arXiv subjects

Yusuke Matsui

Publications and source records attributed to Yusuke Matsui.

At least 19 recordsLinked to original sources

arXiv2026

Poisoning Attacks on the PGM-index

The PGM-index (Ferragina and Vinciguerra, VLDB'20) is one of the most practical learned indexes, owing to its theoretical elegance and consistently strong empirical performance. It is built on optimal piecewise linear approximations (PLAs) that minimize the number of segments. In this paper, we ask how sensitive this optimal PLA itself is to poisoning attacks. We propose PGM-attack, an efficient poisoning attack that sequentially inserts adversarial keys to inflate the resulting number of segments, and we develop a method for deriving theoretical upper bounds on the number of segments attainable under arbitrary insertions. Our experiments show that poisoning only 10% of the keys allows PGM-attack to increase the segment count by up to 120x. On every evaluated instance, our instance-dependent upper bound is at most 1.92x the segment count attained by PGM-attack, certifying that PGM-attack achieves at least 52% of the optimum. This increase in the number of segments enlarges the PGM-index by up to 120x. Moreover, the attack also transfers to other learned indexes, substantially inflating the index size of PLA-based ones in particular. Our results reveal that, despite the optimality of its PLAs, the PGM-index has an intrinsic vulnerability rooted in its optimization objective, motivating robustness-aware objective design for future learned indexes. Our code is publicly available at https://github.com/atsukisato/pgm-attack.

cs.DB↗

arXiv2026

Fast Construction of Learned Count-Min Sketch via Ternary Search

The Learned Count-Min Sketch (LCMS) is a learned data structure that estimates element frequencies in a multiset and has been experimentally shown to outperform classical data structures in the capacity-accuracy trade-off. However, its performance depends heavily on parameter selection. Because systematic optimization has not been adequately discussed, previous approaches relied on inefficient brute-force methods. In this study, we propose a method to rapidly optimize the parameters of the original LCMS. We experimentally confirmed that when the machine learning model performs well enough, using a single hash function is sufficient to optimize the weighted error metric. Based on this, we introduce a ternary search approach to efficiently find the optimal proportion of Unique Buckets. Our method achieves the same performance as brute-force approaches while speeding up parameter optimization by $216$-$729$ times when the machine learning model's performance is sufficient. Furthermore, even when the model's performance is suboptimal, our approach still determines appropriate parameters $221$-$740$ times faster than the brute-force approach.

cs.DS↗

arXiv2026

Towards Vision-Free CIR: Attribute-Augmented Scoring and LLM-Based Reranking for Zero-Shot Composed Image Retrieval

Recent work has shown that "Vision-Free'' approaches (representing images as text) can be effective for standard image retrieval tasks. However, it remains unclear whether this paradigm can effectively handle a more complex, multimodal task, Composed Image Retrieval (CIR), due to the inherent information loss in textual descriptions. In this paper, we introduce a Vision-Free CIR framework that addresses this challenge through two key techniques: (1) Attribute-Augmented Hybrid Scoring, which compensates for lost visual details via explicit attribute matching, and (2) LLM-Based Reranking, which verifies semantic consistency of top candidates. Experiments on the open-domain CIRR dataset show that our approach outperforms existing Zero-shot CIR methods (44.04% R@1, +8.79%). On FashionIQ, our results highlight the trade-off between semantic reasoning and fine-grained visual matching. Ablation studies reveal that both attribute-augmented scoring and LLM-Based Reranking consistently improve performance.

cs.CV↗

arXiv2026

PINE: Pruning Boosted Tree Ensembles with Conformal In-Distribution Prediction Equivalence

Tree ensembles are machine learning models with strong predictive performance and interpretability, and remain widely used for tabular data. Standard pruning methods for tree ensembles typically optimize an accuracy-compression trade-off and may change a subset of predictions, potentially compromising decision consistency. Faithful pruning methods address this issue by preserving prediction equivalence over the entire input space, but this requirement leads to lower compression ratios. We propose PINE, a pruning method that provides strong guarantees within an in-distribution region. PINE preserves prediction equivalence within this region and controls the region size using a single parameter $\alpha$ via conformal calibration. Experiments on 12 public tabular datasets show that PINE improves the compression ratio by up to 30% while preserving predictions at a comparable level to existing faithful pruning methods.

