Enhanced and Efficient Reasoning in Large Learning Models

Leslie G. Valiant

arXiv:2605.14036 · 2026-05-16 공개 · arXiv · PDF

large-language-models world-model unary-relational-integracode robust-logic relational-reasoning polynomial-time-learning efficient-reasoning relational-rules

Abstract

In current Large Language Models we can trust the production of smoothly flowing prose on the basis of the principles of machine learning. However, there is no comparably principled basis to justify trust in the content of the text produced. It appears to be conventional wisdom that addressing this issue by adding more principled reasoning is not computationally affordable. Here we propose a principled method of reasoning that is efficient enough to be practical for large language models. Further, the method allows the retention of much of the currently used software and hardware base. Our method for improving the functioning of large language models consists of a first stage of preprocessing that recodes the data to a Unary Relational Integracode that is more explicit about the relationships among the objects described in the text, followed as a second stage by a standard but possibly streamlined machine learning process that then also learns to predict these relationships. The method may be viewed as realizing a world model and applying beyond natural language, to vision and actions, for example, where the multiple properties of an object referred to in an input are brought together explicitly, rather than remaining distributed in the various references to it in the input. We articulate its advantages in terms of Robust Logic, a system for performing principled chaining on learned, and hence uncertain, information. We show that this recoding has the surprising and fortuitous property that, while succinct, it makes the task of learning a core subset of relational rules that hold in the world described in the training data polynomial time learnable in a defined sense, the polynomial depending on the complexity of the rule. This gives support for sound reasoning within each single call of the learned classifier as well as between multiple calls.

한국어 요약

📋 한 줄 요약

**[LLM Reasoning / Foundations]** LLM이 매끄러운 문장을 넘어 신뢰할 수 있는 추론을 하도록, 데이터를 Unary Relational Integracode로 재부호화한 뒤 표준 학습을 수행하는 원칙적·실용적 추론 방법을 제안한다.

🎯 핵심 기여도

💡 핵심 아이디어

자연어 표면에 산재된 객체 속성을 명시적으로 한데 모으는 "world model 같은 표현"으로 입력을 재부호화하면, 학습기는 분포 패턴 외에 객체 간 관계 규칙도 함께 배울 수 있고, 그 결과 추론이 단순한 표면 외삽이 아닌 학습된 규칙 적용이 된다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: LLM에 강한 추론 능력을 부여하기 위해 모델을 통째로 바꾸지 않고도 데이터 표현 차원에서 접근할 수 있는 길을 제시. **한계**: 본 발표는 제안과 이론 분석 중심으로, 대규모 실험적 검증은 아직 제한적이며, 효과적인 unary relational encoder 설계 자체의 난이도가 남는다.

🚀 실용적 활용