arXiv subjects

Terence L. van Zyl

Publications and source records attributed to Terence L. van Zyl.

At least 19 recordsLinked to original sources

arXiv2026

A Time-Synchronous Prequential Evaluation Framework for Next-Week Student Disengagement Prediction from Moodle LMS Logs

Retrospective evaluation of early-warning models may not reflect information available when predictions are issued and outcomes become observable. We developed a time-synchronous prequential framework for online prediction of next-week student disengagement, operationalised as inactivity onset, from Moodle logs. Ordered by calendar time, models were updated only with available outcomes and remained unchanged for all predictions at each prediction cutoff; development, calibration and evaluation were chronologically separate. We compared stochastic gradient descent logistic regression (SGD-LR), online sequential extreme learning machine (OSELM) and echo state network (ESN) across predictive performance, fixed, frozen and rolling threshold policies, calendar-time variation and four behavioural feature sets. Held-out evaluation comprised 49,329 predictions with 37.68% inactivity-onset prevalence. Discrimination was similar (average precision 0.665-0.679; ROC-AUC 0.755-0.774). OSELM had the highest recall, SGD-LR the highest precision and specificity, and ESN the highest Matthews correlation coefficient and balanced accuracy. Lower thresholds increased recall and alert volume, whereas rolling adaptation did not improve MCC over frozen thresholds. Weekly performance and calibrated-probability distributions varied without sustained deterioration. During model selection, cumulative-and-temporal features achieved the highest individual-set AP, with modest gains from combining all feature groups. These findings indicate that streaming early-warning evaluation should consider calibration, threshold-dependent performance, alert burden and calendar-time variation alongside aggregate discrimination.

cs.LG↗

arXiv2026

Longitudinal Learner Participation Profiling in Online Learning: Heterogeneous Consensus Clustering under Unequal Observation Windows

Learning management system traces are clustered to identify participation profiles, but unequal observation windows may cause profiles to reflect duration rather than participation alone. Using 16,448 complete learner--course sequences from 4,599 learners across 152 courses at a South African public university, we evaluated eight participation measures. All remained related to observation-window length even when expressed as rates or shares. Course-grouped adjustment reduced these relationships. Five retained measures were evaluated using K-means, Gaussian mixtures, Ward agglomerative clustering and K-medoids, with standardised measures and principal-component summaries. Evidence supported a broad two-profile grouping and a more detailed adjusted four-profile consensus: Gradually declining, Early-concentrated, Cyclical variable and Sustained active. Profile membership accounted for 2.87% of the variation in duration in the primary consensus, versus 38.18% using original measures. Mean assignment support was 0.739; 8.35% of assignments fell below 0.50, with greatest uncertainty in Sustained active. Two week-by-week sequence analyses agreed strongly with each other but only modestly with the four-profile consensus, indicating that the profiles combined multiple aspects of participation rather than distinct sequence patterns. Profiling should therefore assess duration effects, compare clustering methods and participation summaries, and report assignment consistency.

cs.LG↗

arXiv2026

Multi-Object Tracking Consistently Improves Wildlife Inference

Camera traps have become a common tool for wildlife monitoring efforts in ecological research and biodiversity conservation. Wildlife classification models have benefited from the increase in wildlife visual data. These models reach high levels of accuracy on curated, high-quality datasets. However, their performance remains sensitive to real-world environmental constraints. They often produce inconsistent predictions when performing inference on temporally coherent sequences. The predicted label for a single individual shifts rapidly between frames. This study exploits the temporal nature of camera-trap data to augment inferred predictions from a wildlife classification model. Specifically, we adopt several standard Multi-Object Tracking (MOT) models to link detections across consecutive frames. The curated trajectories are used to fuse the softmax class probabilities. The fused probability score produces a single consensus class label estimate that overrides misclassifications caused by noise. The analysis of the experimental results shows that our proposed strategy improves over a standalone classifier over all datasets and for each metric. Specifically, the best-performing MOT models gain a weighted F1-Score of 5.1%, 3.1% and 2.0% over the classifier across three MOT datasets.

