Wonseok Choi

Chengxin Liu, Wonseok Choi, Chenshuang Zhang et al.

Vision-Language Models (VLMs) have demonstrated strong capability in a wide range of tasks such as visual recognition, document parsing, and visual grounding. Nevertheless, recent work shows that while VLMs often manage to capture the correct image region corresponding to the question, they do not necessarily produce the correct answers. In this work, we demonstrate that this misalignment could be attributed to suboptimal information flow within VLMs, where text tokens distribute too much attention to irrelevant visual tokens, leading to incorrect answers. Based on the observation, we show that modulating the information flow during inference can improve the perception capability of VLMs. The idea is that text tokens should only be associated with important visual tokens during decoding, eliminating the interference of irrelevant regions. To achieve this, we propose a token dynamics-based method to determine the importance of visual tokens, where visual tokens that exhibit distinct activation patterns during different decoding stages are viewed as important. We apply our approach to representative open-source VLMs and evaluate on various datasets, including visual question answering, visual grounding and counting, optical character recognition, and object hallucination. The results show that our approach significantly improves the performance of baselines. Project page: https://cxliu0.github.io/AIF/.

Moon Ye-Bin, Nam Hyeon-Woo, Wonseok Choi et al.

Vision language models (VLMs) perceive the world through a combination of a visual encoder and a large language model (LLM). The visual encoder, pre-trained on large-scale vision-text datasets, provides zero-shot generalization to visual data, and the LLM endows its high reasoning ability to VLMs. It leads VLMs to achieve high performance on wide benchmarks without fine-tuning, exhibiting zero or few-shot capability. However, recent studies show that VLMs are vulnerable to hallucination. This undesirable behavior degrades reliability and credibility, thereby making users unable to fully trust the output from VLMs. To enhance trustworthiness and better tackle the hallucination of VLMs, we curate a new evaluation dataset, called the BEfore-AFter hallucination dataset (BEAF), and introduce new metrics: True Understanding (TU), IGnorance (IG), StuBbornness (SB), and InDecision (ID). Unlike prior works that focus only on constructing questions and answers, the key idea of our benchmark is to manipulate visual scene information by image editing models and to design the metrics based on scene changes. This allows us to clearly assess whether VLMs correctly understand a given scene by observing the ability to perceive changes. We also visualize image-wise object relationship by virtue of our two-axis view: vision and text. Upon evaluating VLMs with our dataset, we observed that our metrics reveal different aspects of VLM hallucination that have not been reported before. Project page: \url{https://beafbench.github.io/}

Moon Ye-Bin, Nam Hyeon-Woo, Wonseok Choi et al.

Data imbalance in training data often leads to biased predictions from trained models, which in turn causes ethical and social issues. A straightforward solution is to carefully curate training data, but given the enormous scale of modern neural networks, this is prohibitively labor-intensive and thus impractical. Inspired by recent developments in generative models, this paper explores the potential of synthetic data to address the data imbalance problem. To be specific, our method, dubbed SYNAuG, leverages synthetic data to equalize the unbalanced distribution of training data. Our experiments demonstrate that, although a domain gap between real and synthetic data exists, training with SYNAuG followed by fine-tuning with a few real samples allows to achieve impressive performance on diverse tasks with different data imbalance issues, surpassing existing task-specific methods for the same purpose.

Nam Hyeon-Woo, Moon Ye-Bin, Wonseok Choi et al.

Vision language models (VLMs) have shown promising reasoning capabilities across various benchmarks; however, our understanding of their visual perception remains limited. In this work, we propose an eye examination process to investigate how a VLM perceives images, specifically focusing on key elements of visual recognition, from primitive color and shape to semantic levels. To this end, we introduce a dataset named LENS to guide a VLM to follow the examination and check its readiness. Once the model is ready, we conduct the examination. Through this examination, we quantify and visualize VLMs' sensitivities to color and shape, and semantic matching. Our findings reveal that VLMs have varying sensitivity to different colors while consistently showing insensitivity to green across different VLMs. Also, we found different shape sensitivity and semantic recognition depending on LLM's capacity despite using the same fixed visual encoder. Our analyses and findings have potential to inspire the design of VLMs and the pre-processing of visual input to VLMs for improving application performance.

Wonseok Choi, Hyunah Yu, Jongmin Kim et al.

Fully homomorphic encryption (FHE) enables secure computation on encrypted data, mitigating privacy concerns in cloud and edge environments. However, due to its high compute and memory demands, extensive acceleration research has been pursued across diverse hardware platforms, especially GPUs. In this paper, we perform a microarchitectural analysis of CKKS, a popular FHE scheme, on modern GPUs. Focusing on the memory hierarchy, we demonstrate that dominant kernels remain bound by the on-chip L2 cache despite its high bandwidth, exposing a persistent inner memory wall beyond the conventional off-chip DRAM bottleneck. Further, we reveal that the overall CKKS throughput is constrained by low per-kernel hardware utilization, caused by insufficient intra-kernel parallelism. Motivated by these findings, we introduce Theodosian, a set of complementary, memory-aware optimizations that improve cache efficiency and reduce runtime overheads. Theodosian achieves 1.45--1.83x performance improvements over a highly optimized baseline, Cheddar, across representative CKKS workloads. On an RTX 5090, we reduce the bootstrapping latency for 32,768 complex numbers from 22.1ms to 15.2ms, and further to 12.8ms with additional algorithmic optimizations, establishing a new state-of-the-art GPU performance to the best of our knowledge.

Wonduk Seo, Wonseok Choi, Junseo Koh et al.

