Kibum Kim

Sukwon Yun, Kibum Kim, Kanghoon Yoon et al.

Existing Graph Neural Networks (GNNs) usually assume a balanced situation where both the class distribution and the node degree distribution are balanced. However, in real-world situations, we often encounter cases where a few classes (i.e., head class) dominate other classes (i.e., tail class) as well as in the node degree perspective, and thus naively applying existing GNNs eventually fall short of generalizing to the tail cases. Although recent studies proposed methods to handle long-tail situations on graphs, they only focus on either the class long-tailedness or the degree long-tailedness. In this paper, we propose a novel framework for training GNNs, called Long-Tail Experts for Graphs (LTE4G), which jointly considers the class long-tailedness, and the degree long-tailedness for node classification. The core idea is to assign an expert GNN model to each subset of nodes that are split in a balanced manner considering both the class and degree long-tailedness. After having trained an expert for each balanced subset, we adopt knowledge distillation to obtain two class-wise students, i.e., Head class student and Tail class student, each of which is responsible for classifying nodes in the head classes and tail classes, respectively. We demonstrate that LTE4G outperforms a wide range of state-of-the-art methods in node classification evaluated on both manual and natural imbalanced graphs. The source code of LTE4G can be found at https://github.com/SukwonYun/LTE4G.

Kanghoon Yoon, Kibum Kim, Jinyoung Moon et al.

Recent scene graph generation (SGG) frameworks have focused on learning complex relationships among multiple objects in an image. Thanks to the nature of the message passing neural network (MPNN) that models high-order interactions between objects and their neighboring objects, they are dominant representation learning modules for SGG. However, existing MPNN-based frameworks assume the scene graph as a homogeneous graph, which restricts the context-awareness of visual relations between objects. That is, they overlook the fact that the relations tend to be highly dependent on the objects with which the relations are associated. In this paper, we propose an unbiased heterogeneous scene graph generation (HetSGG) framework that captures relation-aware context using message passing neural networks. We devise a novel message passing layer, called relation-aware message passing neural network (RMP), that aggregates the contextual information of an image considering the predicate type between objects. Our extensive evaluations demonstrate that HetSGG outperforms state-of-the-art methods, especially outperforming on tail predicate classes.

Kibum Kim, Dongmin Hyun, Sukwon Yun et al.

The long-tailed problem is a long-standing challenge in Sequential Recommender Systems (SRS) in which the problem exists in terms of both users and items. While many existing studies address the long-tailed problem in SRS, they only focus on either the user or item perspective. However, we discover that the long-tailed user and item problems exist at the same time, and considering only either one of them leads to sub-optimal performance of the other one. In this paper, we propose a novel framework for SRS, called Mutual Enhancement of Long-Tailed user and item (MELT), that jointly alleviates the long-tailed problem in the perspectives of both users and items. MELT consists of bilateral branches each of which is responsible for long-tailed users and items, respectively, and the branches are trained to mutually enhance each other, which is trained effectively by a curriculum learning-based training. MELT is model-agnostic in that it can be seamlessly integrated with existing SRS models. Extensive experiments on eight datasets demonstrate the benefit of alleviating the long-tailed problems in terms of both users and items even without sacrificing the performance of head users and items, which has not been achieved by existing methods. To the best of our knowledge, MELT is the first work that jointly alleviates the long-tailed user and item problems in SRS.

Kibum Kim, Kanghoon Yoon, Jaehyeong Jeon et al.

Weakly-Supervised Scene Graph Generation (WSSGG) research has recently emerged as an alternative to the fully-supervised approach that heavily relies on costly annotations. In this regard, studies on WSSGG have utilized image captions to obtain unlocalized triplets while primarily focusing on grounding the unlocalized triplets over image regions. However, they have overlooked the two issues involved in the triplet formation process from the captions: 1) Semantic over-simplification issue arises when extracting triplets from captions, where fine-grained predicates in captions are undesirably converted into coarse-grained predicates, resulting in a long-tailed predicate distribution, and 2) Low-density scene graph issue arises when aligning the triplets in the caption with entity/predicate classes of interest, where many triplets are discarded and not used in training, leading to insufficient supervision. To tackle the two issues, we propose a new approach, i.e., Large Language Model for weakly-supervised SGG (LLM4SGG), where we mitigate the two issues by leveraging the LLM's in-depth understanding of language and reasoning ability during the extraction of triplets from captions and alignment of entity/predicate classes with target data. To further engage the LLM in these processes, we adopt the idea of Chain-of-Thought and the in-context few-shot learning strategy. To validate the effectiveness of LLM4SGG, we conduct extensive experiments on Visual Genome and GQA datasets, showing significant improvements in both Recall@K and mean Recall@K compared to the state-of-the-art WSSGG methods. A further appeal is that LLM4SGG is data-efficient, enabling effective model training with a small amount of training images.

