 | RTagger · Brodatz-textured MNIST, 50k labeled | segmentation accuracy (AMI): 0.55 -> 0.80 ± 0.01 | Helped |
| iterate recurrent inference from iteration 1 to iteration 3 | RTagger · Brodatz-textured MNIST, 50k labeled | classification error (%): 18.2 -> 5.9 ± 0.2 | Helped |
| use recurrent inference rather than Tagger | perceptual grouping · Moving MNIST | training success | Mixed |
| use marginal note probabilities instead of a joint distribution | RLadder · Polyphonic music datasets | negative log-likelihood: 6.19 ± 0.02 vs 5.00; 2.42 ± 0.03 vs 1.62; 5.69 ± 0.02 vs 3.90 | Hurt |
Closest papers by meaning
- Combining Recurrent, Convolutional, and Continuous-time Models with Linear State Space Layers (NeurIPS 2021 · similarity 0.86)
- Temporal Latent Bottleneck: Synthesis of Fast and Slow Processing Mechanisms in Sequence Learning (NeurIPS 2022 · similarity 0.86)
- Deconstructing the Ladder Network Architecture (ICML 2016 · similarity 0.85)
- A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music (ICML 2018 · similarity 0.85)
- Learning dynamics in linear recurrent neural networks (ICML 2025 · similarity 0.83)
- Complex priors and flexible inference in recurrent circuits with dendritic nonlinearities (ICLR 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.