Federated Recommendation (FedRec) enables privacy-preserving recommendation by training models across user devices without sharing raw interaction data. However, personalized FedRec still faces a trade-off between personalization and resource efficiency. In this paper, we revisit personalization from the perspective of the local optimization horizon. We find that sufficiently long local optimization can form user-specific item structures, but frequent synchronization or full replacement can disrupt them. Based on this observation, we propose
FedRKG
, a model-agnostic framework that forms personalized item embeddings through
Long-Horizon Local Optimization
and injects common global knowledge through intermittent
Regularized Knowledge Guidance
. This design incorporates shared collaborative signals without replacing local embeddings or maintaining dual representations. We further introduce
Adaptive Guidance
to control the influence of global knowledge at the user–item interaction level. Extensive experiments show that FedRKG consistently outperforms state-of-the-art FedRec baselines while reducing storage and communication costs. Our code is available at
https://github.com/Jaehyung-Lim/FedRKG_JIIS26_SCIE
.
Jaehyung Lim, Wonbin Kweon, Woojoo Kim et al.· Journal of Intelligence and...· 0 citations
Vision-Language-Action (VLA) models provide a versatile foundation for general robotic manipulation, yet they exhibit significant brittleness when confronted with novel task distributions. While In-Context Imitation Learning (ICIL) offers a training-free alternative, existing frameworks suffer from an adaptation bottleneck that hinders the effective translation of expert context to executable actions. This failure originates from superficial retrieval mechanisms and an inherent behavioral inertia that anchors the policy to its pre-trained priors. To address these limitations, we present RA-VLA, a retrieval-augmented VLA framework that integrates behavior-aligned context retrieval with a grounded execution pipeline. By enforcing faithful adherence to functional cues within a scalable architecture, RA-VLA facilitates seamless task adaptation while preserving inference efficiency. Our empirical evaluations across the LIBERO benchmark and a real-world UR5e environment demonstrate that RA-VLA achieves superior success rates and computational efficiency, establishing a robust framework for training-free robotic adaptation.
Sanghwan Jang, Minjin Jeon, Minsoo Kim et al.· 1 citation
Sequential recommenders often struggle with sparse and noisy histories, limiting generalization to unseen interactions. Knowledge distillation mitigates this by transferring dense supervision from a teacher to a student. However, most distillation methods run teacher and student independently, then match student outputs or representations to the teacher. Such supervision entangles student-component effects, blurring whether weak generalization stems from unreliable embeddings, overfitted encoding, or co-adaptation to sparse histories. In this paper, we propose Graft-Oriented Distillation (GOD), a component-level distillation framework for improved generalization through grafting. Grafting denotes replacing selected frozen-teacher components with trainable student counterparts to build hybrid source models. GOD uses these hybrid models to evaluate student embeddings with the teacher encoder and the student encoder with teacher embeddings, providing component-level feedback. At inference, GOD uses only the student, incurring no additional cost. Across three real-world datasets, GOD outperforms state-of-the-art baselines by up to 13.92%.
Woojoo Kim, Junyoung Kim, Jaehyung Lim et al.· 0 citations