Research map / Architectures and efficiency / RNNs and normalization
Understanding the Representation and Computation of Multilayer Perceptrons: A Case Study in Speech Recognition
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 |
|---|---|---|---|
| Use wider hidden layers (256 -> 2048 nodes, with batch norm and dropout) | 5-layer MLP acoustic model · WSJ eval93 | frame-wise accuracy: 80.56% -> 81.91% | Helped |
| Use wider hidden layers (256 -> 2048 nodes) | 5-layer MLP · WSJ TR284 | frame-wise accuracy: 81.43% -> 84.53% | Helped |
| Go deeper from input features to HL5 (256-node net) | LDA probe · WSJ eval93 | LDA accuracy: 56.04% -> 75.65% | Helped |
| Go deeper from input features to HL5 (2048-node net) | LDA probe · WSJ eval93 | LDA accuracy: 56.04% -> 80.04% | Helped |
| Use 2048 instead of 256 nodes at layer HL1 | LDA probe · WSJ dev93 | LDA accuracy: 67.38% -> 74.58% | Helped |
| Use 2048 instead of 256 nodes at layer HL5 | LDA probe · WSJ dev93 | LDA accuracy: 76.78% -> 82.70% | Helped |
Closest papers by meaning
- Untangling in Invariant Speech Recognition (NeurIPS 2019 · similarity 0.90)
- AND: Audio Network Dissection for Interpreting Deep Acoustic Models (ICML 2024 · similarity 0.86)
- Towards Interpretability Without Sacrifice: Faithful Dense Layer Decomposition with Mixture of Decoders (NeurIPS 2025 · similarity 0.84)
- The Brain's Bitter Lesson: Scaling Speech Decoding With Self-Supervised Learning (ICML 2025 · similarity 0.84)
- Transcoders find interpretable LLM feature circuits (NeurIPS 2024 · similarity 0.83)
- Unveiling Induction Heads: Provable Training Dynamics and Feature Learning in Transformers (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.