Research map / Architectures and efficiency / RNNs and normalization
Generalized Tensor Models for Recurrent Neural 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 |
|---|---|---|---|
| Replace product nonlinearity with ReLU (ablation) | Generalized RNN · MNIST | accuracy: 97.39 -> 97.45 | Helped |
| Replace product nonlinearity with log-sum-exp (ablation) | Generalized RNN · MNIST | accuracy: 97.39 -> 97.68 | Helped |
| Replace product nonlinearity with addition (ablation) | Generalized RNN · MNIST | accuracy: 97.39 -> 96.28 | Hurt |
| Replace product nonlinearity with Euclidean norm (ablation) | Generalized RNN · MNIST | accuracy: 97.39 -> 96.44 | Hurt |
| Replace product nonlinearity with ReLU (ablation) | Generalized RNN · CIFAR-10 | accuracy: 43.08 -> 48.09 | Helped |
| Replace product nonlinearity with log-sum-exp (ablation) | Generalized RNN · CIFAR-10 | accuracy: 43.08 -> 55.37 | Helped |
| Replace product nonlinearity with addition (ablation) | Generalized RNN · CIFAR-10 | accuracy: 43.08 -> 57.18 | Helped |
| Replace product nonlinearity with Euclidean norm (ablation) | Generalized RNN · CIFAR-10 | accuracy: 43.08 -> 49.04 | Helped |
| Replace product nonlinearity with ReLU (ablation) | Generalized RNN · IMDB | accuracy: 83.33 -> 84.35 | Helped |
| Replace product nonlinearity with log-sum-exp (ablation) | Generalized RNN · IMDB | accuracy: 83.33 -> 82.25 | Hurt |
| Replace product nonlinearity with addition (ablation) | Generalized RNN · IMDB | accuracy: 83.33 -> 81.28 | Hurt |
| Replace product nonlinearity with Euclidean norm (ablation) | Generalized RNN · IMDB | accuracy: 83.33 -> 79.76 | Hurt |
Closest papers by meaning
- Architectural Complexity Measures of Recurrent Neural Networks (NeurIPS 2016 · similarity 0.88)
- Capacity and Trainability in Recurrent Neural Networks (ICLR 2017 · similarity 0.88)
- On the Provable Generalization of Recurrent Neural Networks (NeurIPS 2021 · similarity 0.88)
- Recurrent neural networks: vanishing and exploding gradients are not the end of the story (NeurIPS 2024 · similarity 0.88)
- Exploring Sparsity in Recurrent Neural Networks (ICLR 2017 · similarity 0.88)
- A Tensor Decomposition Perspective on Second-order RNNs (ICML 2024 · similarity 0.87)
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.