Jaemin Yoo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jaemin Yoo has a total of 136 co-authors. Here are the top co-authors by number of shared publications:
- Kijung Shin
KAIST, Seoul, Republic of Korea
10 shared publications - Christos Faloutsos
Carnegie Mellon University
4 shared publications - Geon Lee
KAIST, Seoul, Republic of Korea
4 shared publications Soo Yong Lee
Kim Jaechul Graduate School of AI KAIST Seoul South Korea
4 shared publicationsSunwoo Kim
Kim Jaechul Graduate School of AI KAIST Seoul South Korea
4 shared publications- Kyungho Kim
KAIST, Seoul, Republic of Korea
3 shared publications
Latest Publications
- Open Access
- 6 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 4452-4463
RouteRec: Behavior-Guided Sparse Routing for Sequential Recommendation
Sessionized interaction histories contain behavioral patterns that can improve sequential recommendation. However, existing models process all sessions through the same parameterized blocks, regardless of their behavioral differences. Mixture of Experts (MoE) enables conditional computation, but it leaves open what should guide expert allocation. We propose RouteRec, a sequential recommender that uses observed session behavior as the routing criterion. RouteRec summarizes four types of behavioral evidence from sessionized histories: interaction tempo, item-group focus, repetition and carryover, and popularity tendency. It uses these cues to route computation at macro, mid, and micro scopes. Cue-derived scores first select expert groups; within each selected group, the current backbone state then refines expert selection. Across six public datasets and 18 dataset--metric combinations, RouteRec ranks first in 12 and second in three, yielding the best overall average rank of 1.61 compared with 4.11 for the next-best baseline. Additional analyses suggest that the behavioral cues guide expert allocation beyond added capacity and produce routing patterns aligned with observed behavior. Our code is available at https://github.com/jy1559/RouteRec.
- 6 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 4452-4463
Open AccessRouteRec: Behavior-Guided Sparse Routing for Sequential Recommendation
Sessionized interaction histories contain behavioral patterns that can improve sequential recommendation. However, existing models process all sessions through the same parameterized blocks, regardless of their behavioral differences. Mixture of Experts (MoE) enables conditional computation, but it leaves open what should guide expert allocation. We propose RouteRec, a sequential recommender that uses observed session behavior as the routing criterion. RouteRec summarizes four types of behavioral evidence from sessionized histories: interaction tempo, item-group focus, repetition and carryover, and popularity tendency. It uses these cues to route computation at macro, mid, and micro scopes. Cue-derived scores first select expert groups; within each selected group, the current backbone state then refines expert selection. Across six public datasets and 18 dataset--metric combinations, RouteRec ranks first in 12 and second in three, yielding the best overall average rank of 1.61 compared with 4.11 for the next-best baseline. Additional analyses suggest that the behavioral cues guide expert allocation beyond added capacity and produce routing patterns aligned with observed behavior. Our code is available at https://github.com/jy1559/RouteRec.
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 2461-2472
Node4All: Learning Node Representation Beyond Datasets
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 2461-2472
Open AccessNode4All: Learning Node Representation Beyond Datasets
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 3884-3888
ItemRAG: Item-Based Retrieval-Augmented Generation for LLM-Based Recommendation
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 3884-3888
Open AccessItemRAG: Item-Based Retrieval-Augmented Generation for LLM-Based Recommendation
- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 6009-6020
PULSE: Socially-Aware User Representation Modeling Toward Parameter-Efficient Graph Collaborative Filtering
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 6009-6020
Open AccessPULSE: Socially-Aware User Representation Modeling Toward Parameter-Efficient Graph Collaborative Filtering
- Open Access
Feature-Centric Unsupervised Node Representation Learning Without Homophily Assumption
Unsupervised node representation learning aims to obtain meaningful node embeddings without relying on node labels. To achieve this, graph convolution, which aggregates information from neighboring nodes, is commonly employed to encode node features and graph topology. However, excessive reliance on graph convolution can be suboptimal-especially in non-homophilic graphs-since it may yield unduly similar embeddings for nodes that differ in their features or topological properties. As a result, adjusting the degree of graph convolution usage has been actively explored in supervised learning settings, whereas such approaches remain underexplored in unsupervised scenarios. To tackle this, we propose FUEL, which adaptively learns the adequate degree of graph convolution usage by aiming to enhance intra-class similarity and inter-class separability in the embedding space. Since classes are unknown, FUEL leverages node features to identify node clusters and treats these clusters as proxies for classes. Through extensive experiments using 15 baseline methods and 14 benchmark datasets, we demonstrate the effectiveness of FUEL in downstream tasks, achieving state-of-the-art performance across graphs with diverse levels of homophily.