cs.CV↗

arXiv2026

Complexity of Linear Regions in Self-supervised Deep ReLU Networks

There has been growing interest in studying the complexity of Rectified Linear Unit (ReLU) based activation networks. Recent work investigates the evolution of the number of piecewise-linear partitions (linear regions) that are formed during training. However, current research is limited to examining the complexity of models trained in a supervised way. Self-Supervised Learning (SSL) differs in that it directly optimises the representation space using a loss function to enhance the model's performance across multiple downstream tasks. This study investigates the local distribution of linear regions produced by SSL models. We demonstrate that the evolution of linear regions correlates with the representation quality by utilising SplineCam to extract two-dimensional polytopes near the data distribution. We track the number, area, eccentricity, and boundaries of regions throughout training. The study compares supervised, contrastive, and self-distillation methods over two standard benchmark datasets, MNIST and FashionMNIST. The analysis of the experimental results shows that self-supervised methods create substantially fewer regions to achieve comparable accuracy to supervised models. Contrastive methods rapidly expand regions over time, whereas self-distillation methods tend to consolidate by merging neighbouring regions. Lastly, we can detect representation collapse early within the geometric space of linear regions. Our analysis suggests that polytopal metrics can serve as reliable indicators of representation quality and model performance.

cs.LG↗

arXiv2026

Improving Wildlife Out-of-Distribution Detection: Africas Big Five

Mitigating human-wildlife conflict seeks to resolve unwanted encounters between these parties. Computer Vision provides a solution to identifying individuals that might escalate into conflict, such as members of the Big Five African animals. However, environments often contain several varied species. The current state-of-the-art animal classification models are trained under a closed-world assumption. They almost always remain overconfident in their predictions even when presented with unknown classes. This study investigates out-of-distribution (OOD) detection of wildlife, specifically the Big Five. To this end, we select a parametric Nearest Class Mean (NCM) and a non-parametric contrastive learning approach as baselines to take advantage of pretrained and projected features from popular classification encoders. Moreover, we compare our baselines to various common OOD methods in the literature. The results show feature-based methods reflect stronger generalisation capability across varying classification thresholds. Specifically, NCM with ImageNet pre-trained features achieves a 2%, 4% and 22% improvement on AUPR-IN, AUPR-OUT and AUTC over the best OOD methods, respectively. The code can be found here https://github.com/pxpana/BIG5OOD

cs.CV↗

arXiv2025

Nearest-Class Mean and Logits Agreement for Wildlife Open-Set Recognition

Current state-of-the-art Wildlife classification models are trained under the closed world setting. When exposed to unknown classes, they remain overconfident in their predictions. Open-set Recognition (OSR) aims to classify known classes while rejecting unknown samples. Several OSR methods have been proposed to model the closed-set distribution by observing the feature, logit, or softmax probability space. A significant drawback of many existing approaches is the requirement to retrain the pre-trained classification model with the OSR-specific strategy. This study contributes a post-processing OSR method that measures the agreement between the models' features and predicted logits. We propose a probability distribution based on an input's distance to its Nearest Class Mean (NCM). The NCM-based distribution is then compared with the softmax probabilities from the logit space to measure agreement between the NCM and the classification head. Our proposed strategy ranks within the top three on two evaluated datasets, showing consistent performance across the two datasets. In contrast, current state-of-the-art methods excel on a single dataset. We achieve an AUROC of 93.41 and 95.35 for African and Swedish animals. The code can be found https://github.com/Applied-Representation-Learning-Lab/OSR.

cs.CV↗

arXiv2025

Wildlife Target Re-Identification Using Self-supervised Learning in Non-Urban Settings

Wildlife re-identification aims to match individuals of the same species across different observations. Current state-of-the-art (SOTA) models rely on class labels to train supervised models for individual classification. This dependence on annotated data has driven the curation of numerous large-scale wildlife datasets. This study investigates self-supervised learning Self-Supervised Learning (SSL) for wildlife re-identification. We automatically extract two distinct views of an individual using temporal image pairs from camera trap data without supervision. The image pairs train a self-supervised model from a potentially endless stream of video data. We evaluate the learnt representations against supervised features on open-world scenarios and transfer learning in various wildlife downstream tasks. The analysis of the experimental results shows that self-supervised models are more robust even with limited data. Moreover, self-supervised features outperform supervision across all downstream tasks. The code is available here https://github.com/pxpana/SSLWildlife.

cs.CV↗

arXiv2023

A Learnheuristic Approach to A Constrained Multi-Objective Portfolio Optimisation Problem

Multi-objective portfolio optimisation is a critical problem researched across various fields of study as it achieves the objective of maximising the expected return while minimising the risk of a given portfolio at the same time. However, many studies fail to include realistic constraints in the model, which limits practical trading strategies. This study introduces realistic constraints, such as transaction and holding costs, into an optimisation model. Due to the non-convex nature of this problem, metaheuristic algorithms, such as NSGA-II, R-NSGA-II, NSGA-III and U-NSGA-III, will play a vital role in solving the problem. Furthermore, a learnheuristic approach is taken as surrogate models enhance the metaheuristics employed. These algorithms are then compared to the baseline metaheuristic algorithms, which solve a constrained, multi-objective optimisation problem without using learnheuristics. The results of this study show that, despite taking significantly longer to run to completion, the learnheuristic algorithms outperform the baseline algorithms in terms of hypervolume and rate of convergence. Furthermore, the backtesting results indicate that utilising learnheuristics to generate weights for asset allocation leads to a lower risk percentage, higher expected return and higher Sharpe ratio than backtesting without using learnheuristics. This leads us to conclude that using learnheuristics to solve a constrained, multi-objective portfolio optimisation problem produces superior and preferable results than solving the problem without using learnheuristics.

cs.LG↗

arXiv2023

Late Meta-learning Fusion Using Representation Learning for Time Series Forecasting

Meta-learning, decision fusion, hybrid models, and representation learning are topics of investigation with significant traction in time-series forecasting research. Of these two specific areas have shown state-of-the-art results in forecasting: hybrid meta-learning models such as Exponential Smoothing - Recurrent Neural Network (ES-RNN) and Neural Basis Expansion Analysis (N-BEATS) and feature-based stacking ensembles such as Feature-based FORecast Model Averaging (FFORMA). However, a unified taxonomy for model fusion and an empirical comparison of these hybrid and feature-based stacking ensemble approaches is still missing. This study presents a unified taxonomy encompassing these topic areas. Furthermore, the study empirically evaluates several model fusion approaches and a novel combination of hybrid and feature stacking algorithms called Deep-learning FORecast Model Averaging (DeFORMA). The taxonomy contextualises the considered methods. Furthermore, the empirical analysis of the results shows that the proposed model, DeFORMA, can achieve state-of-the-art results in the M4 data set. DeFORMA, increases the mean Overall Weighted Average (OWA) in the daily, weekly and yearly subsets with competitive results in the hourly, monthly and quarterly subsets. The taxonomy and empirical results lead us to argue that significant progress is still to be made by continuing to explore the intersection of these research areas.

cs.LG↗

arXiv2023

Incremental Class Learning using Variational Autoencoders with Similarity Learning

Catastrophic forgetting in neural networks during incremental learning remains a challenging problem. Previous research investigated catastrophic forgetting in fully connected networks, with some earlier work exploring activation functions and learning algorithms. Applications of neural networks have been extended to include similarity learning. Understanding how similarity learning loss functions would be affected by catastrophic forgetting is of significant interest. Our research investigates catastrophic forgetting for four well-known similarity-based loss functions during incremental class learning. The loss functions are Angular, Contrastive, Center, and Triplet loss. Our results show that the catastrophic forgetting rate differs across loss functions on multiple datasets. The Angular loss was least affected, followed by Contrastive, Triplet loss, and Center loss with good mining techniques. We implemented three existing incremental learning techniques, iCaRL, EWC, and EBLL. We further proposed a novel technique using Variational Autoencoders (VAEs) to generate representation as exemplars passed through the network's intermediate layers. Our method outperformed three existing state-of-the-art techniques. We show that one does not require stored images (exemplars) for incremental learning with similarity learning. The generated representations from VAEs help preserve regions of the embedding space used by prior knowledge so that new knowledge does not ``overwrite'' it.

cs.LG↗

arXiv2022

Towards a methodology for addressing missingness in datasets, with an application to demographic health datasets

Missing data is a common concern in health datasets, and its impact on good decision-making processes is well documented. Our study's contribution is a methodology for tackling missing data problems using a combination of synthetic dataset generation, missing data imputation and deep learning methods to resolve missing data challenges. Specifically, we conducted a series of experiments with these objectives; $a)$ generating a realistic synthetic dataset, $b)$ simulating data missingness, $c)$ recovering the missing data, and $d)$ analyzing imputation performance. Our methodology used a gaussian mixture model whose parameters were learned from a cleaned subset of a real demographic and health dataset to generate the synthetic data. We simulated various missingness degrees ranging from $10 \%$, $20 \%$, $30 \%$, and $40\%$ under the missing completely at random scheme MCAR. We used an integrated performance analysis framework involving clustering, classification and direct imputation analysis. Our results show that models trained on synthetic and imputed datasets could make predictions with an accuracy of $83 \%$ and $80 \%$ on $a) $ an unseen real dataset and $b)$ an unseen reserved synthetic test dataset, respectively. Moreover, the models that used the DAE method for imputed yielded the lowest log loss an indication of good performance, even though the accuracy measures were slightly lower. In conclusion, our work demonstrates that using our methodology, one can reverse engineer a solution to resolve missingness on an unseen dataset with missingness. Moreover, though we used a health dataset, our methodology can be utilized in other contexts.

cs.LG↗

arXiv2022

Multi-Modal Recommendation System with Auxiliary Information

Context-aware recommendation systems improve upon classical recommender systems by including, in the modelling, a user's behaviour. Research into context-aware recommendation systems has previously only considered the sequential ordering of items as contextual information. However, there is a wealth of unexploited additional multi-modal information available in auxiliary knowledge related to items. This study extends the existing research by evaluating a multi-modal recommendation system that exploits the inclusion of comprehensive auxiliary knowledge related to an item. The empirical results explore extracting vector representations (embeddings) from unstructured and structured data using data2vec. The fused embeddings are then used to train several state-of-the-art transformer architectures for sequential user-item representations. The analysis of the experimental results shows a statistically significant improvement in prediction accuracy, which confirms the effectiveness of including auxiliary information in a context-aware recommendation system. We report a 4% and 11% increase in the NDCG score for long and short user sequence datasets, respectively.

cs.IR↗

arXiv2022

Pareto Driven Surrogate (ParDen-Sur) Assisted Optimisation of Multi-period Portfolio Backtest Simulations

Portfolio management is a multi-period multi-objective optimisation problem subject to a wide range of constraints. However, in practice, portfolio management is treated as a single-period problem partly due to the computationally burdensome hyper-parameter search procedure needed to construct a multi-period Pareto frontier. This study presents the \gls{ParDen-Sur} modelling framework to efficiently perform the required hyper-parameter search. \gls{ParDen-Sur} extends previous surrogate frameworks by including a reservoir sampling-based look-ahead mechanism for offspring generation in \glspl{EA} alongside the traditional acceptance sampling scheme. We evaluate this framework against, and in conjunction with, several seminal \gls{MO} \glspl{EA} on two datasets for both the single- and multi-period use cases. Our results show that \gls{ParDen-Sur} can speed up the exploration for optimal hyper-parameters by almost $2\times$ with a statistically significant improvement of the Pareto frontiers, across multiple \glspl{EA}, for both datasets and use cases.

cs.NE↗

arXiv2022

Knowledge Graph Fusion for Language Model Fine-tuning

Language Models such as BERT have grown in popularity due to their ability to be pre-trained and perform robustly on a wide range of Natural Language Processing tasks. Often seen as an evolution over traditional word embedding techniques, they can produce semantic representations of text, useful for tasks such as semantic similarity. However, state-of-the-art models often have high computational requirements and lack global context or domain knowledge which is required for complete language understanding. To address these limitations, we investigate the benefits of knowledge incorporation into the fine-tuning stages of BERT. An existing K-BERT model, which enriches sentences with triplets from a Knowledge Graph, is adapted for the English language and extended to inject contextually relevant information into sentences. As a side-effect, changes made to K-BERT for accommodating the English language also extend to other word-based languages. Experiments conducted indicate that injected knowledge introduces noise. We see statistically significant improvements for knowledge-driven tasks when this noise is minimised. We show evidence that, given the appropriate task, modest injection with relevant, high-quality knowledge is most performant.

