Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
DISCO Nets : DISsimilarity COefficients 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 deterministic BASE (sigma=1) with DISCO gamma=0.5 | DISCO Nets · Hand pose estimation, test set | ProbLoss (mm): 103.8 -> 83.8 | Helped |
| replace BASE with DISCO gamma=0.5 | DISCO Nets · Hand pose estimation, test set | MeJEE (mm): 25.2 -> 20.9 | Helped |
| replace BASE with DISCO gamma=0.5 | DISCO Nets · Hand pose estimation, test set | FF (80mm): 86.040 -> 94.438 | Helped |
| increase BASE noise sigma from 1 to 10 (ablation) | BASE · Hand pose estimation, test set | ProbLoss (mm): 103.8 -> 96.3 | Helped |
| increase BASE noise sigma from 1 to 10 (ablation) | BASE · Hand pose estimation, test set | MaJEE (mm): 52.7 -> 53.2 | Hurt |
| raise gamma from 0 to 0.5 (ablation) | DISCO Nets · Hand pose estimation, test set | ProbLoss (mm): 92.9 -> 83.8 | Helped |
| raise gamma from 0 to 0.25 (ablation) | DISCO Nets · Hand pose estimation, test set | ProbLoss (mm): 92.9 -> 89.9 | Helped |
| raise gamma from 0.25 to 0.5 (ablation) | DISCO Nets · Hand pose estimation, test set | MaJEE (mm): 46.4 -> 45.1 | Helped |
| all-zero noise at test time (ablation) | DISCO Nets · Hand pose estimation, test set | pointwise metrics (MeJEE, MaJEE) | No clear effect |
| random init instead of DISCO-initialised fixed conv layers for cGAN (ablation) | cGAN · Hand pose estimation, test set | ProbLoss (mm): 128.9 -> 442.7 | Hurt |
| cGAN init fixed vs DISCO gamma=0.5 | cGAN init, fixed · Hand pose estimation, test set | ProbLoss (mm): 83.8 -> 128.9 | Hurt |
| train loss differs from task loss (toy Gaussian example) (ablation) | Gaussian fit · Mixture of Gaussians toy | Delta_A task loss: 13.7 -> 11.6 | Helped |
Closest papers by meaning
- A Probabilistic Approach to Neural Network Pruning (ICML 2021 · similarity 0.85)
- HyperSPNs: Compact and Expressive Probabilistic Circuits (NeurIPS 2021 · similarity 0.85)
- Depth Uncertainty in Neural Networks (NeurIPS 2020 · similarity 0.85)
- Lightweight Probabilistic Deep Networks (CVPR 2018 · similarity 0.85)
- On the Hardness of Training Deep Neural Networks Discretely (AAAI 2025 · similarity 0.84)
- Make Me a BNN: A Simple Strategy for Estimating Bayesian Uncertainty from Pre-trained Models (CVPR 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.