Selected work

Representative Papers

Shiha Distribution: Statistical Properties and Applications to Reliability Engineering and Environmental Data

Feb 02, 2026

This study addresses the challenge posed by lifetime data that often exhibit skewness and heavy- or light-tailed characteristics, for which existing distributional models lack both flexibility and analytical tractability. To bridge this gap, the authors propose a novel two-parameter Shiha distribution and systematically derive its key statistical properties, including the moment-generating function, quantile function, entropy, and stress-strength reliability—providing, for the first time, a closed-form expression for the latter. Through comprehensive Monte Carlo simulations and empirical analyses of multiple real-world lifetime datasets, the Shiha distribution demonstrates superior fitting accuracy and model adaptability compared to established distributions, highlighting its practical utility and modeling advantages in reliability engineering and environmental science.

1 citationsRead paper

Nonvolatile Charge-Domain Attention with HZO Ferroelectric Capacitors: A Simulation-Based Device-to-System Evaluation

May 27, 2026

Transformer decoding is hindered by the high energy consumption and latency associated with attention computation and KV cache data movement. This work proposes a ferroelectric charge-domain computing unit (FCDC), which, for the first time, leverages HZO-based ferroelectric capacitors to perform non-volatile charge-domain vector-matrix multiplication. The design supports both full-die integration and a dedicated KV coprocessor deployment mode, augmented by a refresh-free KV cache residency mechanism. Through device-to-system co-modeling and multi-scale simulation across twelve pretrained large language models, the approach demonstrates minimal degradation—perplexity increases by no more than 2.9% and attention accuracy loss remains below 1% for models of at least 7B parameters. Compared to GPU baselines, it achieves 1.36–35× higher energy efficiency in RAG and agent tasks, and exceeds 41× in conversational residency scenarios.

0 citationsRead paper

Unit Shiha Distribution and its Applications to Engineering and Medical Data

Feb 04, 2026

This study addresses the limited flexibility of existing unit interval distributions in capturing diverse skewness patterns and complex hazard rate shapes, such as bathtub-shaped or J-shaped profiles. To overcome this, the authors propose a novel Unit Shiha (USh) distribution, constructed by applying an inverse exponential transformation to the Shiha distribution. This is the first unit distribution capable of unifying the modeling of both left- and right-skewed data alongside multiple hazard rate forms. Parameter inference is conducted via maximum likelihood estimation, complemented by analyses of moments, quantile functions, entropy, and stress-strength reliability. Simulation studies demonstrate favorable estimation performance, and empirical evaluations on four real-world engineering and medical datasets show that the USh distribution consistently achieves significantly better goodness-of-fit than established unit distributions, confirming its superior flexibility and practical utility.

0 citationsRead paper

Transformer-Based Model for Multilingual Hope Speech Detection

Jan 31, 2026

This study addresses hope speech detection in English and German contexts, aiming to enhance the automatic identification of positive expressions in multilingual settings. For the first time, we apply RoBERTa (monolingual) and XLM-RoBERTa (English–German bilingual) to this task, leveraging their Transformer-based architectures and fine-tuning strategies for effective detection. Experimental results demonstrate that the proposed approach achieves an accuracy of 81.8% and a weighted F1-score of 0.818 on the English dataset, and 78.5% accuracy with a weighted F1-score of 0.786 in the multilingual English–German setting. These findings confirm the efficacy of pretrained language models in recognizing positively valenced text and highlight their strong potential for cross-lingual transfer in affective computing tasks.

0 citationsRead paper

A Unified AI, Embedded, Simulation, and Mechanical Design Approach to an Autonomous Delivery Robot

Dec 26, 2025

This work addresses two critical challenges in fully autonomous delivery robots deployed on resource-constrained embedded platforms: (1) insufficient real-time performance of AI perception algorithms, and (2) low-latency, reliable communication between ROS 2 and embedded motion controllers. To this end, we propose a heterogeneous computing framework that tightly couples AI-based perception with hard real-time motion control. We design a firmware-level emergency motor shutdown mechanism for fault tolerance and an AWS IoT–enabled remote reliability monitoring system. Furthermore, we implement a cross-platform, low-latency communication protocol between Raspberry Pi 5 and ESP32/FreeRTOS, integrated with PID-based closed-loop motor control. The complete system demonstrates end-to-end perception, path planning, and motion control in realistic outdoor environments, sustaining continuous, stable delivery operations. It achieves industrial-grade robustness, determinism, and deployability on cost-effective embedded hardware.