Jaehyeong Jeon, Kibum Kim, Kanghoon Yoon et al.

The scene graph generation (SGG) task involves detecting objects within an image and predicting predicates that represent the relationships between the objects. However, in SGG benchmark datasets, each subject-object pair is annotated with a single predicate even though a single predicate may exhibit diverse semantics (i.e., semantic diversity), existing SGG models are trained to predict the one and only predicate for each pair. This in turn results in the SGG models to overlook the semantic diversity that may exist in a predicate, thus leading to biased predictions. In this paper, we propose a novel model-agnostic Semantic Diversity-aware Prototype-based Learning (DPL) framework that enables unbiased predictions based on the understanding of the semantic diversity of predicates. Specifically, DPL learns the regions in the semantic space covered by each predicate to distinguish among the various different semantics that a single predicate can represent. Extensive experiments demonstrate that our proposed model-agnostic DPL framework brings significant performance improvement on existing SGG models, and also effectively understands the semantic diversity of predicates.

Kibum Kim, Jiwan Kim, Kyle Min et al.

Video Large Language Models (Video LLMs) incur high inference latency due to a large number of visual tokens provided to LLMs. To address this, training-free visual token pruning has emerged as a solution to reduce computational costs; however, existing methods are primarily validated on Multiple-Choice Question Answering (MCQA) benchmarks, where coarse-grained cues often suffice. In this work, we reveal that these methods suffer a sharp performance collapse on fine-grained understanding tasks requiring precise visual grounding, such as hallucination evaluation. To explore this gap, we conduct a systematic analysis and identify sink tokens--semantically uninformative tokens that attract excessive attention--as a key obstacle to fine-grained video understanding. When these sink tokens survive pruning, they distort the model's visual evidence and hinder fine-grained understanding. Motivated by these insights, we propose Sink-Token-aware Pruning (SToP), a simple yet effective plug-and-play method that introduces a sink score to quantify each token's tendency to behave as a sink and applies this score to existing spatial and temporal pruning methods to suppress them, thereby enhancing video understanding. To validate the effectiveness of SToP, we apply it to state-of-the-art pruning methods (VisionZip, FastVid, and Holitom) and evaluate it across diverse benchmarks covering hallucination, open-ended generation, compositional reasoning, and MCQA. Our results demonstrate that SToP significantly boosts performance, even when pruning up to 90% of visual tokens.

Kanghoon Yoon, Yeonjun In, Namkyeong Lee et al.

Graph neural networks (GNN) are vulnerable to adversarial attacks, which aim to degrade the performance of GNNs through imperceptible changes on the graph. However, we find that in fact the prevalent meta-gradient-based attacks, which utilizes the gradient of the loss w.r.t the adjacency matrix, are biased towards training nodes. That is, their meta-gradient is determined by a training procedure of the surrogate model, which is solely trained on the training nodes. This bias manifests as an uneven perturbation, connecting two nodes when at least one of them is a labeled node, i.e., training node, while it is unlikely to connect two unlabeled nodes. However, these biased attack approaches are sub-optimal as they do not consider flipping edges between two unlabeled nodes at all. This means that they miss the potential attacked edges between unlabeled nodes that significantly alter the representation of a node. In this paper, we investigate the meta-gradients to uncover the root cause of the uneven perturbations of existing attacks. Based on our analysis, we propose a Meta-gradient-based attack method using contrastive surrogate objective (Metacon), which alleviates the bias in meta-gradient using a new surrogate loss. We conduct extensive experiments to show that Metacon outperforms existing meta gradient-based attack methods through benchmark datasets, while showing that alleviating the bias towards training nodes is effective in attacking the graph structure.

Jiwan Kim, Kibum Kim, Wonjoong Kim et al.

Recently, visual token pruning has been studied to handle the vast number of visual tokens in Multimodal Large Language Models. However, we observe that while existing pruning methods perform reliably on simple visual understanding, they struggle to effectively generalize to complex visual reasoning tasks, a critical gap underexplored in previous studies. Through a systematic analysis, we identify Relevant Visual Information Shift (RVIS) during decoding as the primary failure driver. To address this, we propose Decoding-stage Shift-aware Token Pruning (DSTP), a training-free add-on framework that enables existing pruning methods to align visual tokens with shifting reasoning requirements during the decoding stage. Extensive experiments demonstrate that DSTP significantly mitigates performance degradation of pruning methods in complex reasoning tasks, while consistently yielding performance gains even across visual understanding benchmarks. Furthermore, DSTP demonstrates effectiveness across diverse state-of-the-art architectures, highlighting its generalizability and efficiency with minimal computational overhead.

