Overcoming Catastrophic Forgetting in Tabular Data Classification: A Pseudorehearsal-based approach

TL;DR

This paper introduces TRIL3, a novel continual learning framework for tabular data that uses generative models and incremental algorithms to prevent catastrophic forgetting without storing old data, outperforming existing methods.

Contribution

The paper presents TRIL3, a new pseudorehearsal-based method combining XuILVQ and modified DNDF for effective lifelong learning on tabular data without retaining old samples.

Findings

TRIL3 outperforms other continual learning methods in tabular data classification.

TRIL3 achieves comparable performance using only 50% synthetic data.

The framework effectively prevents catastrophic forgetting in tabular data tasks.

Abstract

Continual learning (CL) poses the important challenge of adapting to evolving data distributions without forgetting previously acquired knowledge while consolidating new knowledge. In this paper, we introduce a new methodology, coined as Tabular-data Rehearsal-based Incremental Lifelong Learning framework (TRIL3), designed to address the phenomenon of catastrophic forgetting in tabular data classification problems. TRIL3 uses the prototype-based incremental generative model XuILVQ to generate synthetic data to preserve old knowledge and the DNDF algorithm, which was modified to run in an incremental way, to learn classification tasks for tabular data, without storing old samples. After different tests to obtain the adequate percentage of synthetic data and to compare TRIL3 with other CL available proposals, we can conclude that the performance of TRIL3 outstands other options in the literature using only 50% of synthetic data.