Research map / Point clouds / Point cloud classification
Deep Sets
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 permutation-equivariant layers with fully connected layers after set pooling (ablation) | DeepSets outlier detector · CelebA set anomaly detection | accuracy: ~6.3% -> 75% | Helped |
| Use image features to condition set model (non-blind vs blind) (ablation) | DeepSets · COCO-Tag | Recall@10: 9.0 -> 31.4 | Helped |
| DeepSets vs w2v NN blind baseline | DeepSets (blind) · COCO-Tag | MRR: 0.021 -> 0.044 | Helped |
| Increase point-cloud size from 100 to 5000 points (ablation) | DeepSets · ModelNet40 | accuracy: 82 ± 2% -> 90 ± .3% | Helped |
| DeepSets vs MLP on red-shift scatter | DeepSets · Red-shift | scatter: 0.026 -> 0.023 | Helped |
| Sum pooling DeepSets vs max pooling NN-max (ablation) | DeepSets · LDA-3k | Recall@10: 5.30 -> 6.04 | Helped |
| DeepSets vs w2v-Near on small LDA-1k | DeepSets · LDA-1k | Recall@10: 6.00 -> 5.53 | Hurt |
| DeepSets vs w2v-Near on LDA-5k | DeepSets · LDA-5k | Recall@10: 4.03 -> 5.54 | Helped |
| DeepSets vs best competitor on ESPgame recall | DeepSets · ESPgame | recall: 32 -> 34 | Helped |
| DeepSets vs best competitor on ESPgame precision | DeepSets · ESPgame | precision: 44 -> 39 | Hurt |
| DeepSets vs RNNs (LSTM, GRU) on digit sum with longer test sets | DeepSets · Sum of digits | accuracy | Helped |
| DeepSets vs SDM on high-dimensional population statistics with many sets | DeepSets · Entropy/mutual information | MSE | Mixed |
Closest papers by meaning
- Set Norm and Equivariant Skip Connections: Putting the Deep in Deep Sets (ICML 2022 · similarity 0.87)
- Shape Constraints for Set Functions (ICML 2019 · similarity 0.86)
- Janossy Pooling: Learning Deep Permutation-Invariant Functions for Variable-Size Inputs (ICLR 2019 · similarity 0.85)
- Learning Representations of Sets through Optimized Permutations (ICLR 2019 · similarity 0.85)
- Regularizing Towards Permutation Invariance In Recurrent Models (NeurIPS 2020 · similarity 0.84)
- Asymptotics of Learning with Deep Structured (Random) Features (ICML 2024 · 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.