Matthew Peroni

Matthew Peroni, Marharyta Kurban, Sun Young Yang et al.

With increasing interest in applying machine learning to develop healthcare solutions, there is a desire to create interpretable deep learning models for survival analysis. In this paper, we extend the Neural Additive Model (NAM) by incorporating pairwise feature interaction networks and equip these models with loss functions that fit both proportional and non-proportional extensions of the Cox model. We show that within this extended framework, we can construct non-proportional hazard models, which we call TimeNAM, that significantly improve performance over the standard NAM model architecture on benchmark survival datasets. We apply these model architectures to data from the Electronic Health Record (EHR) database of Seoul National University Hospital Gangnam Center (SNUHGC) to build an interpretable neural network survival model for gastric cancer prediction. We demonstrate that on both benchmark survival analysis datasets, as well as on our gastric cancer dataset, our model architectures yield performance that matches, or surpasses, the current state-of-the-art black-box methods.

Matthew Peroni, Franck Le, Vadim Sheinin

The development of tabular foundation models (TFMs) has accelerated in recent years, showing strong potential to outperform traditional ML methods for structured data. A key finding is that TFMs can be pretrained entirely on synthetic datasets, opening opportunities to design data generators that encourage desirable model properties. Prior work has mainly focused on crafting high-quality priors over generators to improve overall pretraining performance. Our insight is that parameterizing the generator distribution enables an adversarial robustness perspective: during training, we can adapt the generator to emphasize datasets that are particularly challenging for the model. We formalize this by introducing an optimality gap measure, given by the difference between TFM performance and the best achievable performance as estimated by strong baselines such as XGBoost, CatBoost, and Random Forests. Building on this idea, we propose Robust Tabular Foundation Models (RTFM), a model-agnostic adversarial training framework. Applied to the TabPFN V2 classifier, RTFM improves benchmark performance, with up to a 6% increase in mean normalized AUC over the original TabPFN and other baseline algorithms, while requiring less than 100k additional synthetic datasets. These results highlight a promising new direction for targeted adversarial training and fine-tuning of TFMs using synthetic data alone.

Dimitris Bertsimas, Lisa Everest, Jiayi Gu et al.

Decision Trees have remained a popular machine learning method for tabular datasets, mainly due to their interpretability. However, they lack the expressiveness needed to handle highly nonlinear or unstructured datasets. Motivated by recent advances in tree-based machine learning (ML) techniques and first-order optimization methods, we introduce Generalized Soft Trees (GSTs), which extend soft decision trees (STs) and are capable of processing images directly. We demonstrate their advantages with respect to tractability, performance, and interpretability. We develop a tractable approach to growing GSTs, given by the DeepTree algorithm, which, in addition to new regularization terms, produces high-quality models with far fewer nodes and greater interpretability than traditional soft trees. We test the performance of our GSTs on benchmark tabular and image datasets, including MIMIC-IV, MNIST, Fashion MNIST, CIFAR-10 and Celeb-A. We show that our approach outperforms other popular tree methods (CART, Random Forests, XGBoost) in almost all of the datasets, with Convolutional Trees having a significant edge in the hardest CIFAR-10 and Fashion MNIST datasets. Finally, we explore the interpretability of our GSTs and find that even the most complex GSTs are considerably more interpretable than deep neural networks. Overall, our approach of Generalized Soft Trees provides a tractable method that is high-performing on (un)structured datasets and preserves interpretability more than traditional deep learning methods.