Yeonjun In, Kanghoon Yoon, Kibum Kim et al.

Recent studies have revealed that GNNs are vulnerable to adversarial attacks. To defend against such attacks, robust graph structure refinement (GSR) methods aim at minimizing the effect of adversarial edges based on node features, graph structure, or external information. However, we have discovered that existing GSR methods are limited by narrowassumptions, such as assuming clean node features, moderate structural attacks, and the availability of external clean graphs, resulting in the restricted applicability in real-world scenarios. In this paper, we propose a self-guided GSR framework (SG-GSR), which utilizes a clean sub-graph found within the given attacked graph itself. Furthermore, we propose a novel graph augmentation and a group-training strategy to handle the two technical challenges in the clean sub-graph extraction: 1) loss of structural information, and 2) imbalanced node degree distribution. Extensive experiments demonstrate the effectiveness of SG-GSR under various scenarios including non-targeted attacks, targeted attacks, feature attacks, e-commerce fraud, and noisy node labels. Our code is available at https://github.com/yeonjun-in/torch-SG-GSR.

Yeonjun In, Wonjoong Kim, Kanghoon Yoon et al.

As the use of large language model (LLM) agents continues to grow, their safety vulnerabilities have become increasingly evident. Extensive benchmarks evaluate various aspects of LLM safety by defining the safety relying heavily on general standards, overlooking user-specific standards. However, safety standards for LLM may vary based on a user-specific profiles rather than being universally consistent across all users. This raises a critical research question: Do LLM agents act safely when considering user-specific safety standards? Despite its importance for safe LLM use, no benchmark datasets currently exist to evaluate the user-specific safety of LLMs. To address this gap, we introduce U-SafeBench, a benchmark designed to assess user-specific aspect of LLM safety. Our evaluation of 20 widely used LLMs reveals current LLMs fail to act safely when considering user-specific safety standards, marking a new discovery in this field. To address this vulnerability, we propose a simple remedy based on chain-of-thought, demonstrating its effectiveness in improving user-specific safety. Our benchmark and code are available at https://github.com/yeonjun-in/U-SafeBench.

Sein Kim, Hongseok Kang, Kibum Kim et al.

Large Language Models (LLMs) have recently emerged as promising tools for recommendation thanks to their advanced textual understanding ability and context-awareness. Despite the current practice of training and evaluating LLM-based recommendation (LLM4Rec) models under a sequential recommendation scenario, we found that whether these models understand the sequential information inherent in users' item interaction sequences has been largely overlooked. In this paper, we first demonstrate through a series of experiments that existing LLM4Rec models do not fully capture sequential information both during training and inference. Then, we propose a simple yet effective LLM-based sequential recommender, called LLM-SRec, a method that enhances the integration of sequential information into LLMs by distilling the user representations extracted from a pre-trained CF-SRec model into LLMs. Our extensive experiments show that LLM-SRec enhances LLMs' ability to understand users' item interaction sequences, ultimately leading to improved recommendation performance. Furthermore, unlike existing LLM4Rec models that require fine-tuning of LLMs, LLM-SRec achieves state-of-the-art performance by training only a few lightweight MLPs, highlighting its practicality in real-world applications. Our code is available at https://github.com/Sein-Kim/LLM-SRec.

Jiwan Kim, Hongseok Kang, Sein Kim et al.

Multi-modal recommender systems (MRSs) have achieved notable success in improving personalization by leveraging diverse modalities such as images, text, and audio. However, two key challenges remain insufficiently addressed: (1) Insufficient consideration of missing modality scenarios and (2) the overlooking of unique characteristics of modality features. These challenges result in significant performance degradation in realistic situations where modalities are missing. To address these issues, we propose Disentangling and Generating Modality Recommender (DGMRec), a novel framework tailored for missing modality scenarios. DGMRec disentangles modality features into general and specific modality features from an information-based perspective, enabling richer representations for recommendation. Building on this, it generates missing modality features by integrating aligned features from other modalities and leveraging user modality preferences. Extensive experiments show that DGMRec consistently outperforms state-of-the-art MRSs in challenging scenarios, including missing modalities and new item settings as well as diverse missing ratios and varying levels of missing modalities. Moreover, DGMRec's generation-based approach enables cross-modal retrieval, a task inapplicable for existing MRSs, highlighting its adaptability and potential for real-world applications. Our code is available at https://github.com/ptkjw1997/DGMRec.