cs.LG↗

arXiv2026

CIRCLED: A Multi-turn CIR Dataset with Consistent Dialogues across Domains

Existing Multi-Turn Composed Image Retrieval (MTCIR) datasets lack dialogue-historyconsistency and are restricted to the fashion domain. To address these limitations, we construct CIRCLED by extending FashionIQ, CIRR, and CIRCO. In CIRCLED, the query ateach turn progressively approaches the target image. Data are generated via a CIReVLbased retrieval pipeline and curated with multiple filters on retrieval success, turn length, consistency, and information redundancy to ensure quality. In total, we collect 22,608 multiturn sessions across nine subsets, substantially exceeding Multi-turn FashionIQ (11,505 sessions) in both scale and generality. We further apply multiple baseline methods and quantitatively assess retrieval accuracy on CIRCLED. Our work provides a practical, highquality benchmark to facilitate future research on multi-turn CIR. The dataset is publicly available at https://huggingface.co/datasets/tk1441/CIRCLED, and the code at https://github.com/mti-lab/circled.

cs.CV↗

arXiv2026

Mathematical Foundations of Poisoning Attacks on Linear Regression over Cumulative Distribution Functions

Learned indexes are a class of index data structures that enable fast search by approximating the cumulative distribution function (CDF) using machine learning models (Kraska et al., SIGMOD'18). However, recent studies have shown that learned indexes are vulnerable to poisoning attacks, where injecting a small number of poison keys into the training data can significantly degrade model accuracy and reduce index performance (Kornaropoulos et al., SIGMOD'22). In this work, we provide a rigorous theoretical analysis of poisoning attacks targeting linear regression models over CDFs, one of the most basic regression models and a core component in many learned indexes. Our main contributions are as follows: (i) We present a theoretical proof characterizing the optimal single-point poisoning attack and show that the existing method yields the optimal attack. (ii) We show that in multi-point attacks, the existing greedy approach is not always optimal, and we rigorously derive the key properties that an optimal attack should satisfy. (iii) We propose a method to compute an upper bound of the multi-point poisoning attack's impact and empirically demonstrate that the loss under the greedy approach is often close to this bound. Our study deepens the theoretical understanding of attack strategies against linear regression models on CDFs and provides a foundation for the theoretical evaluation of attacks and defenses on learned indexes.

cs.LG↗

arXiv2025

Optimized Learned Count-Min Sketch

Count-Min Sketch (CMS) is a memory-efficient data structure for estimating the frequency of elements in a multiset. Learned Count-Min Sketch (LCMS) enhances CMS with a machine learning model to reduce estimation error under the same memory usage, but suffers from slow construction due to empirical parameter tuning and lacks theoretical guarantees on intolerable error probability. We propose Optimized Learned Count-Min Sketch (OptLCMS), which partitions the input domain and assigns each partition to its own CMS instance, with CMS parameters analytically derived for fixed thresholds, and thresholds optimized via dynamic programming with approximate feasibility checks. This reduces the need for empirical validation, enabling faster construction while providing theoretical guarantees under these assumptions. OptLCMS also allows explicit control of the allowable error threshold, improving flexibility in practice. Experiments show that OptLCMS builds faster, achieves lower intolerable error probability, and matches the estimation accuracy of LCMS.

cs.LG↗

arXiv2025

How Should We Evaluate Data Deletion in Graph-Based ANN Indexes?

Approximate Nearest Neighbor Search (ANNS) has recently gained significant attention due to its many applications, such as Retrieval-Augmented Generation. Such applications require ANNS algorithms that support dynamic data, so the ANNS problem on dynamic data has attracted considerable interest. However, a comprehensive evaluation methodology for data deletion in ANNS has yet to be established. This study proposes an experimental framework and comprehensive evaluation metrics to assess the efficiency of data deletion for ANNS indexes under practical use cases. Specifically, we categorize data deletion methods in graph-based ANNS into three approaches and formalize them mathematically. The performance is assessed in terms of accuracy, query speed, and other relevant metrics. Finally, we apply the proposed evaluation framework to Hierarchical Navigable Small World, one of the state-of-the-art ANNS methods, to analyze the effects of data deletion, and propose Deletion Control, a method which dynamically selects the appropriate deletion method under a required search accuracy.

cs.LG↗

arXiv2025

RouteExtract: A Modular Pipeline for Extracting Routes from Paper Maps

Paper maps remain widely used for hiking and sightseeing because they contain curated trails and locally relevant annotations that are often missing from digital navigation applications such as Google Maps. We propose a pipeline to extract navigable trails from scanned maps, enabling their use in GPS-based navigation. Our method combines georeferencing, U-Net-based binary segmentation, graph construction, and an iterative refinement procedure using a routing engine. We evaluate the full end-to-end pipeline as well as individual components, showing that the approach can robustly recover trail networks from diverse map styles and generate GPS routes suitable for practical use.

