Research map / Label-efficient and robust learning / Recognition and long-tail
Adversarially-Regularized Mixed Effects Deep Learning (ARMED) Models Improve Interpretability, Performance, and Generalization on Clustered (non-iid) Data
Kevin P. Nguyen, Alex Treacher, Albert Montillo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-DFNN · Spiral simulation 3 | accuracy (%): 68.5 -> 74.5 | Improved |
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-DFNN · Spiral simulation 1 | accuracy (%): 79.3 -> 78.8 | No clear effect |
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-DFNN · Spiral simulation 2 | accuracy (%): 69.9 -> 65.0 | No clear effect |
| use true cluster membership instead of randomized Z (ablation) | ARMED-DFNN · Spiral simulation 2 | accuracy (%): 50.6 -> 65.0 | Improved |
| use true cluster membership instead of randomized Z (ablation) | ARMED-DFNN · Spiral simulation 1 | accuracy (%): 76.7 -> 78.8 | Improved |
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-DFNN · MCI conversion (seen sites) | AUROC: 0.919 -> 0.926 | Improved |
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-DFNN · MCI conversion (unseen sites) | sensitivity (%): 65.4 -> 72.4 | Improved |
| use true site membership instead of randomized Z (ablation) | ARMED-DFNN · MCI conversion (seen sites) | AUROC: 0.889 -> 0.926 | Improved |
| use Z-predictor inferred site membership instead of random (ablation) | ARMED-DFNN · MCI conversion (unseen sites) | sensitivity (%): 69.8 -> 72.4 | Improved |
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-CNN · AD diagnosis MRI (seen sites) | AUROC: 0.816 -> 0.900 | Improved |
| add adversarial regularization of the fixed effects subnetwork (ablation) | ARMED-CNN · AD diagnosis MRI (unseen sites) | AUROC: 0.655 -> 0.645 | No clear effect |
| use true site membership instead of randomized Z (ablation) | ARMED-CNN · AD diagnosis MRI (seen sites) | AUROC: 0.585 -> 0.900 | Improved |
Closest papers by meaning
- Learning to cluster neuronal function (NeurIPS 2025 · similarity 0.83)
- Adjustment for Confounding using Pre-Trained Representations (ICML 2025 · similarity 0.83)
- An amortized approach to non-linear mixed-effects modeling based on neural posterior estimation (ICML 2024 · similarity 0.82)
- The Effect of Intrinsic Dataset Properties on Generalization: Unraveling Learning Differences Between Natural and Medical Images (ICLR 2024 · similarity 0.82)
- CASL-VAE: Learning Structured Latent Variables from Unpaired Data for Semi-supervised Clustering and Paired Sample Generation (NeurIPS 2026 · similarity 0.82)
- Adversarially Robust Deep Multi-View Clustering: A Novel Attack and Defense Framework (ICML 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