Yeonjun In, Kanghoon Yoon, Sukwon Yun et al.

In real-world applications, node features in graphs often contain noise from various sources, leading to significant performance degradation in GNNs. Although several methods have been developed to enhance robustness, they rely on the unrealistic assumption that noise in node features is independent of the graph structure and node labels, thereby limiting their applicability. To this end, we introduce a more realistic noise scenario, dependency-aware noise on graphs (DANG), where noise in node features create a chain of noise dependencies that propagates to the graph structure and node labels. We propose a novel robust GNN, DA-GNN, which captures the causal relationships among variables in the data generating process (DGP) of DANG using variational inference. In addition, we present new benchmark datasets that simulate DANG in real-world applications, enabling more practical research on robust GNNs. Extensive experiments demonstrate that DA-GNN consistently outperforms existing baselines across various noise scenarios, including both DANG and conventional noise models commonly considered in this field. Our code is available at https://github.com/yeonjun-in/torch-DA-GNN.

Kibum Kim, Kanghoon Yoon, Yeonjun In et al.

Existing Video Scene Graph Generation (VidSGG) studies are trained in a fully supervised manner, which requires all frames in a video to be annotated, thereby incurring high annotation cost compared to Image Scene Graph Generation (ImgSGG). Although the annotation cost of VidSGG can be alleviated by adopting a weakly supervised approach commonly used for ImgSGG (WS-ImgSGG) that uses image captions, there are two key reasons that hinder such a naive adoption: 1) Temporality within video captions, i.e., unlike image captions, video captions include temporal markers (e.g., before, while, then, after) that indicate time related details, and 2) Variability in action duration, i.e., unlike human actions in image captions, human actions in video captions unfold over varying duration. To address these issues, we propose a Natural Language-based Video Scene Graph Generation (NL-VSGG) framework that only utilizes the readily available video captions for training a VidSGG model. NL-VSGG consists of two key modules: Temporality-aware Caption Segmentation (TCS) module and Action Duration Variability-aware caption-frame alignment (ADV) module. Specifically, TCS segments the video captions into multiple sentences in a temporal order based on a Large Language Model (LLM), and ADV aligns each segmented sentence with appropriate frames considering the variability in action duration. Our approach leads to a significant enhancement in performance compared to simply applying the WS-ImgSGG pipeline to VidSGG on the Action Genome dataset. As a further benefit of utilizing the video captions as weak supervision, we show that the VidSGG model trained by NL-VSGG is able to predict a broader range of action classes that are not included in the training data, which makes our framework practical in reality.

Kanghoon Yoon, Kibum Kim, Jaehyung Jeon et al.

Scene Graph Generation (SGG) research has suffered from two fundamental challenges: the long-tailed predicate distribution and semantic ambiguity between predicates. These challenges lead to a bias towards head predicates in SGG models, favoring dominant general predicates while overlooking fine-grained predicates. In this paper, we address the challenges of SGG by framing it as multi-label classification problem with partial annotation, where relevant labels of fine-grained predicates are missing. Under the new frame, we propose Retrieval-Augmented Scene Graph Generation (RA-SGG), which identifies potential instances to be multi-labeled and enriches the single-label with multi-labels that are semantically similar to the original label by retrieving relevant samples from our established memory bank. Based on augmented relations (i.e., discovered multi-labels), we apply multi-prototype learning to train our SGG model. Several comprehensive experiments have demonstrated that RA-SGG outperforms state-of-the-art baselines by up to 3.6% on VG and 5.9% on GQA, particularly in terms of F@K, showing that RA-SGG effectively alleviates the issue of biased prediction caused by the long-tailed distribution and semantic ambiguity of predicates.

Jiwan Kim, Kibum Kim, Sangwoo Seo et al.