Open AccessFeature-Centric Unsupervised Node Representation Learning Without Homophily Assumption
Unsupervised node representation learning aims to obtain meaningful node embeddings without relying on node labels. To achieve this, graph convolution, which aggregates information from neighboring nodes, is commonly employed to encode node features and graph topology. However, excessive reliance on graph convolution can be suboptimal-especially in non-homophilic graphs-since it may yield unduly similar embeddings for nodes that differ in their features or topological properties. As a result, adjusting the degree of graph convolution usage has been actively explored in supervised learning settings, whereas such approaches remain underexplored in unsupervised scenarios. To tackle this, we propose FUEL, which adaptively learns the adequate degree of graph convolution usage by aiming to enhance intra-class similarity and inter-class separability in the embedding space. Since classes are unknown, FUEL leverages node features to identify node clusters and treats these clusters as proxies for classes. Through extensive experiments using 15 baseline methods and 14 benchmark datasets, we demonstrate the effectiveness of FUEL in downstream tasks, achieving state-of-the-art performance across graphs with diverse levels of homophily.
- Open Access
Fully Inductive Node Representation Learning via Graph View Transformation
Generalizing a pretrained model to unseen datasets without retraining is an essential step toward a foundation model. However, achieving such cross-dataset, fully inductive inference is difficult in graph-structured data where feature spaces vary widely in both dimensionality and semantics. Any transformation in the feature space can easily violate the inductive applicability to unseen datasets, strictly limiting the design space of a graph model. In this work, we introduce the view space, a novel representational axis in which arbitrary graphs can be naturally encoded in a unified manner. We then propose Graph View Transformation (GVT), a node- and feature-permutation-equivariant mapping in the view space. GVT serves as the building block for Recurrent GVT, a fully inductive model for node representation learning. Pretrained on OGBN-Arxiv and evaluated on 27 node-classification benchmarks, Recurrent GVT outperforms GraphAny, the prior fully inductive graph model, by +8.93% and surpasses 12 individually tuned GNNs by at least +3.30%. These results establish the view space as a principled and effective ground for fully inductive node representation learning.
Open AccessFully Inductive Node Representation Learning via Graph View Transformation
Generalizing a pretrained model to unseen datasets without retraining is an essential step toward a foundation model. However, achieving such cross-dataset, fully inductive inference is difficult in graph-structured data where feature spaces vary widely in both dimensionality and semantics. Any transformation in the feature space can easily violate the inductive applicability to unseen datasets, strictly limiting the design space of a graph model. In this work, we introduce the view space, a novel representational axis in which arbitrary graphs can be naturally encoded in a unified manner. We then propose Graph View Transformation (GVT), a node- and feature-permutation-equivariant mapping in the view space. GVT serves as the building block for Recurrent GVT, a fully inductive model for node representation learning. Pretrained on OGBN-Arxiv and evaluated on 27 node-classification benchmarks, Recurrent GVT outperforms GraphAny, the prior fully inductive graph model, by +8.93% and surpasses 12 individually tuned GNNs by at least +3.30%. These results establish the view space as a principled and effective ground for fully inductive node representation learning.
- Open Access
Simple and Behavior-Driven Augmentation for Recommendation with Rich Collaborative Signals
Contrastive learning (CL) has been widely used for enhancing the performance of graph collaborative filtering (GCF) for personalized recommendation. Since data augmentation plays a crucial role in the success of CL, previous works have designed augmentation methods to remove noisy interactions between users and items in order to generate effective augmented views. However, the ambiguity in defining ''noisiness'' presents a persistent risk of losing core information and generating unreliable data views, while increasing the overall complexity of augmentation. In this paper, we propose Simple Collaborative Augmentation for Recommendation (SCAR), a novel and intuitive augmentation method designed to maximize the effectiveness of CL for GCF. Instead of removing information, SCAR leverages collaborative signals extracted from user-item interactions to generate pseudo-interactions, which are then either added to or used to replace existing interactions. This results in more robust representations while avoiding the pitfalls of overly complex augmentation modules. We conduct experiments on four benchmark datasets and show that SCAR outperforms previous CL-based GCF methods as well as other state-of-the-art self-supervised learning approaches across key evaluation metrics. SCAR exhibits strong robustness across different hyperparameter settings and is particularly effective in sparse data scenarios.
