URSABench: A System for Comprehensive Benchmarking of Bayesian Deep Neural Network Models and Inference Methods
URSABench is an open-source benchmarking suite for assessing Bayesian deep learning models and inference methods, focusing on uncertainty quantification, robustness, scalability, and accuracy in classification tasks for both server and edge GPUs.
Apr 1, 2022
EDITS: Modeling and Mitigating Data Bias for Graph Neural Networks
Apr 1, 2022
Impact of Parameter Sparsity on Stochastic Gradient MCMC Methods for Bayesian Deep Learning
Feb 1, 2022
Runtime Monitoring of Deep Neural Networks Using Top-Down Context Models Inspired by Predictive Processing and Dual Process Theory
Jan 1, 2022
Scaling Hamiltonian Monte Carlo Inference for Bayesian Neural Networks with Symmetric Splitting
We introduce a new symmetric integration scheme for split Hamiltonian Monte Carlo (HMC) that enables efficient inference for Bayesian neural networks on large datasets. Our method outperforms stochastic gradient MCMC in accuracy and uncertainty quantification, demonstrating HMC as a viable option for large-scale machine learning problems.
Dec 1, 2021
Robust Decision-Making in the Internet of Battlefield Things Using Bayesian Neural Networks
Dec 1, 2021
Improving Differential Evolution through Bayesian Hyperparameter Optimization
Jun 1, 2021
Trinity: Trust, Resilience and Interpretability of Machine Learning Models
Jan 1, 2021
Toward Safe Decision-Making via Uncertainty Quantification in Machine Learning
Jan 1, 2021
Machine learning raw network traffic detection
Jan 1, 2021