Recently, efficient Multimodal Large Language Models (MLLMs) have gained significant attention as a solution to their high computational complexity, making them more practical for real-world applications. In this regard, the knowledge distillation (KD) approach has emerged as a promising alternative, which transfers the rich visual and linguistic knowledge from a larger model (teacher) to a smaller model (student). However, we observe that existing KD methods struggle to effectively distill the teacher MLLM's rich visual perception abilities to the student, a challenge that has been largely overlooked in previous studies. Through a systematic analysis, we identify visual attention misalignment between student and teacher as the main cause of this issue. Based on this insight, we propose CompoDistill, a novel KD framework that explicitly aligns the student's visual attention with that of the teacher to enhance the student's visual perception abilities. Our extensive experiments show that CompoDistill significantly improves performance on compositional reasoning tasks that require visual perception abilities while maintaining strong performance on visual question answering tasks, as done in existing studies. Furthermore, CompoDistill demonstrates effectiveness with a more advanced backbone, highlighting its generalizability.

Heewoong Noh, Namkyeong Lee, Gyoung S. Na et al.

Spectral analysis provides crucial clues for the elucidation of unknown materials. Among various techniques, infrared spectroscopy (IR) plays an important role in laboratory settings due to its high accessibility and low cost. However, existing approaches often fail to reflect expert analytical processes and lack flexibility in incorporating diverse types of chemical knowledge, which is essential in real-world analytical scenarios. In this paper, we propose IR-Agent, a novel multi-agent framework for molecular structure elucidation from IR spectra. The framework is designed to emulate expert-driven IR analysis procedures and is inherently extensible. Each agent specializes in a specific aspect of IR interpretation, and their complementary roles enable integrated reasoning, thereby improving the overall accuracy of structure elucidation. Through extensive experiments, we demonstrate that IR-Agent not only improves baseline performance on experimental IR spectra but also shows strong adaptability to various forms of chemical information.

Kibum Kim, Sein Kim, Hongseok Kang et al.

Large Language Models (LLMs) have recently emerged as a powerful backbone for recommender systems. Existing LLM-based recommender systems take two different approaches for representing items in natural language, i.e., Attribute-based Representation and Description-based Representation. In this work, we aim to address the trade-off between efficiency and effectiveness that these two approaches encounter, when representing items consumed by users. Based on our interesting observation that there is a significant information overlap between images and descriptions associated with items, we propose a novel method, Image is all you need for LLM-based Recommender system (I-LLMRec). Our main idea is to leverage images as an alternative to lengthy textual descriptions for representing items, aiming at reducing token usage while preserving the rich semantic information of item descriptions. Through extensive experiments, we demonstrate that I-LLMRec outperforms existing methods in both efficiency and effectiveness by leveraging images. Moreover, a further appeal of I-LLMRec is its ability to reduce sensitivity to noise in descriptions, leading to more robust recommendations.

Kibum Kim, Kanghoon Yoon, Yeonjun In et al.

Scene graph generation (SGG) models have suffered from inherent problems regarding the benchmark datasets such as the long-tailed predicate distribution and missing annotation problems. In this work, we aim to alleviate the long-tailed problem of SGG by utilizing unannotated triplets. To this end, we introduce a Self-Training framework for SGG (ST-SGG) that assigns pseudo-labels for unannotated triplets based on which the SGG models are trained. While there has been significant progress in self-training for image recognition, designing a self-training framework for the SGG task is more challenging due to its inherent nature such as the semantic ambiguity and the long-tailed distribution of predicate classes. Hence, we propose a novel pseudo-labeling technique for SGG, called Class-specific Adaptive Thresholding with Momentum (CATM), which is a model-agnostic framework that can be applied to any existing SGG models. Furthermore, we devise a graph structure learner (GSL) that is beneficial when adopting our proposed self-training framework to the state-of-the-art message-passing neural network (MPNN)-based SGG models. Our extensive experiments verify the effectiveness of ST-SGG on various SGG models, particularly in enhancing the performance on fine-grained predicate classes.

Hojun Aan, Sangsun Han, Hyeonkyu Kim et al.

Many researchers have developed VR systems for people with visual impairments by using various audio feedback techniques. However, there has been much less study of collaborative VR systems in which people with visual impairments and people with able-body can participate together. Therefore, we developed a VR showdown game which is similar to a real Showdown game in which two players can play together in the same virtual environment. We incorporate auditory distance perception using the HRTF (Head Related Transform Function) based on a spatial position in VR. We developed two modes in the showdown game. One is the PVA (Player vs. Agent) mode in which people with visual impairments can play alone and the PVP (Player vs. Player) mode in which people with visual impairments can play with another player in the network environment. We conducted our user studies by comparing the performances of people with visual impairments and people with able-body. The user study results show that people with visual impairments won 67.6% of the games when competing against people with able-body. This paper reports an example of a collaborative VR system for people with visual impairments and also design guideline for developing VR systems for people with visual impairments.