arXiv subjects

Akihiro Yoshida

Publications and source records attributed to Akihiro Yoshida.

4 recordsLinked to original sources

arXiv2026

OneComp: One-Line Revolution for Generative AI Model Compression

Deploying foundation models is increasingly constrained by memory footprint, latency, and hardware costs. Post-training compression can mitigate these bottlenecks by reducing the precision of model parameters without significantly degrading performance; however, its practical implementation remains challenging as practitioners navigate a fragmented landscape of quantization algorithms, precision budgets, data-driven calibration strategies, and hardware-dependent execution regimes. We present OneComp, an open-source compression framework that transforms this expert workflow into a reproducible, resource-adaptive pipeline. Given a model identifier and available hardware, OneComp automatically inspects the model, plans mixed-precision assignments, and executes progressive quantization stages, ranging from layer-wise compression to block-wise refinement and global refinement. A key architectural choice is treating the first quantized checkpoint as a deployable pivot, ensuring that each subsequent stage improves the same model and that quality increases as more compute is invested. By converting state-of-the-art compression research into an extensible, open-source, hardware-aware pipeline, OneComp bridges the gap between algorithmic innovation and production-grade model deployment.

cs.LG↗

arXiv2025

Enhancing Quantum-ready QUBO-based Suppression for Object Detection with Appearance and Confidence Features

Quadratic Unconstrained Binary Optimization (QUBO)-based suppression in object detection is known to have superiority to conventional Non-Maximum Suppression (NMS), especially for crowded scenes where NMS possibly suppresses the (partially-) occluded true positives with low confidence scores. Whereas existing QUBO formulations are less likely to miss occluded objects than NMS, there is room for improvement because existing QUBO formulations naively consider confidence scores and pairwise scores based on spatial overlap between predictions. This study proposes new QUBO formulations that aim to distinguish whether the overlap between predictions is due to the occlusion of objects or due to redundancy in prediction, i.e., multiple predictions for a single object. The proposed QUBO formulation integrates two features into the pairwise score of the existing QUBO formulation: i) the appearance feature calculated by the image similarity metric and ii) the product of confidence scores. These features are derived from the hypothesis that redundant predictions share a similar appearance feature and (partially-) occluded objects have low confidence scores, respectively. The proposed methods demonstrate significant advancement over state-of-the-art QUBO-based suppression without a notable increase in runtime, achieving up to 4.54 points improvement in mAP and 9.89 points gain in mAR.

cs.CV↗

arXiv2022

Long-Term Optimal Delivery Planning for Replacing the Liquefied Petroleum Gas Cylinder

In the daily operation of liquefied petroleum gas service, gas providers visit customers and replace cylinders if the gas is about to run out. For a long time, frequent visits to customers were required because they could not determine the amount of remaining gas without a staff visit and observation. To solve this problem, smart meters are started to be employed to acquire gas consumption more frequently without visiting customers. In this study, we construct a system to optimize plans for cylinder replacement, and evaluate it with a large-scale field test. We propose an algorithm to create a replacement plan with three steps: estimating the replacement date, acquiring the customer list for replacement, and determining the delivery route. A more accurate estimation of the replacement date can be acquired with a smart meter, which is used for making a customer list for replacement. The formulation for making a customer list enables the gas provider to replace cylinders some days before the date when the gas would run out. It can suppress the concentration of replacements on certain days. Large-scale verification experiments were performed with more than 1,000 customers in Chiba prefecture in Japan. In the field test, the gas provider incorporated the system into its replacement operations. Moreover, the replacement plans developed by the proposed system were compared with that by the gas provider. Our system reduced the number of gas cylinders with gas shortage, the number of visits without replacement due to plenty of gas remaining, and the working duration per customer, which shows that our system benefits both gas providers and customers.

eess.SY↗

arXiv2020

Nested Subspace Arrangement for Representation of Relational Data

Studies on acquiring appropriate continuous representations of discrete objects, such as graphs and knowledge base data, have been conducted by many researchers in the field of machine learning. In this study, we introduce Nested SubSpace (NSS) arrangement, a comprehensive framework for representation learning. We show that existing embedding techniques can be regarded as special cases of the NSS arrangement. Based on the concept of the NSS arrangement, we implement a Disk-ANChor ARrangement (DANCAR), a representation learning method specialized to reproducing general graphs. Numerical experiments have shown that DANCAR has successfully embedded WordNet in ${\mathbb R}^{20}$ with an F1 score of 0.993 in the reconstruction task. DANCAR is also suitable for visualization in understanding the characteristics of graphs.

cs.LG↗