Research map / Learning theory and algorithms / Prediction and decision losses
AutoEncoder by Forest
Ji Feng, Zhi-Hua Zhou
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 unsupervised rather than supervised tree construction (ablation) | eForest · MNIST test | MSE: 1386.96 -> 27.39 (500 trees); 701.99 -> 6.86 (1,000 trees) | Improved |
| Use unsupervised rather than supervised tree construction (ablation) | eForest · CIFAR-10 test | MSE: 1623.93 -> 579.337 (500 trees); 567.64 -> 153.68 (1,000 trees) | Improved |
| Increase the forest from 500 to 1,000 trees (ablation) | eForest supervised · MNIST test | MSE: 1386.96 -> 701.99 | Improved |
| Increase the forest from 500 to 1,000 trees (ablation) | eForest unsupervised · MNIST test | MSE: 27.39 -> 6.86 | Improved |
| Increase the forest from 500 to 1,000 trees (ablation) | eForest supervised · CIFAR-10 test | MSE: 1623.93 -> 567.64 | Improved |
| Increase the forest from 500 to 1,000 trees (ablation) | eForest unsupervised · CIFAR-10 test | MSE: 579.337 -> 153.68 | Improved |
| Use unsupervised rather than supervised tree construction (ablation) | eForest · IMDB test | cosine distance: 0.1132 -> 0.0070 (500 trees); 0.0676 -> 0.0023 (1,000 trees) | Improved |
| Increase the forest from 500 to 1,000 trees (ablation) | eForest unsupervised · IMDB test | cosine distance: 0.0070 -> 0.0023 | Improved |
| Use eForest rather than MLP-AE | auto-encoding · MNIST test | MSE: 163.97 (MLP2) -> 6.86 (eForest unsupervised, 1,000 trees) | Improved |
| Use eForest rather than MLP-AE | auto-encoding · IMDB test | cosine distance: 0.512 -> 0.0023 | Improved |
| Train a reusable eForest rather than an MLP-AE | model reuse across datasets · CIFAR-10-trained model on MNIST test | MSE: 1898.76 -> 90.43 | Improved |
| Train a reusable eForest rather than an MLP-AE | model reuse across datasets · MNIST-trained model on Omniglot test | MSE: 596.24 -> 12.80 | Improved |
Closest papers by meaning
- Supervised autoencoders: Improving generalization performance with unsupervised regularizers (NeurIPS 2018 · similarity 0.84)
- Semantic-Aware Auto-Encoders for Self-Supervised Representation Learning (CVPR 2022 · similarity 0.83)
- Autoencoding Random Forests (NeurIPS 2025 · similarity 0.82)
- Deep Quantization: Encoding Convolutional Activations With Deep Generative Model (CVPR 2017 · similarity 0.82)
- Scaling and evaluating sparse autoencoders (ICLR 2025 · similarity 0.82)
- Improved Baselines with Representation Autoencoders (NeurIPS 2026 · similarity 0.82)
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.