cs.CV↗

arXiv2025

Region-Wise Correspondence Prediction between Manga Line Art Images

Understanding region-wise correspondences between manga line art images is fundamental for high-level manga processing, supporting downstream tasks such as line art colorization and in-between frame generation. Unlike natural images that contain rich visual cues, manga line art consists only of sparse black-and-white strokes, making it challenging to determine which regions correspond across images. In this work, we introduce a new task: predicting region-wise correspondence between raw manga line art images without any annotations. To address this problem, we propose a Transformer-based framework trained on large-scale, automatically generated region correspondences. The model learns to suppress noisy matches and strengthen consistent structural relationships, resulting in robust patch-level feature alignment within and across images. During inference, our method segments each line art and establishes coherent region-level correspondences through edge-aware clustering and region matching. We construct manually annotated benchmarks for evaluation, and experiments across multiple datasets demonstrate both high patch-level accuracy and strong region-level correspondence performance, achieving 78.4-84.4% region-level accuracy. These results highlight the potential of our method for real-world manga and animation applications.

cs.CV↗

arXiv2025

Noisy Label Refinement with Semantically Reliable Synthetic Images

Semantic noise in image classification datasets, where visually similar categories are frequently mislabeled, poses a significant challenge to conventional supervised learning approaches. In this paper, we explore the potential of using synthetic images generated by advanced text-to-image models to address this issue. Although these high-quality synthetic images come with reliable labels, their direct application in training is limited by domain gaps and diversity constraints. Unlike conventional approaches, we propose a novel method that leverages synthetic images as reliable reference points to identify and correct mislabeled samples in noisy datasets. Extensive experiments across multiple benchmark datasets show that our approach significantly improves classification accuracy under various noise conditions, especially in challenging scenarios with semantic label noise. Additionally, since our method is orthogonal to existing noise-robust learning techniques, when combined with state-of-the-art noise-robust training methods, it achieves superior performance, improving accuracy by 30% on CIFAR-10 and by 11% on CIFAR-100 under 70% semantic noise, and by 24% on ImageNet-100 under real-world noise conditions.

cs.CV↗

arXiv2025

PLASMA: A Layout-Aware Benchmark Reveals Memory Layout Matters for Graph-based ANNS on GPU

We propose a $\textbf{P}$latform for $\textbf{L}$ayout-$\textbf{A}$ware $\textbf{S}$earch and $\textbf{M}$emory $\textbf{A}$rrangement ($\textbf{PLASMA}$), a unified evaluation framework for graph-based Approximate Nearest Neighbor Search (ANNS) on GPU that isolates the effects of graph index topology and memory layout. Graph-based ANNS is essential in modern AI applications such as RAG, and GPU utilization is attracting attention for datasets of millions or more vectors. Our framework extracts the topology of arbitrary graph-based indices and enables execution under a unified, GPU-optimized search algorithm, specifying the correspondence between vertex IDs and positions on memory to allow arbitrary vertex orderings. Through comprehensive experiments, we demonstrate that vertex reordering yields up to $80\%$ (typically $10-30\%$) QPS improvement while preserving search accuracy.

cs.IR↗

arXiv2025

LotusFilter: Fast Diverse Nearest Neighbor Search via a Learned Cutoff Table

Approximate nearest neighbor search (ANNS) is an essential building block for applications like RAG but can sometimes yield results that are overly similar to each other. In certain scenarios, search results should be similar to the query and yet diverse. We propose LotusFilter, a post-processing module to diversify ANNS results. We precompute a cutoff table summarizing vectors that are close to each other. During the filtering, LotusFilter greedily looks up the table to delete redundant vectors from the candidates. We demonstrated that the LotusFilter operates fast (0.02 [ms/query]) in settings resembling real-world RAG applications, utilizing features such as OpenAI embeddings. Our code is publicly available at https://github.com/matsui528/lotf.

cs.CV↗

arXiv2025

SVGEditBench V2: A Benchmark for Instruction-based SVG Editing

Vector format has been popular for representing icons and sketches. It has also been famous for design purposes. Regarding image editing, research on vector graphics editing rarely exists in contrast with the raster counterpart. We considered the reason to be the lack of datasets and benchmarks. Thus, we propose SVGEditBench V2, a benchmark dataset for instruction-based SVG editing. SVGEditBench V2 comprises triplets of an original image, a ground truth image, and the editing prompt. We built the dataset by first extracting image pairs from various SVG emoji datasets. Then, we had GPT-4o to create the prompt. We found that triplets gained by this simple pipeline contain varying sorts of editing tasks. Additionally, we performed the editing tasks with existing LLMs and investigated how those current methods can perform SVG editing. Although there were some successful cases, we found that there is a massive room for improvement.

