Selected work

Representative Papers

Subcodes of Lambda-Gabidulin Codes for Compact-Ciphertext Cryptography

Apr 20, 2026

This work addresses the challenges of large ciphertext sizes and security vulnerabilities arising from algebraic structure leakage in post-quantum cryptography by investigating subspace subcodes of Lambda-Gabidulin codes. Through coordinate scaling, these subcodes are linked to classical subcodes of Gabidulin codes, enabling the first explicit linearized polynomial representation. The study fully characterizes their dimension, encoding structure, and algebraic invariants of their base-field matrix images. Leveraging these insights, a novel random subcode generation method is proposed that conceals algebraic invariants, leading to a new LGS-Niederreiter public-key encryption scheme. At security levels of 128, 192, and 256 bits, this scheme achieves the smallest ciphertext size reported to date while maintaining competitive public-key sizes.

0 citationsRead paper

Revenue-Sharing as Infrastructure: A Distributed Business Model for Generative AI Platforms

Mar 20, 2026

This study addresses how the prevailing prepaid pricing models of generative AI platforms raise barriers to entry, stifle innovation, and marginalize developers from emerging economies. To counter this, the paper proposes a “Revenue Share as Infrastructure” (RSI) model, wherein platforms offer AI services at no upfront cost and instead take a share of developers’ application revenues. This approach reconfigures incentive structures and value co-creation dynamics across multi-sided markets by inverting the traditional upstream fee paradigm. By substantially lowering access barriers and better aligning platform and developer interests, RSI fosters greater participation—particularly from low-income countries—and enables the deployment of localized AI applications in high-impact domains such as health and agriculture, especially in regions with high mobile penetration. The model thus unlocks significant potential for inclusive digital innovation and employment generation.

0 citationsRead paper

Secu-Table: a Comprehensive security table dataset for evaluating semantic table interpretation systems

Nov 09, 2025

The cybersecurity domain lacks publicly available, semantically annotated tabular benchmarks—especially for evaluating large language models (LLMs) on semantic table interpretation. Method: We introduce Secu-Table, the first large-scale, manually curated tabular benchmark for cybersecurity, comprising over 1,500 real-world security tables and 15,000+ entities. It is constructed from CVE/CWE vulnerability data and enriched via fine-grained semantic alignment and annotation using Wikidata and the SEPSES Cybersecurity Knowledge Graph (CSKG). Contribution/Results: Secu-Table supports core tasks such as table-to-knowledge-graph mapping and serves as the official benchmark for the SemTab challenge. It enables unified evaluation of both open- and closed-weight LLMs—including Falcon-3B, Mistral-7B, and GPT-4o mini. The dataset, annotation guidelines, and preprocessing code are fully open-sourced, thereby filling a critical gap in the evaluation of semantic table understanding for cybersecurity applications.

0 citationsRead paper

Leveraging deep learning for plant disease identification: a bibliometric analysis in SCOPUS from 2018 to 2024

Feb 04, 2025Journal of Scientific Agriculture

This study addresses critical gaps in generative modeling for plant disease identification using deep learning (2018–2024), including limited application and inconsistent evaluation protocols. Based on 253 publications from the Scopus database, it conducts the first systematic bibliometric analysis, employing co-occurrence analysis, citation network mapping, keyword clustering, and quantitative assessment of four core performance metrics—accuracy, precision, recall, and F1-score. Results identify José M. C. de Toledo and Arnaldo R. Barbedo as central scholars, revealing strong international collaboration clusters. Generative models remain in an early exploratory phase: methodological diversity is high, yet standardized benchmarks are absent. Highly cited works predominantly focus on lightweight CNNs and data augmentation—not generative paradigms. The findings provide empirical foundations for establishing standardized evaluation frameworks, prioritizing algorithmic development directions, and refining research funding policies in this domain.

0 citationsRead paper

Recent publications

Latest Papers

Subcodes of Lambda-Gabidulin Codes for Compact-Ciphertext Cryptography

Apr 20, 2026

This work addresses the challenges of large ciphertext sizes and security vulnerabilities arising from algebraic structure leakage in post-quantum cryptography by investigating subspace subcodes of Lambda-Gabidulin codes. Through coordinate scaling, these subcodes are linked to classical subcodes of Gabidulin codes, enabling the first explicit linearized polynomial representation. The study fully characterizes their dimension, encoding structure, and algebraic invariants of their base-field matrix images. Leveraging these insights, a novel random subcode generation method is proposed that conceals algebraic invariants, leading to a new LGS-Niederreiter public-key encryption scheme. At security levels of 128, 192, and 256 bits, this scheme achieves the smallest ciphertext size reported to date while maintaining competitive public-key sizes.

0 citationsRead paper

Revenue-Sharing as Infrastructure: A Distributed Business Model for Generative AI Platforms

Mar 20, 2026

This study addresses how the prevailing prepaid pricing models of generative AI platforms raise barriers to entry, stifle innovation, and marginalize developers from emerging economies. To counter this, the paper proposes a “Revenue Share as Infrastructure” (RSI) model, wherein platforms offer AI services at no upfront cost and instead take a share of developers’ application revenues. This approach reconfigures incentive structures and value co-creation dynamics across multi-sided markets by inverting the traditional upstream fee paradigm. By substantially lowering access barriers and better aligning platform and developer interests, RSI fosters greater participation—particularly from low-income countries—and enables the deployment of localized AI applications in high-impact domains such as health and agriculture, especially in regions with high mobile penetration. The model thus unlocks significant potential for inclusive digital innovation and employment generation.

0 citationsRead paper

Secu-Table: a Comprehensive security table dataset for evaluating semantic table interpretation systems

Nov 09, 2025

The cybersecurity domain lacks publicly available, semantically annotated tabular benchmarks—especially for evaluating large language models (LLMs) on semantic table interpretation. Method: We introduce Secu-Table, the first large-scale, manually curated tabular benchmark for cybersecurity, comprising over 1,500 real-world security tables and 15,000+ entities. It is constructed from CVE/CWE vulnerability data and enriched via fine-grained semantic alignment and annotation using Wikidata and the SEPSES Cybersecurity Knowledge Graph (CSKG). Contribution/Results: Secu-Table supports core tasks such as table-to-knowledge-graph mapping and serves as the official benchmark for the SemTab challenge. It enables unified evaluation of both open- and closed-weight LLMs—including Falcon-3B, Mistral-7B, and GPT-4o mini. The dataset, annotation guidelines, and preprocessing code are fully open-sourced, thereby filling a critical gap in the evaluation of semantic table understanding for cybersecurity applications.

0 citationsRead paper

Leveraging deep learning for plant disease identification: a bibliometric analysis in SCOPUS from 2018 to 2024

Feb 04, 2025Journal of Scientific Agriculture

This study addresses critical gaps in generative modeling for plant disease identification using deep learning (2018–2024), including limited application and inconsistent evaluation protocols. Based on 253 publications from the Scopus database, it conducts the first systematic bibliometric analysis, employing co-occurrence analysis, citation network mapping, keyword clustering, and quantitative assessment of four core performance metrics—accuracy, precision, recall, and F1-score. Results identify José M. C. de Toledo and Arnaldo R. Barbedo as central scholars, revealing strong international collaboration clusters. Generative models remain in an early exploratory phase: methodological diversity is high, yet standardized benchmarks are absent. Highly cited works predominantly focus on lightweight CNNs and data augmentation—not generative paradigms. The findings provide empirical foundations for establishing standardized evaluation frameworks, prioritizing algorithmic development directions, and refining research funding policies in this domain.

0 citationsRead paper