Poster De Conférence Année : 2024

Adapting TabPFN for Zero-Inflated Metagenomic Data

Résumé

This paper introduces a novel prior assumption for TabPFN—a meta-learning method designed to approximate Bayesian inference on synthetic datasets generated from a predefined prior—aimed at better accommodating the unique zero-inflated distributions characteristic of metagenomic data. We modify the model's prior assumptions without changing its architecture by generating synthetic training data replicating the sparsity and variability inherent in these datasets. Preliminary results from metagenomic classification tasks show significant improvements in predictive performance, exceeding that of the original TabPFN and competing with state-of-the-art methods. This work emphasizes the necessity of tailoring PFN priors to align with the specific statistical properties of biomedical data, thereby enhancing their effectiveness in precision medicine.

Fichier non déposé

Dates et versions

hal-05197388 , version 1 (02-08-2025)

Identifiants

  • HAL Id : hal-05197388 , version 1

Citer

Giulia Perciballi, Federica Granese, Ahmad Fall, Farida Zehraoui, Edi Prifti, et al.. Adapting TabPFN for Zero-Inflated Metagenomic Data. NeurIPS 2024 Third Table Representation Learning Workshop, Dec 2024, Vancouver, Canada. , 2024. ⟨hal-05197388⟩
142 Consultations
0 Téléchargements

Partager

  • More