Reliable uncertainty estimate for antibiotic resistance classification with Stochastic Gradient Langevin Dynamics
This addresses the need for reliable uncertainty estimates in antibiotic resistance monitoring, which is critical for global health, but it appears incremental as it applies an existing method (SGLD) to a specific domain.
The paper tackled the problem of poor uncertainty estimates from deep learning models for antibiotic resistance classification when tested on out-of-distribution data, and found that using Stochastic Gradient Langevin Dynamics (SGLD) provided better uncertainty estimates compared to traditional methods like Adam.
Antibiotic resistance monitoring is of paramount importance in the face of this on-going global epidemic. Deep learning models trained with traditional optimization algorithms (e.g. Adam, SGD) provide poor posterior estimates when tested against out-of-distribution (OoD) antibiotic resistant/non-resistant genes. In this paper, we introduce a deep learning model trained with Stochastic Gradient Langevin Dynamics (SGLD) to classify antibiotic resistant genes. The model provides better uncertainty estimates when tested against OoD data compared to traditional optimization methods such as Adam.