South African Tuberculosis Vaccine Initiative
UCT Computer Science Research Document ArchiveNot a member yet
1270 research outputs found
Sort by
Evolving Swarm-Robotic Behavioral Allocations
Full text linkThis study investigates comparative methods for two-step collective
behavior evolution (evolving group behaviors from pre-evolved
behaviors), to encourage the evolution of behavioral diversity in
swarm-robotic applications. Specifically, we investigate behavioral
diversity evolution given pre-evolved behaviors in collective behaviors
that are effective across increasingly complex and difficult
collective herding task environments. Results indicate that specific
complements of pre-evolved (lower task-performance) collective
herding behaviors was suitable for achieving high task performance
across all environments and task difficulty levels. Results support
the efficacy of the two-step approach for evolving behaviorally
heterogeneous groups in collective behavior tasks that benefit from
groups comprising various complementary behaviors
Semi-automatic linguistic annotation for lexicography with machine learning
Full text linkMany terminology lists for South African languages provide only the translations of each term without grammatical information such as part-of-speech and noun class. This makes it difficult to know how to use these words correctly in context. Annotating these terminology lists manually requires linguistic expertise and can be costly, time-intensive, and error prone. Instead, we propose annotating these terminology lists using a machine-learning classifier. An expert will then review the generated output to ensure accuracy. We will apply this approach to isiXhosa and integrate the results into the IsiXhosa.click online dictionary (Marquard 2024). This progresses the annotation of lexicographic works by making it easier to input linguistic information in dictionaries
Engagement of Parents with the Aurora Child Health Chatbot: A Conversation Log Analysis Study
Full text linkChatbots have the potential to support child health by answering parents doubts and providing tailored information. However, prior work has not studied the deployment of chatbots for this setting. We analysed how parents used the Aurora Facebook Messenger chatbot, designed for Portuguese parents, with an optional subscription to professional support. Our analysis investigated chatbot use, discussed topics, and conversation topics, drawing on user engagement and conversation metrics, text-mining, user satisfaction scores, and conversation content analysis. Results revealed 718 active users (out of 1043), with peak activity during lunchtime and late at night. Most queries pertained to critical situations, including infant sleep (80%), (breast)feeding (13%), or healthcare-related issues (7%). Aurora handled in-domain questions appropriately, but struggled to answer multi-topic queries. Subscription users had 243% more interactions and 162% more extended use of the chatbot. Our research underscores the importance of offering timely and personalised messaging to meet parents’ needs
On the Feasibility of LLM-based Automated Generation and Filtering of Competency Questions for Ontologies
No full textCompetency questions for ontologies are used in a number of ontology development tasks. The questions’ sentences structure have been analysed to inform ontology authoring and validation. One of the problems to make this a seamless process is the hurdle of writing good CQs manually or offering automated assistance in writing CQs. In this paper, we propose an enhanced and automated pipeline where one can trace meticulously through each step, using a mini-corpus, T5, and the SQuAD dataset to generate questions, and the CLaRO controlled language, semantic similarity, and other steps for filtering. This was evaluated with two corpora of different genre in the same broad domain and evaluated with domain experts. The final output questions across the experiments were around 25% for scope and relevance and 45% of unproblematic quality. Technically, it provided ample insight into trade-offs in generation and filtering, where relaxing filtering increased sentence structure diversity but also led to more spurious sentences that required additional processing
Move With Me: Co-designing a Tangible User Interface To Promote Physical Activity Among Rural South African Children
Full text linkChildhood physical inactivity and obesity are growing concerns globally, including in rural South Africa, where children often lack access to safe play spaces and well-resourced Early Childhood Development (ECD) centers and face increasing screen time. These challenges limit opportunities for physical development in early childhood. To address this, we developed Move With Me, a wearable tangible user interface (TUI) in the form of a superhero cape that encourages physical activity as a probe for further investigation. While many commercial technologies that encourage movement exist, they are unsuitable to rural communities due to their high cost and reliance on stable internet and electricity infrastructure. In response to these challenges, we co-designed a more contextually relevant solution with rural South African mothers using the technology probe to support the ideation and co-design process. The final cape design integrates motion sensors, LED lights, Bluetooth Low Energy (BLE), and a solar rechargeable battery pack, delivering real-time feedback without internet connectivity or stable electricity. A companion mobile app gamifies movement with culturally relevant animations to engage children. We found that co-design empowered mothers to tailor the technology to their context, suggesting affordable components and re-purposing existing smartphone features such as sound and animations instead of using costly electronics incorporated into the cape. Their contributions led to a low-cost, offline-capable, and personalized TUI. Our work demonstrates how wearable TUIs, developed through inclusive design, can support physical activity in resource-limited settings. We contribute practical insights for designing sustainable technologies for rural contexts, emphasizing affordability, low power consumption, offline functionality, and the value of embedding co-creation throughout the design process
Evolutionary Deep-Learning Malware Classifiers
Full text linkMalware attacks remain a critical cyber-security concern, necessitating robust solutions for both individual users and organizations. Deep learning methods have become pervasive
tools for malware detection and classification. However, the evolution of malware into sophisticated forms that aim to elude detection poses a formidable challenge to traditional deep learning methods. Existing techniques for generating adversarial
samples often rely on manual feature extraction and white-box models, introducing a gap between the generated samples and real-world scenarios. In response to these challenges, we propose an innovative approach leveraging evolutionary learning for the generation of adversarial samples. Our approach uses a three step process for malware detection. First, a trained deep-learning malware classifier categorizes samples as benign or malicious.