Large Language Models (LLMs) increasingly support culturally sensitive decision making, yet often exhibit misalignment due to skewed pretraining data and the absence of structured value representations. Existing methods can steer outputs, but often lack demographic grounding and treat values as independent, unstructured signals, reducing consistency and interpretability. We propose OG-MAR, an Ontology-Guided Multi-Agent Reasoning framework. OG-MAR summarizes respondent-specific values from the World Values Survey (WVS) and constructs a global cultural ontology by eliciting relations over a fixed taxonomy via competency questions. At inference time, it retrieves ontology-consistent relations and demographically similar profiles to instantiate multiple value-persona agents, whose outputs are synthesized by a judgment agent that enforces ontology consistency and demographic proximity. Experiments on regional social-survey benchmarks across four LLM backbones show that OG-MAR improves cultural alignment and robustness over competitive baselines, while producing more transparent reasoning traces.

Hankyeol Lee, Gawon Seo, Wonseok Choi et al.

The performance of vision-language models (VLMs), such as CLIP, in visual classification tasks, has been enhanced by leveraging semantic knowledge from large language models (LLMs), including GPT. Recent studies have shown that in zero-shot classification tasks, descriptors incorporating additional cues, high-level concepts, or even random characters often outperform those using only the category name. In many classification tasks, while the top-1 accuracy may be relatively low, the top-5 accuracy is often significantly higher. This gap implies that most misclassifications occur among a few similar classes, highlighting the model's difficulty in distinguishing between classes with subtle differences. To address this challenge, we introduce a novel concept of comparative descriptors. These descriptors emphasize the unique features of a target class against its most similar classes, enhancing differentiation. By generating and integrating these comparative descriptors into the classification framework, we refine the semantic focus and improve classification accuracy. An additional filtering process ensures that these descriptors are closer to the image embeddings in the CLIP space, further enhancing performance. Our approach demonstrates improved accuracy and robustness in visual classification tasks by addressing the specific challenge of subtle inter-class differences.

Hyesung Ji, Hyunah Yu, Jongmin Kim et al.

Private information retrieval (PIR) allows private database queries but is hindered by intense server-side computation and memory traffic. Modern lattice-based PIR protocols typically involve three phases: ExpandQuery (expanding a query into encrypted indices), RowSel (encrypted row selection), and ColTor (recursive "column tournament" for final selection). ExpandQuery and ColTor primarily perform number-theoretic transforms (NTTs), whereas RowSel reduces to large-scale independent matrix-matrix multiplications (GEMMs). GPUs are theoretically ideal for these tasks, provided multi-client batching is used to achieve high throughput. However, batching fundamentally reshapes performance bottlenecks; while it amortizes database access costs, it expands working sets beyond the L2 cache capacity, causing divergent memory behaviors and excessive DRAM traffic. We present GPIR, a GPU-accelerated PIR system that rethinks kernel design, data layout, and execution scheduling. We introduce a stage-aware hybrid execution model that dynamically switches between operation-level kernels, which execute each primitive operation separately, and stage-level kernels, which fuse all operations within a protocol stage into a single kernel to maximize on-chip data reuse. For RowSel, we identify a performance gap caused by a structural mismatch between NTT-driven data layouts and tiled GEMM access patterns, which is exacerbated by multi-client batching. We resolve this through a transposed-layout GEMM design and fine-grained pipelining. Finally, we extend GPIR to multi-GPU systems, scaling both query throughput and database capacity with negligible communication overhead. GPIR achieves up to 305.7x higher throughput than PIRonGPU, the state-of-the-art GPU implementation.

Wonseok Choi, Sohwi Lim, Nam Hyeon-Woo et al.

Instance-level image retrieval aims to find images containing the same object as a given query, despite variations in size, position, or appearance. To address this challenging task, we propose Patchify, a simple yet effective patch-wise retrieval framework that offers high performance, scalability, and interpretability without requiring fine-tuning. Patchify divides each database image into a small number of structured patches and performs retrieval by comparing these local features with a global query descriptor, enabling accurate and spatially grounded matching. To assess not just retrieval accuracy but also spatial correctness, we introduce LocScore, a localization-aware metric that quantifies whether the retrieved region aligns with the target object. This makes LocScore a valuable diagnostic tool for understanding and improving retrieval behavior. We conduct extensive experiments across multiple benchmarks, backbones, and region selection strategies, showing that Patchify outperforms global methods and complements state-of-the-art reranking pipelines. Furthermore, we apply Product Quantization for efficient large-scale retrieval and highlight the importance of using informative features during compression, which significantly boosts performance. Project website: https://wons20k.github.io/PatchwiseRetrieval/

Jusop Choi, Wonseok Choi, William Aiken et al.

Since Bitcoin appeared in 2009, over 6,000 different cryptocurrency projects have followed. The cryptocurrency world may be the only technology where a massive number of competitors offer similar services yet claim unique benefits, including scalability, fast transactions, and security. But are these projects really offering unique features and significant enhancements over their competitors? To answer this question, we conducted a large-scale empirical analysis of code maintenance activities, originality and security across 592 crypto projects. We found that about half of these projects have not been updated for the last six months; over two years, about three-quarters of them disappeared, or were reported as scams or inactive. We also investigated whether 11 security vulnerabilities patched in Bitcoin were also patched in other projects. We found that about 80% of 510 C-language-based cryptocurrency projects have at least one unpatched vulnerability, and the mean time taken to fix the vulnerability is 237.8 days. Among those 510 altcoins, we found that at least 157 altcoins are likely to have been forked from Bitcoin, about a third of them containing only slight changes from the Bitcoin version from which they were forked. As case studies, we did a deep dive into 20 altcoins (e.g., Litecoin, FujiCoin, and Feathercoin) similar to the version of Bitcoin used for the fork. About half of them did not make any technically meaningful change - failing to comply with the promises (e.g., about using Proof of Stake) made in their whitepapers.