Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Nonparametric Bayesian Deep Networks with Local Competition
Konstantinos Panousis, Sotirios Chatzis, Sergios Theodoridis
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 SB-LWTA with two competing units instead of ReLU | LeNet-300-100 · MNIST | error (%): 1.60 -> 1.31 | Improved |
| use SB-LWTA with four competing units instead of ReLU | LeNet-300-100 · MNIST | error (%): 1.60 -> 1.34 | Improved |
| reduce weight bit precision using inferred posterior variance | SB-LWTA/4 units · MNIST, LeNet-300-100 | error (%): 1.34 -> 1.5 | Hurt |
| replace LWTA blocks with ReLU units (ablation) | SB-LWTA · MNIST, LeNet-300-100 with connections thresholded | Error (%): 1.7 -> 1.75 | Mixed |
| increase competitors per LWTA block from two to four (ablation) | SB-LWTA · MNIST, LeNet-300-100 | error (%): 1.31 -> 1.34 | Hurt |
| replace LWTA blocks with ReLU units (ablation) | SB-LWTA · MNIST, LeNet-5-Caffe | Error (%): 0.9 -> 0.9 | No clear effect |
| use four competing feature maps instead of two (ablation) | SB-LWTA · MNIST, LeNet-5-Caffe | Error (%): 0.9 -> 0.8 | Improved |
| infer and omit unnecessary connections using utility indicators (ablation) | SB-LWTA (2 units) · MNIST, LeNet-300-100 | Error (%): 1.31 -> 1.7 | Mixed |
| infer and omit unnecessary convolutional components | SB-LWTA-2 · MNIST, LeNet-5-Caffe | Error (%): 0.9 | No clear effect |
Closest papers by meaning
- Parsimonious Bayesian deep networks (NeurIPS 2018 · similarity 0.89)
- Bayesian Deep Learning via Subnetwork Inference (ICML 2021 · similarity 0.88)
- Encoding the Latent Posterior of Bayesian Neural Networks for Uncertainty Quantification (TPAMI 2023 · similarity 0.87)
- On the Expressiveness of Approximate Inference in Bayesian Neural Networks (NeurIPS 2020 · similarity 0.87)
- A Symmetry-Aware Exploration of Bayesian Neural Network Posteriors (ICLR 2024 · similarity 0.86)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 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.
