LGFeb 10, 2024

In-Context Data Distillation with TabPFN

MILA
arXiv:2402.06971v115 citationsh-index: 16
Originality Incremental advance
AI Analysis

This addresses a key limitation for applying TabPFN in real-world tabular data scenarios, though it is an incremental improvement on an existing method.

The paper tackles the data size constraint of TabPFN, a transformer model for tabular data, by introducing in-context data distillation (ICD), which optimizes its context to handle larger datasets with fixed memory, achieving strong performance on 48 large tabular datasets against tree-based and deep learning models.

Foundation models have revolutionized tasks in computer vision and natural language processing. However, in the realm of tabular data, tree-based models like XGBoost continue to dominate. TabPFN, a transformer model tailored for tabular data, mirrors recent foundation models in its exceptional in-context learning capability, being competitive with XGBoost's performance without the need for task-specific training or hyperparameter tuning. Despite its promise, TabPFN's applicability is hindered by its data size constraint, limiting its use in real-world scenarios. To address this, we present in-context data distillation (ICD), a novel methodology that effectively eliminates these constraints by optimizing TabPFN's context. ICD efficiently enables TabPFN to handle significantly larger datasets with a fixed memory budget, improving TabPFN's quadratic memory complexity but at the cost of a linear number of tuning steps. Notably, TabPFN, enhanced with ICD, demonstrates very strong performance against established tree-based models and modern deep learning methods on 48 large tabular datasets from OpenML.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes