Open Research Repository
The Open Research Repository is the University’s online open access repository for collecting, maintaining and disseminating the scholarly output of the University.
Contribute to the Open Research Repository
Communities in DSpace
Select a community to browse its collections.
Recent Submissions
Item type: Publication , Access status: Metadata only , CECNN: COPULA-ENHANCED CONVOLUTIONAL NEURAL NETWORKS IN JOINT PREDICTION OF REFRACTION ERROR AND AXIAL LENGTH BASED ON ULTRA-WIDEFIELD FUNDUS IMAGES(2025) Zhong, Chong; Li, Yang; Yang, Danjuan; Li, Meiyan; Zhou, Xingtao; Fu, Bo; Liu, Catherine C.; Welsh, A. H.The ultra-widefield (UWF) fundus image is an attractive 3D biomarker in AI-aided myopia screening because it provides much richer myopia-related information. Though axial length (AL) has been acknowledged to be highly related to the two key targets of myopia screening, spherical equivalence (SE) measurement and high myopia diagnosis, its prediction based on the UWF fundus image is rarely considered. To save the high expense and time costs of measuring SE and AL, we propose the Copula-enhanced Convolutional Neural Network (CeCNN), a one-stop UWF-based ophthalmic AI framework to jointly predict SE, AL, and myopia status. The CeCNN formulates a multiresponse regression that relates multiple dependent discrete-continuous responses and the image covariate, where the nonlinearity of the association is modeled by a backbone CNN. To thoroughly describe the dependence structure among the responses, we model and incorporate the conditional dependence among responses in a CNN through a new copula-likelihood loss. We provide statistical interpretations of the conditional dependence among responses and reveal that such dependence is beyond the dependence explained by the image covariate. We heuristically justify that the proposed loss can enhance the estimation efficiency of the CNN weights. We apply the CeCNN to the UWF dataset collected by us and demonstrate that the CeCNN sharply enhances the predictive capability of various backbone CNNs. Our study supports the ophthalmology view that, besides SE, AL is an important measure of myopia.Item type: Publication , Access status: Metadata only , The ATO ALife Family File(2025) Fischer, Daniel; Holland, Justin; Chan, Marc; Polidano, Cain; Vu, Ha; Wilkins, RogerThe ATO ALife Family File (ALife-Family) combines rich family connections and administrative superannuation and tax data that opens new opportunities for understanding joint financial decisions between partners, intergenerational income mobility and some aspects of wealth mobility. In this study, we describe the construction of the ALife Family File, compare it to the HILDA Survey and demonstrate its value as a tool for intergenerational analysis. Our analysis shows that the characteristics of couples in ALife-Family are consistent with those in HILDA but that the former has the advantage of better coverage of superannuation and income data. Preliminary analysis highlights both opportunities and challenges of using ALife-Family when examining intergenerational issues.Item type: Publication , Access status: Metadata only , Further remarks on permissible covariance structures for simultaneous retention of BLUEs in linear models(2023-05-30) Haslett, Stephen J.; Isotalo, Jarkko; Markiewicx, Augustyn; Puntanen, SimoWe consider the partitioned linear model M12 (V0) = {y, X1 β1 + X2 β2, V0 } and the corresponding small model M1 (V0) = {y, X1 β1, V0 }. We define the set V1/12 of nonnegative definite matrices V such that every representation of the best linear unbiased estimator, BLUE, of µ1 = X1 β1 under M12 (V0) remains BLUE under M12 (V). Correspondingly, we can characterize the set V1 of matrices V such that every BLUE of µ1 = X1 β1 under M1 (V0) remains BLUE under M1 (V). In this paper we focus on the mutual relations between the sets V1 and V1/12 .Item type: Publication , Access status: Metadata only , AI and Human Creativity on Content Creation: Analyzing Watermark and Detection Policies(Association for Information Systems, 2024) Wu, Zhenhua; Hu, Lin; Chen, Pei YuThis paper investigates the dynamic interaction between artificial intelligence (AI) and human creativity on digital content platforms, focusing on how AI affects content creators' incentives and consumer welfare through content quality. We develop a model to analyze AI's impact, particularly its substitutive and complementary relationships with human skills, and examine the effectiveness of AI watermarking and detection policies. Our findings indicate that AI benefits low-skill creators when human-input costs are low but may push high-skill creators out of the market when costs are high, reducing overall content quality. AI watermarking, which marks AI-generated content, supports high-skill creators if human-input costs are moderate, thus enhancing content quality. Both watermarking and AI detection services improve content quality and consumer satisfaction, with the effectiveness depending on watermark removal rates. High-skill creators prefer watermarking when removal rates are high and detection services when they are low, benefiting consumers and platforms alike.Item type: Publication , Access status: Metadata only , Dynamic consumption and portfolio choice with heterogeneous discounting and information learning(2026) Zhang, Wuqi; Wang, Ning; Wang, Hao; Zhang, NiThis paper presents a portfolio optimization problem incorporating independent standard Brownian motions, with a focus on wealth allocation across bonds, stocks, and money market accounts. Investors utilize an information variable derived from historical performance during a specified learning horizon to inform their decisions. The model incorporates heterogeneous discounting, where the utilities from consumption and terminal wealth are valued differently. By solving the extended Hamilton–Jacobi–Bellman (HJB) equations, we derive time-consistent strategies for power, exponential, and logarithmic utility functions, respectively. Finally, numerical examples are provided to illustrate the impacts of several model parameters on optimal strategies.