Enhancing Generative Decoders with Stochastic Training on Biomedical Data with Missingness
Shen, X.; Bjerregaard, A.; Li, Y.; Krogh, A.
Show abstract
Biomedical datasets are often heterogeneous and affected by noise or missing values. Deep Generative Decoders (DGD) provide a promising framework for latent representation learning, but their standard training procedure relies on sample-level stochastic gradient descent (SGD), which performs poorly with incomplete data. To address this, we introduce two stochastic training strategies -- Nested SGD and Feature Dropout -- that incorporate feature-level randomness into optimization. Evaluations on biomedical tabular datasets demonstrate that Nested SGD improves robustness under missingness the best, while Feature Dropout not only improves but also accelerates convergence with lower computational cost. These results suggest that feature-level stochasticity is a practical way to strengthen biomedical AI pipelines.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- STAMP: Simultaneous Training and Model Pruning for Low Data Regimes in Medical Image Segmentation 93%
- A Deep Graph Neural Network Architecture for Modelling Spatio-temporal Dynamics in resting-state functional MRI Data 93%
- A Framework for Falsifiable Explanations of Machine Learning Models with an Application in Computational Pathology 92%
Similar papers in this journal
Similar papers in this journal
- Uncertainty in Deep Learning for EEG under Dataset Shifts 94%
- Stability of feature selection utilizing Graph Convolutional Neural Network and Layer-wise Relevance Propagation 92%
- Graph Neural Network Modelling as a potentially effective Method for predicting and analyzing Procedures based on Patient Diagnoses 92%
Similar papers in this journal
- Generalized Radiograph Representation Learning via Cross-supervision between Images and Free-text Radiology Reports 93%
- Improving protein function prediction with synthetic feature samples created by generative adversarial networks 92%
- COSIME: Cooperative multi-view integration with Scalable and Interpretable Model Explainer 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.