Research map / Point clouds / Point cloud classification
LDMNet: Low Dimensional Manifold Regularized 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 |
|---|---|---|---|
| Use LDMNet instead of DropOut regularization | MNIST classifier · MNIST, 50 training/class | testing accuracy: 92.31% -> 95.57% | Helped |
| Use LDMNet instead of DropOut regularization | MNIST classifier · MNIST, 6000 training/class | testing accuracy: 99.41% -> 99.39% | Hurt |
| Use LDMNet instead of the strongest listed conventional regularizer | SVHN classifier · SVHN, 50 training/class | testing accuracy: 71.94% -> 74.64% | Helped |
| Use LDMNet instead of the strongest listed conventional regularizer | CIFAR-10 classifier · CIFAR-10, 50 training/class | testing accuracy: 35.94% -> 41.55% | Helped |
| Fine-tune the VGG-16 model by regularizing its output layer with LDMNet | CIFAR-10 VGG-16 model trained with weight decay and DropOut · CIFAR-10, full data | testing accuracy: 87.72% -> 88.21% | Helped |
| Learn a nonlinear embedding with LDMNet rather than using the VGG-face features directly | VGG-face features · CASIA NIR-VIS 2.0 | rank-1 identification rate: 74.51% to 85.02% | Helped |
| Use LDMNet rather than low-rank embedding | VGG-face features · CASIA NIR-VIS 2.0 | rank-1 identification rate: 80.69 ± 1.02 -> 85.02 ± 0.86 | Helped |
| Use weight decay or DropOut instead of LDMNet for the nonlinear embedding | VGG-face features · CASIA NIR-VIS 2.0 | rank-1 identification rate: 85.02 ± 0.86 vs 63.87 ± 1.33; 66.97 ± 1.31 | Hurt |
Closest papers by meaning
- Deep Networks and the Multiple Manifold Problem (ICLR 2021 · similarity 0.88)
- Approximating Latent Manifolds in Neural Networks via Vanishing Ideals (ICML 2025 · similarity 0.85)
- Graph Geometry-Preserving Autoencoders (ICML 2024 · similarity 0.85)
- Dual Manifold Regularization Steered Robust Representation Learning for Point Cloud Analysis (AAAI 2025 · similarity 0.85)
- Matrix Manifold Neural Networks++ (ICLR 2024 · similarity 0.85)
- Nonparametric Classification on Low Dimensional Manifolds using Overparameterized Convolutional Residual Networks (NeurIPS 2024 · similarity 0.85)
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.