Second, an evolutionary adversarial learning approach trains and generates new malware samples. Third, competitive coevolution facilitates automated adaptation of malware detection agents that are robust against attacks. We evaluate the efficacy of our approach for adaptive malware detection via benchmark evaluations with an established deep-learning classifier
Designing and Contextualising Probes for African Languages
Full text linkPretrained language models (PLMs) for African languages are continually improving, but the reasons behind these advances remain unclear. This paper presents the first systematic investigation into how knowledge about African languages is encoded in PLMs. We train layer-wise probes for six typologically diverse African languages to analyse how linguistic features are distributed. We also design control tasks, a way to interpret probe performance, for the MasakhaPOS dataset. We find PLMs adapted for African languages to encode more linguistic information about target languages than massively multilingual PLMs. Our results reaffirm previous findings that token-level syntactic information concentrates in middle-to-last layers, while sentence-level semantic information is distributed across all layers. Through control tasks and probing baselines, we confirm that performance reflects the internal knowledge of PLMs rather than probe memorisation. Our study applies established interpretability techniques to African-language PLMs. In doing so, we highlight the internal mechanisms underlying the success of strategies like active learning and multilingual adaptation
Cross-Lingual Knowledge Projection and Knowledge Enhancement for Zero-Shot Question Answering in Low-Resource Languages
Full text linkKnowledge bases (KBs) in low-resource languages (LRLs) are often incomplete, posing a challenge for developing effective question answering systems over KBs in those languages. On the other hand, the size of training corpora for LRL language models is also limited, restricting the ability to do zero-shot question answering using multilingual language models.
To address these issues, we propose a two-fold approach. First, we introduce LeNS-Align, a novel cross-lingual mapping technique which improves the quality of word alignments extracted from parallel English-LRL text by combining lexical alignment, named entity recognition, and semantic alignment. LeNS-Align is applied to perform cross-lingual projection of KB triples.
Second, we leverage the projected KBs to enhance multilingual language models' question answering capabilities by augmenting the models with Graph Neural Networks embedding the projected knowledge.
We apply our approach to map triples from two existing English KBs, ConceptNet and DBpedia, to create comprehensive LRL knowledge bases for four low-resource South African languages.
Evaluation on three translated test sets show that our approach improves zero-shot question answering accuracy by up to 17\% compared to baselines without KB access. The results highlight how our approach contributes to bridging the knowledge gap for low-resource languages by expanding knowledge coverage and question answering capabilities
Preparing an isiNdebele LSP Dictionary: From Idea to design
Full text linkThis article discusses the plan for the compilation of an online IsiNdebele Dictionary, an isiNdebele Language Special Purposes (LSP). IsiNdebele is a lesser resourced language with a very limited corpora available. The isiNdebele LSP dictionary will be the first online dictionary that will be compiled by the isiNdebele Dictionary Unit. The software for the proposed online dictionary will be adapted from the isiZulu. The dictionary will be made available through an online software interface. It will be an open source and will be tailored to the isiNdebele grammar, bird terminology list, covid-19 terminology list and other terminology lists. Its online interface will be in isiNdebele and it is appropriate for mobile use
Automating Damage Recovery in a Legged Robot
Full text linkAutonomous robots are increasingly used in remote
and hazardous environments, where automated recovery given
damage to sensory-actuator systems would be extremely beneficial.
Such robots must therefore have controllers that continue
to function effectively given unexpected hardware malfunctions
and damage. We evaluate various controller types (oscillatorstyle
central pattern generators and artificial neural networks),
for producing adaptable gait behaviors. These controller types
are run for hexapod robot gait control in concert with the
Intelligent Trial and Error (IT&E) and Map-Elites algorithm
to maintain behavioral diversity. Specifically, we investigate the
impact of behavior map-size in MAP-Elites (the first phase of
the IT&E algorithm), in company with various controller types
for multiple leg failures scenarios using a simulated hexapod
robot. Results support previous work demonstrating a trade-off
between adapted gait speed and controller adaptability across
leg-damage scenarios, where map-size is crucial for generating
behavioral diversity required for adaptation