cs.CL↗

arXiv2022

Fusion of Sentiment and Asset Price Predictions for Portfolio Optimization

The fusion of public sentiment data in the form of text with stock price prediction is a topic of increasing interest within the financial community. However, the research literature seldom explores the application of investor sentiment in the Portfolio Selection problem. This paper aims to unpack and develop an enhanced understanding of the sentiment aware portfolio selection problem. To this end, the study uses a Semantic Attention Model to predict sentiment towards an asset. We select the optimal portfolio through a sentiment-aware Long Short Term Memory (LSTM) recurrent neural network for price prediction and a mean-variance strategy. Our sentiment portfolio strategies achieved on average a significant increase in revenue above the non-sentiment aware models. However, the results show that our strategy does not outperform traditional portfolio allocation strategies from a stability perspective. We argue that an improved fusion of sentiment prediction with a combination of price prediction and portfolio optimization leads to an enhanced portfolio selection strategy.

q-fin.PM↗

arXiv2021

A Statistics and Deep Learning Hybrid Method for Multivariate Time Series Forecasting and Mortality Modeling

Hybrid methods have been shown to outperform pure statistical and pure deep learning methods at forecasting tasks and quantifying the associated uncertainty with those forecasts (prediction intervals). One example is Exponential Smoothing Recurrent Neural Network (ES-RNN), a hybrid between a statistical forecasting model and a recurrent neural network variant. ES-RNN achieves a 9.4\% improvement in absolute error in the Makridakis-4 Forecasting Competition. This improvement and similar outperformance from other hybrid models have primarily been demonstrated only on univariate datasets. Difficulties with applying hybrid forecast methods to multivariate data include ($i$) the high computational cost involved in hyperparameter tuning for models that are not parsimonious, ($ii$) challenges associated with auto-correlation inherent in the data, as well as ($iii$) complex dependency (cross-correlation) between the covariates that may be hard to capture. This paper presents Multivariate Exponential Smoothing Long Short Term Memory (MES-LSTM), a generalized multivariate extension to ES-RNN, that overcomes these challenges. MES-LSTM utilizes a vectorized implementation. We test MES-LSTM on several aggregated coronavirus disease of 2019 (COVID-19) morbidity datasets and find our hybrid approach shows consistent, significant improvement over pure statistical and deep learning methods at forecast accuracy and prediction interval construction.

cs.LG↗

arXiv2021

Feature-weighted Stacking for Nonseasonal Time Series Forecasts: A Case Study of the COVID-19 Epidemic Curves

We investigate ensembling techniques in forecasting and examine their potential for use in nonseasonal time-series similar to those in the early days of the COVID-19 pandemic. Developing improved forecast methods is essential as they provide data-driven decisions to organisations and decision-makers during critical phases. We propose using late data fusion, using a stacked ensemble of two forecasting models and two meta-features that prove their predictive power during a preliminary forecasting stage. The final ensembles include a Prophet and long short term memory (LSTM) neural network as base models. The base models are combined by a multilayer perceptron (MLP), taking into account meta-features that indicate the highest correlation with each base model's forecast accuracy. We further show that the inclusion of meta-features generally improves the ensemble's forecast accuracy across two forecast horizons of seven and fourteen days. This research reinforces previous work and demonstrates the value of combining traditional statistical models with deep learning models to produce more accurate forecast models for time-series from different domains and seasonality.

cs.LG↗

arXiv2021

Multivariate Anomaly Detection based on Prediction Intervals Constructed using Deep Learning

It has been shown that deep learning models can under certain circumstances outperform traditional statistical methods at forecasting. Furthermore, various techniques have been developed for quantifying the forecast uncertainty (prediction intervals). In this paper, we utilize prediction intervals constructed with the aid of artificial neural networks to detect anomalies in the multivariate setting. Challenges with existing deep learning-based anomaly detection approaches include $(i)$ large sets of parameters that may be computationally intensive to tune, $(ii)$ returning too many false positives rendering the techniques impractical for use, $(iii)$ requiring labeled datasets for training which are often not prevalent in real life. Our approach overcomes these challenges. We benchmark our approach against the oft-preferred well-established statistical models. We focus on three deep learning architectures, namely, cascaded neural networks, reservoir computing and long short-term memory recurrent neural networks. Our finding is deep learning outperforms (or at the very least is competitive to) the latter.

cs.LG↗