AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents
Petr Anokhin, Nikita Semenov, Artyom Sorokin, Dmitry Evseev, Mikhail Burtsev, Evgeny Burnaev
2024-07-08
Summary
This paper talks about AriGraph, a new method that helps autonomous agents (like AI systems) learn and remember information better by using a structured memory graph that combines different types of memories while they explore their environment.
What's the problem?
The main problem is that current methods for helping AI agents remember things often use unstructured memory systems, which means they just keep a long list of past experiences. This approach can make it hard for the AI to reason and plan effectively because it doesn't organize the information in a useful way. As a result, these agents struggle with complex decision-making tasks.
What's the solution?
To solve this issue, the authors developed AriGraph, which allows the agent to create a memory graph. This graph integrates two types of memory: semantic memory (general knowledge) and episodic memory (specific experiences). By organizing information in this way, the agent can quickly find and use relevant knowledge based on its current situation and goals. The researchers tested their approach using the Ariadne LLM agent in text-based environments and found that it performed much better than traditional methods, even handling complex tasks without prior training.
Why it matters?
This research is important because it enhances how AI agents learn and make decisions, making them more effective in real-world applications. By improving memory organization, AriGraph can help AI systems perform better in various tasks, such as cooking or cleaning, which require understanding and reasoning about their environment. This advancement could lead to more capable and autonomous AI systems in the future.
Abstract
Advancements in generative AI have broadened the potential applications of Large Language Models (LLMs) in the development of autonomous agents. Achieving true autonomy requires accumulating and updating knowledge gained from interactions with the environment and effectively utilizing it. Current LLM-based approaches leverage past experiences using a full history of observations, summarization or retrieval augmentation. However, these unstructured memory representations do not facilitate the reasoning and planning essential for complex decision-making. In our study, we introduce AriGraph, a novel method wherein the agent constructs a memory graph that integrates semantic and episodic memories while exploring the environment. This graph structure facilitates efficient associative retrieval of interconnected concepts, relevant to the agent's current state and goals, thus serving as an effective environmental model that enhances the agent's exploratory and planning capabilities. We demonstrate that our Ariadne LLM agent, equipped with this proposed memory architecture augmented with planning and decision-making, effectively handles complex tasks on a zero-shot basis in the TextWorld environment. Our approach markedly outperforms established methods such as full-history, summarization, and Retrieval-Augmented Generation in various tasks, including the cooking challenge from the First TextWorld Problems competition and novel tasks like house cleaning and puzzle Treasure Hunting.