28 articles tagged with #memory-architecture. AI-curated summaries with sentiment analysis and key takeaways from 50+ sources.
Researchers introduce HORMA, a hierarchical memory system for LLM agents that organizes experience into structured hierarchies with linked summaries and raw trajectories. The system achieves 22% token efficiency on long tasks while maintaining performance, addressing critical limitations in how language model agents manage working memory for multi-step reasoning.
Researchers introduce Continual Learning Bench (CL-Bench), the first comprehensive benchmark for evaluating whether LLM-based AI systems genuinely improve through sequential experience across real-world domains. Testing frontier models reveals significant gaps in current continual learning capabilities, with systems frequently overfitting to immediate observations and failing to reuse knowledge effectively.
Researchers introduce MemGuard, a framework that addresses memory contamination in long-term memory-augmented large language models by organizing memories into functional types and selectively retrieving only relevant evidence. The approach improves hallucination reduction by up to 28.27% while reducing memory token usage by 5.8x, advancing the reliability of AI systems that maintain persistent memory across extended interactions.
Researchers present a biologically-inspired memory architecture for LLM agents that addresses persistent memory management across long interaction horizons. The system incorporates six cognitive mechanisms including sleep-phase consolidation and interference-based forgetting, achieving 97.2% retention precision with 58% storage reduction on a VSCode dataset and matching retrieval accuracy on streaming evaluations.
Researchers introduce MAGIC-Video, a training-free framework that enables multimodal AI systems to process and reason about ultra-long videos spanning days or weeks by combining a structured memory graph with narrative chains. The system outperforms existing baselines on multiple benchmarks, addressing a critical limitation where current LLMs can only handle tens of minutes of video despite having million-token context windows.
Researchers propose E-mem, a new framework for LLM agent memory that reconstructs episodic context instead of compressing it, enabling more rigorous reasoning over extended tasks. The approach uses multiple assistant agents managing uncompressed memory while a master agent coordinates planning, achieving 54% F1 on benchmarks with 70% lower token costs than existing methods.
Researchers introduce soul.py, an open-source architecture addressing catastrophic forgetting in AI agents by distributing identity across multiple memory systems rather than centralizing it. The framework implements persistent identity through separable components and a hybrid RAG+RLM retrieval system, drawing inspiration from how human memory survives neurological damage.
Researchers present Synthius-Mem, a brain-inspired AI memory system that achieves 94.4% accuracy on the LoCoMo benchmark while maintaining 99.6% adversarial robustness—preventing hallucinations about facts users never shared. The system outperforms existing approaches by structuring persona extraction across six cognitive domains rather than treating memory as raw dialogue retrieval, reducing token consumption by 5x.
Researchers propose VaCoAl, a hyperdimensional computing architecture that combines sparse distributed memory with Galois-field algebra to address limitations in modern AI systems like catastrophic forgetting and the binding problem. The deterministic system demonstrates emergent properties equivalent to spike-timing-dependent plasticity and achieves multi-hop reasoning across 25.5M paths in knowledge graphs, positioning it as a complementary third paradigm to large language models.
Researchers propose MGA (Memory-Driven GUI Agent), a minimalist AI framework that improves GUI automation by decoupling long-horizon tasks into independent steps linked through structured state memory. The approach addresses critical limitations in current multimodal AI agents—context overload and architectural redundancy—while maintaining competitive performance with reduced complexity.
Researchers introduce LLMA-Mem, a memory framework for LLM multi-agent systems that balances team size with lifelong learning capabilities. The study reveals that larger agent teams don't always perform better long-term, and smaller teams with better memory design can outperform larger ones while reducing costs.
Researchers introduce D-MEM, a biologically-inspired memory architecture for AI agents that uses dopamine-like reward prediction error routing to dramatically reduce computational costs. The system reduces token consumption by over 80% and eliminates quadratic scaling bottlenecks by selectively processing only high-importance information through cognitive restructuring.
Researchers developed HyMEM, a brain-inspired hybrid memory system that significantly improves GUI agents' ability to interact with computers. The system uses graph-based structured memory combining symbolic nodes with trajectory embeddings, enabling smaller 7B/8B models to match or exceed performance of larger closed-source models like GPT-4o.
Researchers developed KernelSkill, a multi-agent framework that optimizes GPU kernel performance using expert knowledge rather than trial-and-error approaches. The system achieved 100% success rates and significant speedups (1.92x to 5.44x) over existing methods, addressing a critical bottleneck in AI system efficiency.
Researchers propose a new 'Memory-as-Ontology' paradigm for AI agents that treats memory as the foundation of digital existence rather than just a functional tool. The approach introduces Animesis, a Constitutional Memory Architecture designed for persistent digital citizens whose identities must survive across model transitions and extended lifecycles.
Researchers introduce REVERIEMEM, a three-layer memory architecture that enables large language model-based character agents to maintain perspective-bounded knowledge and distinct personalities when roleplaying in book-based narratives. The system addresses key limitations in current LLM roleplay systems by preventing characters from accessing facts outside their perspective and eliminating flattened, monotonous characterization.
Researchers introduce Membox, a hierarchical memory architecture for LLM agents that organizes dialogue history by topic continuity rather than semantic proximity. The system uses Topic Loom to group related turns and Trace Weaver to link events across sessions, achieving 13-19 percentage point F1 improvements over existing memory systems like Mem0 and A-MEM.
Researchers demonstrate that spatial memory systems for language agents must fundamentally separate memory recall from visibility computation, using occlusion testing as a validation method. The study shows that geometry-based weighting outperforms traditional blending approaches, and introduces a ray-casting technique to properly handle occluded spatial information.
ConMem introduces a training-free framework for multi-agent systems that uses structured memory cards and relation-aware graphs to improve adaptation without additional training. The approach reduces inference overhead by over 80% and prunes more than 50% of candidate expansions while maintaining performance across multiple benchmarks.
Researchers introduce MARDoc, a Memory-Aware Refinement Agent framework that improves multimodal long-document question answering by decoupling the task into three specialized agents (Explorer, Refiner, Reflector) that maintain structured memory instead of accumulated interaction history. The approach reduces context noise while preserving critical evidence, outperforming baseline systems on benchmark datasets.
Researchers introduce SaliMory, a framework that trains language models to manage structured memory for conversational AI agents through hierarchical reward processes and contrastive refinement. The approach reduces memory-related failures by one-third and achieves over 10% improvement in accuracy while doubling personalization rates.
Researchers introduce SegTreeMem, a novel memory architecture for long-horizon conversational AI agents that organizes conversation history using temporally-ordered segment trees instead of purely semantic similarity. The system demonstrates improved performance across multiple benchmarks by preserving chronological order while enabling hierarchical retrieval, with ablation studies confirming that temporal sequencing is critical to the approach's effectiveness.
Researchers introduce DeltaMem, a novel memory framework for LLM-based agents that organizes experiences into residual trees to reduce redundancy and improve decision-making. The system stores task skills and environmental knowledge separately, using delta nodes to capture incremental variations of core experiences, with automatic consolidation mechanisms enabling self-organization.
Researchers introduce VitaBench 2.0, a new benchmark for evaluating how well large language models can act as personalized and proactive agents during extended user interactions. The benchmark reveals that current state-of-the-art models struggle significantly with real-world personalization tasks, exposing a substantial gap between current AI capabilities and practical requirements for long-term user collaboration.
Researchers propose a Multi-Memory Segment System (MMS) that improves how AI agents generate and store long-term memories by moving beyond simple summarization. The system creates structured retrieval and contextual memory units inspired by cognitive psychology, enabling more effective historical data utilization and response quality in agent interactions.
