From Volume to Value: Preference-Aligned Memory Construction for On-Device RAG
With the rapid emergence of personal AI agents based on Large Language Models (LLMs), implementing them on-device has become essential for privacy and responsiveness. To handle the inherently personal and context-dependent nature of real-world requests, such agents must ground their generation in device-resident personal context. However, under tight memory budgets, the core bottleneck is what to store so that retrieval remains aligned with the user. We propose EPIC (Efficient Preference-aligned
Record details
Published: 18 May 2026
Source: arXiv
Category: Research
Topics: Privacy · Agents & autonomy
Retrieved: 14 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
It Takes Two: Complementary Self-Distillation for Contextual Integrity in LLMs
arXiv · 18 May 2026
ReacTOD: Bounded Neuro-Symbolic Agentic NLU for Zero-Shot Dialogue State Tracking
arXiv · 18 May 2026
POLAR-Bench: A Diagnostic Benchmark for Privacy-Utility Trade-offs in LLM Agents
arXiv · 18 May 2026
AI Agents May Always Fall for Prompt Injections
arXiv · 17 May 2026
Synthesis and Evaluation of Long-term History-aware Medical Dialogue
arXiv · 19 May 2026
Towards trustworthy agentic AI: a comprehensive survey of safety, robustness, privacy, and system security
arXiv · 17 May 2026
How to cite this record
ethics.ai (18 May 2026), “From Volume to Value: Preference-Aligned Memory Construction for On-Device RAG,” evidence record 4125, https://ethics.ai/record/4125 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.