University of Waikato

Near-infrared (NIR) spectroscopy is a widely used technology in the horticulture industry for non-destructive fruit grading. Partial Least Squares (PLS) regression is the dominant method for producing fruit quality predictions from measured spectra. Alternative deep learning methods have shown promise, but often require large amounts of labelled data to train. This study proposes a semi-supervised method based on Barlow Twins to include unlabelled data in the training process. We adopt the Barlow Twins method by using repeated measurements on the same fruit from different devices as different “views” to encode into the same latent space and combine the encoder network with a regression head for prediction. Our approach demonstrates improved performance over PLS with up to 17% lower RMSE, especially when the labelled data is limited. The Barlow loss function also improves calibration transfer results. • Novel application of Barlow Twins contrastive learning to NIR spectroscopy using repeated measurements from different devices. • Semi-supervised learning allows for unlabelled spectra to assist in training. • Up to 50% reduction in RMSE in calibration transfer tasks compared to training on the MSE loss only. • The best performance gains were observed at small labelled training sizes. • The Barlow Twins loss was not detrimental at large training set sizes.

The 4th UN General Assembly declaration on non-communicable diseases (NCDs) and mental health, adopted on Dec 15, 2025, was notable for its explicit inclusion of dementia, bringing dementia prevention into focus as an international public policy priority.1 A central question therefore becomes how dementia prevention will be prioritised, resourced, and implemented and how it will be integrated into existing NCD policies and health-system strategies (appendix).

This study examines how firms respond to shifts in economic policy uncertainty (EPU) by analyzing its impact on firm resilience. Using data from 2,660 Chinese A-share listed firms from 2010 to 2022, we find that higher firm-perceived policy uncertainty (FEPU) significantly weakens firm resilience. Drawing on real options and growth options perspectives, the results show that firms become less adaptable when uncertainty leads managers to behave more cautiously and when operational decisions—such as cash reserves and R&D spending—are distorted. Social media pressure, however, softens this negative effect by encouraging greater transparency and discipline. The findings also reveal substantial differences across ownership types, financial conditions, and industry characteristics. The study provides practical insights for managers and policymakers seeking to strengthen organizational resilience in uncertain environments and highlights the need to design governance and communication strategies that help firms remain adaptive when policy risks rise. • Corporate perceived policy uncertainty (FEPU) weakens firms’ ability to adapt and remain resilient. • Managerial behaviour and operations explain how uncertainty reduces resilience. • Social media pressure helps firms counteract the negative effects of FEPU. • Corporate resilience varies by ownership, financial strength, industry, and leadership traits. • Findings offer guidance for building stronger firms under rising policy risks.

Pinch analysis provides a systematic approach to targeting the minimum heating and cooling requirements of industrial processes and to designing heat exchanger networks that approach this target. Decades after its introduction, the method remains essential for energy efficiency, decarbonisation, and retrofit studies across sectors such as chemicals, pulp and paper, food, and refining. In addition, numerous papers have proposed extensions to PA but these are often left isolate and unavailable for future development and implementation. In short, tool development has not kept pace with academic progress and today’s industry requirements. Many practitioners rely on bespoke or basic spreadsheets or legacy desktop applications that make it difficult to scale analyses, embed the results in modern workflows, or experiment with emerging pinch techniques. OpenPinch fills this gap with an open-source Python toolkit that implements advanced pinch analysis and total site integration methods. The library offers typed schemas for robust data ingestion, modular analysis routines that span unit operation to process, site and regional scales, and export paths that maintain compatibility with conventional spreadsheet workflows. As illustrated by a survey of current pinch tools, a distinctive contribution is its comprehensive implementation of (unified) total site heat integration combined with multiple utilities, both isothermal and non-isothermal ones, via a combination of problem-table transformations, temperature-interval segmentation, and iterative assignment. The basis of these algorithms were reported in Tarighaleslami et al. (2017a). This technical note documents the architecture, algorithms, and data flow of OpenPinch, highlighting an end-to-end example of a pulp mill case study as well as key directions for future extension.

Magnetic reconnection plays an important role in the turbulent relaxation of space and astrophysical plasmas, such as the solar corona, solar wind, and Earth’s magnetosheath. Recent studies have shed light on the role of magnetic reconnection as an efficient energy dissipation mechanism in these large-scale turbulent systems. However, the relative role of magnetic reconnection in dissipating turbulent energy in these macroscopic systems is still not fully understood. To investigate these issues, we simulate a turbulent plasma system using magnetohydrodynamic (MHD) simulations. A large number of reconnection sites are found, and their statistical properties are quantified. The study reveals, for the first time, that the distribution of upstream reconnecting fields is strongly correlated with the distribution of global fields at the energy-containing scales. To further explore these relations in weakly collisional systems, we perform a similar analysis on kinetic Particle-in-Cell (PIC) simulations of plasma turbulence and on in situ observations of the terrestrial magnetosheath using the Magnetospheric Multiscale Mission (MMS). Notably, the key conclusions drawn from MHD simulations remain valid in both the kinetic simulations and MMS observations. These findings are expected to significantly refine theoretical estimates of reconnection rates and heating rates resulting from magnetic reconnection.

In the context of the Classification and Regression Trees (CART) algorithm, the efficient splitting of categorical features using standard criteria like GINI and Entropy is well-established. However, using the Mean Absolute Error (MAE) criterion for categorical features has traditionally relied on various numerical encoding methods. This paper demonstrates that unsupervised numerical encoding methods are not viable for MAE criteria. Furthermore, we present a novel and efficient splitting algorithm that addresses the challenges of handling categorical features with the MAE criterion. Our findings underscore the limitations of existing approaches and offer a promising solution to enhance the handling of categorical data in CART algorithms.

The intentional design of robots to evoke creepiness provides a unique lens for studying human perception and willingness to engage. To understand user perceptions and acceptance of robots we developed a robot prototype designed with targeted facial, morphological, and movement features that may be perceived as "creepy". Using the Human-Robot Interaction Evaluation Scale (HRIES) we found that disturbance was moderate towards our intentionally creepy robot with significant participant variation. Furthermore, qualitative results confirmed this polarity, with descriptions ranging from "angry and unfriendly" to "cool and cute". This variability demonstrates that "creepiness" is more subjective than initially anticipated and highlights a key research gap in academic literature with the need for measurement tools which capture negative perceptions in HRI.

Purpose This study analyzes the mechanism by which capitalization impacts green total factor productivity (GTFP) in pig production, revealing its operational pathways. The findings provide empirical evidence relevant for future research and support for green development in the pig farming industry. Design/methodology/approach Based on the provincial panel data of large-, medium- and small-scale pig farms from 2016 to 2023, the capitalization level and GTFP of different-sized pig farms were measured. The impact mechanism and path of capitalization and GTFP of pig farming were empirically examined using fixed effect models, moderated effect models and panel threshold models. Findings Capitalization has significantly enhanced the GTFP of large-scale and medium-scale pig farms. This conclusion remains valid after undergoing multiple robustness tests, but the impact on small-scale farms failed to pass the significance test. The results of the mechanism analysis show that in large-scale pig farms, industrial agglomeration and government support significantly enhance the positive impact of capitalization on the GTFP of pig farming. The panel threshold effect test reveals that in medium-sized livestock farms, the positive impact of capitalization on the GTFP of pig farming exhibits a nonlinear marginal increase. Originality/value This article puts forward policy suggestions such as improving the extensive development model, implementing differentiated and precise measures, and steadily advancing the process of pig farming capitalization.