MAPLE: Multi-Agent Adaptive Planning with Long-Term Memory for Table Reasoning

TL;DR

MAPLE is a multi-agent framework that enhances table question answering by mimicking human reasoning, incorporating error detection, long-term memory, and adaptive planning to outperform existing methods.

Contribution

This paper introduces MAPLE, a novel multi-agent system with feedback and memory components, advancing long-term reasoning and error correction in table question answering.

Findings

Achieves state-of-the-art results on WiKiTQ and TabFact datasets.

Significantly improves reasoning accuracy over existing methods.

Demonstrates effective long-term memory utilization for problem-solving.

Abstract

Table-based question answering requires complex reasoning capabilities that current LLMs struggle to achieve with single-pass inference. Existing approaches, such as Chain-of-Thought reasoning and question decomposition, lack error detection mechanisms and discard problem-solving experiences, contrasting sharply with how humans tackle such problems. In this paper, we propose MAPLE (Multi-agent Adaptive Planning with Long-term mEmory), a novel framework that mimics human problem-solving through specialized cognitive agents working in a feedback-driven loop. MAPLE integrates 4 key components: (1) a Solver using the ReAct paradigm for reasoning, (2) a Checker for answer verification, (3) a Reflector for error diagnosis and strategy correction, and (4) an Archiver managing long-term memory for experience reuse and evolution. Experiments on WiKiTQ and TabFact demonstrate significant improvements over existing methods, achieving state-of-the-art performance across multiple LLM backbones.