- Research Article
- 10.1016/j.ins.2026.123362
Enhancing differentially private distributed k-means clustering through data-driven sensitivity calibration and optimized dropout handling
- Jul 01, 2026
- Information Sciences
- Jong Wook Kim + 1 more +1
Publications from 2021 to 2026
Showing 10 of 277 papers
Enhancing differentially private distributed k-means clustering through data-driven sensitivity calibration and optimized dropout handling
Forecasting-based biomedical time-series data synthesis for open data and robust AI.
Investor trading behavior and intermediate prospect theory value in cross-sectional expected returns
Beyond the boundary: how middle managers shape inter-organizational public R&D teams
Purpose This study examines how the structural configuration and cognitive composition of middle managers influence innovation performance in public research and development (R&D) consortia. It seeks to identify the conditions under which coordination efforts by middle managers facilitate or hinder collaboration outcomes in multi-organizational R&D environments. Design/methodology/approach Combining boundary spanning theory (BST) with a mixed-methods design, the study analyzes 1,150 South Korean energy R&D projects through econometric modeling and validates findings with in-depth interviews from six middle managers. The analysis examines three dimensions: intensity of managerial presence, non-linear performance patterns, and the trade-off between specialty and diversity. Findings The results reveal a U-shaped relationship between middle manager intensity and publication quantity, with both low and high levels associated with superior outcomes compared to moderate configurations. In contrast, excessive managerial presence is linked to lower publication quality. Furthermore, teams composed of managers with shared domain expertise consistently demonstrate higher performance than those characterized by greater cognitive diversity. Originality/value This study reframes BST by introducing structural and compositional conditions that affect its efficacy in public-sector innovation networks. It advances theory by identifying intensity and composition as contingent variables and extends its applicability to policy-driven, multi-institutional R&D contexts. The findings provide design-relevant insights for optimizing managerial roles in complex collaborative projects.
Read moreBAKSHEESH: Similar Yet Different From GIFT (and ZORRO)
We propose a lightweight block cipher named BAKSHEESH, which draws inspiration from GIFT (CHES'17). BAKSHEESH runs for 35 rounds, which is 12.5 per cent fewer rounds compared to GIFT-128 whilst maintaining the same security claims against classical attacks. However, we also present an alternate (but equivalent) angle of BAKSHEESH that follows the heritage of ZORRO (CHES'13). In the GIFT heritage version, BAKSHEESH uses a 4-bit SBox that has a non-trivial Linear Structure (LS). In the alternate ZORRO heritage specification, it employs a 3-bit SBox and realises a partial non-linear layer. BAKSHEESH is suitable for efficient hardware and software implementations, and also offers an edge on side channel countermeasures and other niche applications. For instance, our study on the threshold implementation shows that BAKSHEESH offers a few-fold advantage over other lightweight ciphers. We therefore create a new paradigm of lightweight ciphers through adequate deliberation on the design choice and solidify it with appropriate security analysis and ample implementation/benchmark.
Read moreToward Crypto Agility: Automated Analysis of Quantum-Vulnerable TLS via Packet Inspection
Debunking Optimization Myths in Federated Learning for Medical Image Classification
New Results in Quantum Analysis of LED: Featuring One and Two Oracle Attacks
Deep Learning-Based Resident Change Detection Using Electricity Metering Data
As smart meter technology advances, the advanced metering infrastructure (AMI) enables real-time monitoring of electricity usage at 15-minute intervals. Since prior studies mainly focused on industrial infrastructures, anomaly detection in residential electricity usage remains underexplored. In this paper, we propose a gated recurrent unit (GRU)-based deep learning model to detect resident changes in residential households using the electricity metering data through AMI. Monthly consumption patterns and energy consumption are predicted using actual electric energy metering data collected from 846 apartment households in Seoul, Korea. To detect the resident changes, combined losses are computed by integrating the consumption pattern and consumption energy prediction errors. The experimental results show an accuracy of 82.55%, specificity of 100%, and recall of 65.11% when allowing a +-1-month error margin. The proposed deep leaning model demonstrates the practical feasibility of detecting resident changes in residential households and is expected to contribute to the development of smart grid services.
Read moreThe Impact of Product Environmental Innovation in Process Industries: Evidence from Innovation Efficiency and Performance
This study examines the heterogeneous effects of product environmental innovation on firm-level innovation efficiency and performance in process industries, with a focus on the chemical and electronics sectors. Following the Organisation for Economic Co-operation and Development (OECD)’s Oslo Manual, four types of product environmental innovation are considered: reducing energy use and emissions (RUE), reducing pollution (RP), promoting recycling (PR), and enhancing durability and extending product life (EDEL). Innovation efficiency is evaluated using the input-oriented Banker–Charnes–Cooper (BCC) Data Envelopment Analysis (DEA) model, and regression analyses are applied to test the effects of each innovation type on efficiency and sales outcomes. The results reveal that RUE and EDEL consistently enhance both efficiency and performance, whereas PR has a negative impact on performance, and RP shows no significant effect. These findings demonstrate that product environmental innovation is not a homogeneous construct but yields heterogeneous outcomes depending on type and industry context. The study contributes to the literature by jointly examining efficiency and performance outcomes and by overcoming the limitations of single-metric evaluations, and it provides practical implications by clarifying which innovation types deliver immediate value in business-to-consumer (B2C) markets and which are more relevant for business-to-business (B2B) settings.
Read more