A5016115576- Advanced Database Systems and Queries
- Semantic Web and Ontologies
- Data Management and Algorithms
- Data Stream Mining Techniques
- Data Mining Algorithms and Applications
- Anomaly Detection Techniques and Applications
- Rough Sets and Fuzzy Logic
- Adversarial Robustness in Machine Learning
- Software Engineering Research
- Imbalanced Data Classification Techniques
- Machine Learning and Data Classification
- Software Testing and Debugging Techniques
- Time Series Analysis and Forecasting
- Privacy-Preserving Technologies in Data
- Advanced Clustering Algorithms Research
- Network Security and Intrusion Detection
- Complex Network Analysis Techniques
- Recommender Systems and Techniques
- Service-Oriented Architecture and Web Services
- Topic Modeling
- Spam and Phishing Detection
- Data Quality and Management
- Algorithms and Data Compression
- Natural Language Processing Techniques
- Logic, Reasoning, and Knowledge
University of Auckland
2016-2025
Macquarie University
2008
Victoria University of Wellington
1995-2002
Mississippi State University
2002
National University of Singapore
2000
The University of Melbourne
1993
Federated learning (FL) has emerged as a promising privacy-aware paradigm that allows multiple clients to jointly train model without sharing their private data. Recently, many studies have shown FL is vulnerable membership inference attacks (MIAs) can distinguish the training members of given from non-members. However, existing MIAs ignore source member, i.e., information client owning while it essential explore privacy in beyond examples all clients. The leakage lead severe issues. For...
Large Language Models have shown tremendous performance on a large variety of natural language processing tasks, ranging from text comprehension to common sense reasoning. However, the mechanisms responsible for this success remain opaque, and it is unclear whether LLMs can achieve human-like cognitive capabilities or these models are still fundamentally circumscribed. Abstract reasoning fundamental task cognition, consisting finding applying general pattern few data. Evaluating deep neural...
Current drift detection techniques detect a change in distribution within stream. However, there are no current that analyze the rate of these detected changes. We coin term stream volatility, to describe changes A has high volatility if frequently and low infrequently. particularly interested shift which is (e.g. From volatility). introduce define concept propose novel technique on data streams presence drifts. In experiments we show our algorithm be both fast efficient. also new for called...
Network embedding learns the vector representations of nodes. Most real world networks are heterogeneous and evolve over time. There are, however, no network approaches designed for dynamic so far. Addressing this research gap is beneficial analyzing mining networks. We develop a novel representation learning method, change2vec, which considers as snapshots with different time stamps. Instead processing whole at each stamp, change2vec models changes between two consecutive static by...
In recent years, the integration of machine learning techniques into chemical reaction product prediction has opened new avenues for understanding and predicting behaviour substances. The necessity such predictive methods stems from growing regulatory social awareness environmental consequences associated with persistence accumulation residues. Traditional biodegradation rely on expert knowledge to perform predictions. However, creating this is becoming increasingly prohibitive due...
Forecasting surges in hospital admissions caused by severe respiratory infections is of crucial importance during the winter season to enable proactive management and timely decision-making prevent healthcare system overload. As time series derived from surveillance systems for these cases are sparse encode weak seasonality patterns, machine learning key computing accurate forecasts. The most recent algorithmic advance forecasting adaptation generative pre-trained transformers (GPTs). Those...
<title>Abstract</title> Bias in machine learning models remains a critical challenge, particularly datasets with numeric features where discrimination may be subtle and hard to detect. Existing fairness frameworks rely on expert knowledge of marginalized groups, such as specific racial categorical defining them. Furthermore, most evaluate bias rather than datasets, despite the fact that model can often traced back dataset shortcomings. Our research aims remedy this gap by capturing flaws set...
Recommender systems using Collaborative Filtering techniques are capable of make personalized predictions. However, these highly vulnerable to profile injection attacks. Group attacks that target a group items instead one, and there common attributes among items. Such profiles will have good probability being similar large number user profiles, making them hard detect. We propose novel technique for identifying attack which uses an improved metric based on Degree Similarity with Top...
Recommender systems are highly vulnerable to shilling attacks, both by individuals and groups. Attackers who introduce biased ratings in order affect recommendations, have been shown negatively collaborative filtering (CF) algorithms. Previous research focuses only on the differences between genuine profiles attack profiles, ignoring group characteristics profiles. In this paper, we study use of statistical metrics detect rating patterns attackers Another question is that most existing...
By developing awareness of smartphone activities that the user is performing on their smartphone, such as scrolling feeds, typing and watching videos, we can develop application features are beneficial to users, personalization. It currently not possible access real-time directly, due standard privileges if internal movement sensors detect them, there may be implications for policies. Our research seeks understand whether sensor data from existing inertial measurement unit (IMU) (triaxial...
Machine learning (ML) models have been widely applied to various applications, including image classification, text generation, audio recognition, and graph data analysis. However, recent studies shown that ML are vulnerable membership inference attacks (MIAs), which aim infer whether a record was used train target model or not. MIAs on can directly lead privacy breach. For example, via identifying the fact clinical has associated with certain disease, an attacker owner of disease high...
Recently issued data privacy regulations like GDPR (General Data Protection Regulation) grant individuals the right to be forgotten. In context of machine learning, this requires a model forget about training sample if requested by owner (i.e., unlearning). As an essential step prior unlearning, it is still challenge for tell whether or not her have been used unauthorized party train learning model. Membership inference recently emerging technique identify was target model, and seems...
Clustering is an important data mining task and has been explored extensively by a number of researchers for different application areas such as finding similarities in images, text bio-informatics data. Various optimization techniques have proposed to improve the performance clustering algorithms. In this paper we propose novel algorithm that call evolutionary particle swarm (EPSO)-clustering which based on PSO. The evolution generations where particles are initially uniformly distributed...