Reparameterized Importance Sampling for Robust Variational Bayesian Neural Networks

Authors: Yunfei Long, Zilin Tian, Liguo Zhang, Huosheng Xu

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental results demonstrate the effectiveness of the proposed RIS method in three critical aspects: improved convergence, enhanced predictive performance, and successful uncertainty estimation for out-of-distribution data.
Researcher Affiliation Academia 1College Of Computer Science And Technology, Harbin Engineering University, Harbin, Heilongjiang, China 2Modeling and Emulation in E-Government National Engineering Laboratory, Harbin Engineering University, Harbin, Heilongjiang, China.
Pseudocode Yes Algorithm 1 The first moment propagation in l-th via Reparameterized Importcance Sampling
Open Source Code No The paper does not provide any explicit statement or link for open-source code for the methodology described.
Open Datasets Yes Our experiments on real-world applications include Le Net architecture (Le Cun et al., 1998) for MNIST digit dataset, Res Net20, Res Net56 architecture(He et al., 2016), on CIFAR-10 and CIFAR-100 datasets (Krizhevsky et al., 2009).
Dataset Splits Yes The accuracies represent how well the model recognizes the validation set during training.
Hardware Specification Yes We implement the above Bayesian architecture and train them with RIS and with standard Monte Carlo sampling (...), under the Py Torch framework, on a Titan RTX 28G device, and using the same random seeds.
Software Dependencies No The paper mentions using the 'Py Torch framework' but does not specify a version number or other software dependencies with their versions.
Experiment Setup Yes We optimize the two loss objectives using adam (Kingma & Ba, 2014) for same step size.