Open AccessSimple and Behavior-Driven Augmentation for Recommendation with Rich Collaborative Signals
Contrastive learning (CL) has been widely used for enhancing the performance of graph collaborative filtering (GCF) for personalized recommendation. Since data augmentation plays a crucial role in the success of CL, previous works have designed augmentation methods to remove noisy interactions between users and items in order to generate effective augmented views. However, the ambiguity in defining ''noisiness'' presents a persistent risk of losing core information and generating unreliable data views, while increasing the overall complexity of augmentation. In this paper, we propose Simple Collaborative Augmentation for Recommendation (SCAR), a novel and intuitive augmentation method designed to maximize the effectiveness of CL for GCF. Instead of removing information, SCAR leverages collaborative signals extracted from user-item interactions to generate pseudo-interactions, which are then either added to or used to replace existing interactions. This results in more robust representations while avoiding the pitfalls of overly complex augmentation modules. We conduct experiments on four benchmark datasets and show that SCAR outperforms previous CL-based GCF methods as well as other state-of-the-art self-supervised learning approaches across key evaluation metrics. SCAR exhibits strong robustness across different hyperparameter settings and is particularly effective in sparse data scenarios.
- Open Access
Parameter-Free Hypergraph Neural Network for Few-Shot Node Classification
Few-shot node classification on hypergraphs requires models that generalize from scarce labels while capturing high-order structures. Existing hypergraph neural networks (HNNs) effectively encode such structures but often suffer from overfitting and scalability issues due to complex, black-box architectures. In this work, we propose ZEN (Zero-Parameter Hypergraph Neural Network), a fully linear and parameter-free model that achieves both expressiveness and efficiency. Built upon a unified formulation of linearized HNNs, ZEN introduces a tractable closed-form solution for the weight matrix and a redundancy-aware propagation scheme to avoid iterative training and to eliminate redundant self information. On 11 real-world hypergraph benchmarks, ZEN consistently outperforms eight baseline models in classification accuracy while achieving up to 696x speedups over the fastest competitor. Moreover, the decision process of ZEN is fully interpretable, providing insights into the characteristic of a dataset. Our code and datasets are fully available at https://github.com/chaewoonbae/ZEN.
Open AccessParameter-Free Hypergraph Neural Network for Few-Shot Node Classification
Few-shot node classification on hypergraphs requires models that generalize from scarce labels while capturing high-order structures. Existing hypergraph neural networks (HNNs) effectively encode such structures but often suffer from overfitting and scalability issues due to complex, black-box architectures. In this work, we propose ZEN (Zero-Parameter Hypergraph Neural Network), a fully linear and parameter-free model that achieves both expressiveness and efficiency. Built upon a unified formulation of linearized HNNs, ZEN introduces a tractable closed-form solution for the weight matrix and a redundancy-aware propagation scheme to avoid iterative training and to eliminate redundant self information. On 11 real-world hypergraph benchmarks, ZEN consistently outperforms eight baseline models in classification accuracy while achieving up to 696x speedups over the fastest competitor. Moreover, the decision process of ZEN is fully interpretable, providing insights into the characteristic of a dataset. Our code and datasets are fully available at https://github.com/chaewoonbae/ZEN.
- Open Access
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 3622-3632
Self-Tuning Self-Supervised Image Anomaly Detection
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 3622-3632
Open AccessSelf-Tuning Self-Supervised Image Anomaly Detection
- Open Access
Aggregation Buffer: Revisiting DropEdge with a New Parameter Block
We revisit DropEdge, a data augmentation technique for GNNs which randomly removes edges to expose diverse graph structures during training. While being a promising approach to effectively reduce overfitting on specific connections in the graph, we observe that its potential performance gain in supervised learning tasks is significantly limited. To understand why, we provide a theoretical analysis showing that the limited performance of DropEdge comes from the fundamental limitation that exists in many GNN architectures. Based on this analysis, we propose Aggregation Buffer, a parameter block specifically designed to improve the robustness of GNNs by addressing the limitation of DropEdge. Our method is compatible with any GNN model, and shows consistent performance improvements on multiple datasets. Moreover, our method effectively addresses well-known problems such as degree bias or structural disparity as a unifying solution. Code and datasets are available at https://github.com/dooho00/agg-buffer.
Open AccessAggregation Buffer: Revisiting DropEdge with a New Parameter Block
We revisit DropEdge, a data augmentation technique for GNNs which randomly removes edges to expose diverse graph structures during training. While being a promising approach to effectively reduce overfitting on specific connections in the graph, we observe that its potential performance gain in supervised learning tasks is significantly limited. To understand why, we provide a theoretical analysis showing that the limited performance of DropEdge comes from the fundamental limitation that exists in many GNN architectures. Based on this analysis, we propose Aggregation Buffer, a parameter block specifically designed to improve the robustness of GNNs by addressing the limitation of DropEdge. Our method is compatible with any GNN model, and shows consistent performance improvements on multiple datasets. Moreover, our method effectively addresses well-known problems such as degree bias or structural disparity as a unifying solution. Code and datasets are available at https://github.com/dooho00/agg-buffer.