Search papers, labs, and topics across Lattice.
This paper introduces SCG-MEM, a novel agent memory architecture that reformulates memory access as schema-constrained generation, addressing the limitations of dense retrieval in distinguishing contextually distinct instances. SCG-MEM maintains a dynamic cognitive schema to constrain LLM decoding, ensuring the generation of only valid memory entry keys and preventing structural hallucinations. Experiments on the LoCoMo benchmark demonstrate that SCG-MEM significantly outperforms retrieval-based baselines across all categories, showcasing its effectiveness in improving agent memory and reasoning.
Retrieval-based memory is out: schema-constrained generation ensures agents recall contextually relevant information without hallucinating memory keys, leading to substantial performance gains.
Constructivist epistemology argues that knowledge is actively constructed rather than passively copied. Despite the generative nature of Large Language Models (LLMs), most existing agent memory systems are still based on dense retrieval. However, dense retrieval heavily relies on semantic overlap or entity matching within sentences. Consequently, embeddings often fail to distinguish instances that are semantically similar but contextually distinct, introducing substantial noise by retrieving context-mismatched entries. Conversely, directly employing open-ended generation for memory access risks"Structural Hallucination"where the model generates memory keys that do not exist in the memory, leading to lookup failures. Inspired by this epistemology, we posit that memory is fundamentally organized by cognitive schemas, and valid recall must be a generative process performed within these schematic structures. To realize this, we propose SCG-MEM, a schema-constrained generative memory architecture. SCG-MEM reformulates memory access as Schema-Constrained Generation. By maintaining a dynamic Cognitive Schema, we strictly constrain LLM decoding to generate only valid memory entry keys, providing a formal guarantee against structural hallucinations. To support long-term adaptation, we model memory updates via assimilation (grounding inputs into existing schemas) and accommodation (expanding schemas with novel concepts). Furthermore, we construct an Associative Graph to enable multi-hop reasoning through activation propagation. Experiments on the LoCoMo benchmark show that SCG-MEM substantially improves performance across all categories over retrieval-based baselines.