cs.GR↗

arXiv2025

Cascaded Learned Bloom Filter for Optimizing Model-Filter Size Balance and Fast Rejection

Recent studies have demonstrated that learned Bloom filters (LBFs), which combine machine learning with the classical Bloom filter, can achieve superior memory efficiency. However, two challenges remain: (1) jointly optimizing the sizes of the machine learning model and Bloom filters, and (2) systematically minimizing reject time. We propose the Cascaded Learned Bloom Filter (CLBF), a unified architecture that generalizes existing LBF designs, including Sandwiched LBF and Partitioned LBF. Within this framework, we develop a dynamic programming-based optimizer that explores a discretized parameter space and identifies near-optimal configurations that balance model and filter sizes while achieving fast rejection. Experiments on real-world datasets show that CLBF reduces memory usage by up to 17% and decreases reject time by up to a factor of 65 compared to Partitioned LBF, the state-of-the-art LBF in terms of memory efficiency under a fixed machine learning model. Our code is publicly available at https://github.com/atsukisato/CascadedLBF.

cs.DS↗

arXiv2024

FlexFlood: Efficiently Updatable Learned Multi-dimensional Index

A learned multi-dimensional index is a data structure that efficiently answers multi-dimensional orthogonal queries by understanding the data distribution using machine learning models. One of the existing problems is that the search performance significantly decreases when the distribution of data stored in the data structure becomes skewed due to update operations. To overcome this problem, we propose FlexFlood, a flexible variant of Flood. FlexFlood partially reconstructs the internal structure when the data distribution becomes skewed. Moreover, FlexFlood is the first learned multi-dimensional index that guarantees the time complexity of the update operation. Through experiments using both artificial and real-world data, we demonstrate that the search performance when the data distribution becomes skewed is up to 10 times faster than existing methods. We also found that partial reconstruction takes only about twice as much time as naive data updating.

cs.DS↗

arXiv2024

Fast Construction of Partitioned Learned Bloom Filter with Theoretical Guarantees

Bloom filter is a widely used classic data structure for approximate membership queries. Learned Bloom filters improve memory efficiency by leveraging machine learning, with the partitioned learned Bloom filter (PLBF) being among the most memory-efficient variants. However, PLBF suffers from high computational complexity during construction, specifically $O(N^3k)$, where $N$ and $k$ are hyperparameters. In this paper, we propose three methods: fast PLBF, fast PLBF++, and fast PLBF#, that reduce the construction complexity to $O(N^2k)$, $O(Nk \log N)$, and $O(Nk \log k)$, respectively. Fast PLBF preserves the original PLBF structure and memory efficiency. Although fast PLBF++ and fast PLBF# may have different structures, we theoretically prove they are equivalent to PLBF under ideal data distribution. Furthermore, we theoretically bound the difference in memory efficiency between PLBF and fast PLBF++ for non-ideal scenarios. Experiments on real-world datasets demonstrate that fast PLBF, fast PLBF++, and fast PLBF# are up to 233, 761, and 778 times faster to construct than original PLBF, respectively. Additionally, fast PLBF maintains the same data structure as PLBF, and fast PLBF++ and fast PLBF# achieve nearly identical memory efficiency.

cs.DS↗

arXiv2024

Broadcast Product: Redefining Shape-aligned Element-wise Multiplication and Beyond

Broadcast operations are widely used in scientific computing libraries, yet their mathematical formulation is often implicit and inconsistently represented in machine learning literature. This problem frequently leads to invalid equations when element-wise products are written despite mismatched tensor shapes. In this paper, we formalize such operations by introducing the broadcast product $\boxdot$, which explicitly extends the Hadamard product through shape-aligned element duplication. We provide a rigorous definition of the broadcast product, analyze its algebraic properties, and show how it can be expressed using standard linear algebra. Building on this framework, we formulate least-squares problems and sketch a proof-of-concept broadcast decomposition. As a preliminary illustration, we show that the formalism enables a new family of decompositions with distinct structural properties from conventional tensor decompositions. This work establishes a mathematical foundation for broadcast-aware tensor operations, connecting practical implementations with rigorous tensor analysis.

cs.LG↗