0 citationsRead paper

Recent publications

Latest Papers

Nonvolatile Charge-Domain Attention with HZO Ferroelectric Capacitors: A Simulation-Based Device-to-System Evaluation

May 27, 2026

Transformer decoding is hindered by the high energy consumption and latency associated with attention computation and KV cache data movement. This work proposes a ferroelectric charge-domain computing unit (FCDC), which, for the first time, leverages HZO-based ferroelectric capacitors to perform non-volatile charge-domain vector-matrix multiplication. The design supports both full-die integration and a dedicated KV coprocessor deployment mode, augmented by a refresh-free KV cache residency mechanism. Through device-to-system co-modeling and multi-scale simulation across twelve pretrained large language models, the approach demonstrates minimal degradation—perplexity increases by no more than 2.9% and attention accuracy loss remains below 1% for models of at least 7B parameters. Compared to GPU baselines, it achieves 1.36–35× higher energy efficiency in RAG and agent tasks, and exceeds 41× in conversational residency scenarios.

0 citationsRead paper

Unit Shiha Distribution and its Applications to Engineering and Medical Data

Feb 04, 2026

This study addresses the limited flexibility of existing unit interval distributions in capturing diverse skewness patterns and complex hazard rate shapes, such as bathtub-shaped or J-shaped profiles. To overcome this, the authors propose a novel Unit Shiha (USh) distribution, constructed by applying an inverse exponential transformation to the Shiha distribution. This is the first unit distribution capable of unifying the modeling of both left- and right-skewed data alongside multiple hazard rate forms. Parameter inference is conducted via maximum likelihood estimation, complemented by analyses of moments, quantile functions, entropy, and stress-strength reliability. Simulation studies demonstrate favorable estimation performance, and empirical evaluations on four real-world engineering and medical datasets show that the USh distribution consistently achieves significantly better goodness-of-fit than established unit distributions, confirming its superior flexibility and practical utility.

0 citationsRead paper

Shiha Distribution: Statistical Properties and Applications to Reliability Engineering and Environmental Data

Feb 02, 2026

This study addresses the challenge posed by lifetime data that often exhibit skewness and heavy- or light-tailed characteristics, for which existing distributional models lack both flexibility and analytical tractability. To bridge this gap, the authors propose a novel two-parameter Shiha distribution and systematically derive its key statistical properties, including the moment-generating function, quantile function, entropy, and stress-strength reliability—providing, for the first time, a closed-form expression for the latter. Through comprehensive Monte Carlo simulations and empirical analyses of multiple real-world lifetime datasets, the Shiha distribution demonstrates superior fitting accuracy and model adaptability compared to established distributions, highlighting its practical utility and modeling advantages in reliability engineering and environmental science.

1 citationsRead paper

Transformer-Based Model for Multilingual Hope Speech Detection

Jan 31, 2026

This study addresses hope speech detection in English and German contexts, aiming to enhance the automatic identification of positive expressions in multilingual settings. For the first time, we apply RoBERTa (monolingual) and XLM-RoBERTa (English–German bilingual) to this task, leveraging their Transformer-based architectures and fine-tuning strategies for effective detection. Experimental results demonstrate that the proposed approach achieves an accuracy of 81.8% and a weighted F1-score of 0.818 on the English dataset, and 78.5% accuracy with a weighted F1-score of 0.786 in the multilingual English–German setting. These findings confirm the efficacy of pretrained language models in recognizing positively valenced text and highlight their strong potential for cross-lingual transfer in affective computing tasks.

0 citationsRead paper

A Unified AI, Embedded, Simulation, and Mechanical Design Approach to an Autonomous Delivery Robot

Dec 26, 2025

This work addresses two critical challenges in fully autonomous delivery robots deployed on resource-constrained embedded platforms: (1) insufficient real-time performance of AI perception algorithms, and (2) low-latency, reliable communication between ROS 2 and embedded motion controllers. To this end, we propose a heterogeneous computing framework that tightly couples AI-based perception with hard real-time motion control. We design a firmware-level emergency motor shutdown mechanism for fault tolerance and an AWS IoT–enabled remote reliability monitoring system. Furthermore, we implement a cross-platform, low-latency communication protocol between Raspberry Pi 5 and ESP32/FreeRTOS, integrated with PID-based closed-loop motor control. The complete system demonstrates end-to-end perception, path planning, and motion control in realistic outdoor environments, sustaining continuous, stable delivery operations. It achieves industrial-grade robustness, determinism, and deployability on cost-effective embedded hardware.

0 citationsRead paper