Hello, wondering if anyone’s tried to introduce non-trainable parameters/layers in a bayesian neural network using PyMC3 (similar to how skip connections exist in tensorflow). Essentially what I’m trying to do is replicate the way offsets work in a GLM, but within a neural network.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Application of bayesian neural nets in bayesian optimization | 0 | 329 | January 24, 2023 | |
| Hierarchical Bayesian Neural Networks with Informative Priors by @twiecki | 6 | 2183 | July 15, 2021 | |
| Deep Neural Network Question | 4 | 1060 | February 25, 2019 | |
| Multi-dimensional input Bayesian Neural Network | 0 | 255 | February 20, 2024 | |
| Bayesian Backpropagation | 1 | 1117 | December 25, 2018 |