Research map / Architectures and efficiency / CNN design
Sequence Modeling with Multiresolution Convolutional Memory
Jiaxin Shi, Ke Alexander Wang, Emily Fox
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 |
|---|---|---|---|
| train filters end-to-end instead of fixed wavelet filters (ablation) | MultiresConv · ListOps (smaller model) | accuracy (%): 50.15 -> 52.08 | Improved |
| Xavier initialization instead of wavelet initialization of filters (ablation) | MultiresConv trainable filters · ListOps (smaller model) | accuracy (%): 52.08 -> 51.70 | No clear effect |
| increase MultiresConv depth from 7 to 11 (ablation) | MultiresNet with Xavier init · ListOps (smaller model) | accuracy (%): 51.70 -> 61.07 | Improved |
| increase filter size from 2 to 4 (ablation) | MultiresNet with wavelet init · ListOps (smaller model) | accuracy (%): 52.08 -> 59.23 | Improved |
| uniform-over-time memory instead of resolution fading TreeSelect (ablation) | MultiresNet memory mechanism · ListOps (smaller model) | accuracy (%): 51.70 -> 51.58 | No clear effect |
| MultiresNet architecture (learned filters, tree-structured dilated convs with memory) vs. prior models | sequential CIFAR-10 classification · sCIFAR | accuracy (%): 93.15 vs 92.02 (Liquid-S4, next best) | Improved |
| MultiresNet on PTB-XL ECG classification | PTB-XL · PTB-XL All/Rhythm | AUROC: All 0.938 (tied with S4); Rhythm 0.975 vs S4 0.977 | Mixed |
| MultiresNet without 2D bias on CIFAR-10 autoregressive density estimation | autoregressive CIFAR-10 · CIFAR-10 bpd | bits per dimension: 2.84 vs Sparse Transformer 2.80 (better than S4 2.85 and PixelSNAIL 2.85) | Mixed |
Closest papers by meaning
- Combining Recurrent, Convolutional, and Continuous-time Models with Linear State Space Layers (NeurIPS 2021 · similarity 0.87)
- Efficiently Modeling Long Sequences with Structured State Spaces (ICLR 2022 · similarity 0.87)
- Orchid: Flexible and Data-Dependent Convolution for Sequence Modeling (NeurIPS 2024 · similarity 0.85)
- Reparameterized Multi-Resolution Convolutions for Long Sequence Modelling (NeurIPS 2024 · similarity 0.85)
- Utilizing Image Transforms and Diffusion Models for Generative Modeling of Short and Long Time Series (NeurIPS 2024 · similarity 0.85)
- Fundamental limits of learning in sequence multi-index models and deep attention networks: high-dimensional asymptotics and sharp thresholds (ICML 2025 · similarity 0.84)
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.
