arXiv subjects

Daisuke Yagi

Publications and source records attributed to Daisuke Yagi.

5 recordsLinked to original sources

arXiv2025

LLM-Enhanced Self-Evolving Reinforcement Learning for Multi-Step E-Commerce Payment Fraud Risk Detection

This paper presents a novel approach to e-commerce payment fraud detection by integrating reinforcement learning (RL) with Large Language Models (LLMs). By framing transaction risk as a multi-step Markov Decision Process (MDP), RL optimizes risk detection across multiple payment stages. Crafting effective reward functions, essential for RL model success, typically requires significant human expertise due to the complexity and variability in design. LLMs, with their advanced reasoning and coding capabilities, are well-suited to refine these functions, offering improvements over traditional methods. Our approach leverages LLMs to iteratively enhance reward functions, achieving better fraud detection accuracy and demonstrating zero-shot capability. Experiments with real-world data confirm the effectiveness, robustness, and resilience of our LLM-enhanced RL framework through long-term evaluations, underscoring the potential of LLMs in advancing industrial RL applications.

cs.LG↗

arXiv2024

Multi-task CNN Behavioral Embedding Model For Transaction Fraud Detection

The burgeoning e-Commerce sector requires advanced solutions for the detection of transaction fraud. With an increasing risk of financial information theft and account takeovers, deep learning methods have become integral to the embedding of behavior sequence data in fraud detection. However, these methods often struggle to balance modeling capabilities and efficiency and incorporate domain knowledge. To address these issues, we introduce the multitask CNN behavioral Embedding Model for Transaction Fraud Detection. Our contributions include 1) introducing a single-layer CNN design featuring multirange kernels which outperform LSTM and Transformer models in terms of scalability and domain-focused inductive bias, and 2) the integration of positional encoding with CNN to introduce sequence-order signals enhancing overall performance, and 3) implementing multitask learning with randomly assigned label weights, thus removing the need for manual tuning. Testing on real-world data reveals our model's enhanced performance of downstream transaction models and comparable competitiveness with the Transformer Time Series (TST) model.

cs.LG↗

arXiv2019

An axiomatic nonparametric production function estimator: Modeling production in Japan's cardboard industry

We develop a new approach to estimate a production function based on the economic axioms of the Regular Ultra Passum law and convex non-homothetic input isoquants. Central to the development of our estimator is stating the axioms as shape constraints and using shape constrained nonparametric regression methods. We implement this approach using data from the Japanese corrugated cardboard industry from 1997-2007. Using this new approach, we find most productive scale size is a function of the capital-to-labor ratio and the largest firms operate close to the largest most productive scale size associated with a high capital-to-labor ratio. We measure the productivity growth across the panel periods based on the residuals from our axiomatic model. We also decompose productivity into scale, input mix, and unexplained effects to clarify the sources the productivity differences and provide managers guidance to make firms more productive.

stat.AP↗

arXiv2018

Insights from Machine Learning for Evaluating Production Function Estimators on Manufacturing Survey Data

Organizations like U.S. Census Bureau rely on non-exhaustive surveys to estimate industry-level production functions in years in which a full Census is not conducted. When analyzing data from non-census years, we propose selecting an estimator based on a weighting of its in-sample and predictive performance. We compare Cobb-Douglas functional assumptions to existing nonparametric shape constrained estimators and a newly proposed estimator. For simulated data, we find that our proposed estimator has the lowest weighted errors. For actual data, specifically the 2010 Chilean Annual National Industrial Survey, a Cobb-Douglas specification describes at least 90\% as much variance as the best alternative estimators in practically all cases considered providing two insights: the benefits of using application data for selecting an estimator, and the benefits of structure in noisy data.

stat.AP↗

arXiv2018

Shape constrained kernel-weighted least squares: Application to production function estimation for Chilean manufacturing industries

In this paper we examine a novel way of imposing shape constraints on a local polynomial kernel estimator. The proposed approach is referred to as Shape Constrained Kernel-weighted Least Squares (SCKLS). We prove uniform consistency of the SCKLS estimator with monotonicity and convexity/concavity constraints and establish its convergence rate. The competitiveness of SCKLS is shown in a comprehensive simulation study. Finally, we analyze Chilean manufacturing data using the SCKLS estimator and quantify production in the plastics and wood industries. The results show that exporting firms have significantly higher productivity.

stat.AP↗