提出脑启发式记忆系统，解决大模型幻觉问题并实现高精度与强抗攻击性。
Synthius-Mem: Brain-Inspired Hallucination-Resistant Persona Memory Achieving 94.4% Memory Accuracy and 99.6% Adversarial Robustness on LoCoMo
- 将对话分解为六大认知领域，结构化提取用户人格信息
- 在LoCoMo上达到94.37%准确率与99.55%对抗鲁棒性
- 比全上下文回放节省5倍token，且超越人类表现
为大语言模型智能体提供可靠且不产生幻觉的长期记忆仍是开放难题。现有记忆方法——滑动窗口、摘要、基于嵌入的RAG和扁平事实提取——虽降低令牌开销，但均导致灾难性信息丢失、语义漂移或对用户未披露内容的失控幻觉。根本原因在于架构：所有已发表的记忆系统在LoCoMo基准上都将对话视为对原始或轻度摘要对话段的检索问题，且无一报告对抗鲁棒性（即拒绝回答用户从未披露的事实）。我们提出Synthius-Mem，一种脑启发式的结构化人格记忆系统，采用根本不同的思路：不检索说了什么，而是提取关于一个人知道什么。其全人格提取流程将对话分解为六类认知域（传记、经历、偏好、社交圈、工作、心理测量），各域内去重合并，并通过CategoryRAG以21.79毫秒延迟检索结构化事实。在LoCoMo基准（ACL 2024，10轮对话，1,813个问题）上，Synthius-Mem达成94.37%准确率，超过所有已有系统（包括MemMachine的91.69%，但未报告对抗得分）及人类表现（87.9 F1）。核心记忆事实准确率达98.64%。对抗鲁棒性（即幻觉抵抗能力）达99.55%，是唯一报告该指标的系统。相较于全上下文回放，合成记忆系统减少约5倍令牌消耗，同时实现更高准确率。Synthius-Mem在LoCoMo上取得当前最优性能，据我们所知，是首个既超越人类水平又报告对抗鲁棒性的角色记忆系统。
原文摘要 · Abstract (English)
Providing AI agents with reliable long-term memory that does not hallucinate remains an open problem. Current approaches to memory for LLM agents -- sliding windows, summarization, embedding-based RAG, and flat fact extraction -- each reduce token cost but introduce catastrophic information loss, semantic drift, or uncontrolled hallucination about the user. The structural reason is architectural: every published memory system on the LoCoMo benchmark treats conversation as a retrieval problem over raw or lightly summarized dialogue segments, and none reports adversarial robustness, the ability to refuse questions about facts the user never disclosed. We present Synthius-Mem, a brain-inspired structured persona memory system that takes a fundamentally different approach. Instead of retrieving what was said, Synthius-Mem extracts what is known about the person: a full persona extraction pipeline decomposes conversations into six cognitive domains (biography, experiences, preferences, social circle, work, psychometrics), consolidates and deduplicates per domain, and retrieves structured facts via CategoryRAG at 21.79 ms latency. On the LoCoMo benchmark (ACL 2024, 10 conversations, 1,813 questions), Synthius-Mem achieves 94.37% accuracy, exceeding all published systems including MemMachine (91.69%, adversarial score is not reported) and human performance (87.9 F1). Core memory fact accuracy reaches 98.64%. Adversarial robustness, the hallucination resistance metric that no competing system reports, reaches 99.55%. Synthius-Mem reduces token consumption by ~5x compared to full-context replay while achieving higher accuracy. Synthius-Mem achieves state-of-the-art results on LoCoMo and is, to our knowledge, the only persona memory system that both exceeds human-level performance and reports adversarial robustness.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。