Polychronis Koutsakis

This study was prompted by the scarcity of focused quantitative research on the cybersecurity of SMBs. Our research aimed to understand the factors influencing SMBs' approach to cybersecurity, their level of threat awareness and the importance placed on cybersecurity. It also explored the extent to which NIST CSF practices are implemented by SMBs while also detecting and ranking the prevalent challenges faced by SMBs. Additionally, resources that SMBs turn to for help and guidance were also evaluated. While the survey-based study was on Western Australian SMBs, the results are of more general and wider interest. Our study found the lack of funds to be the biggest hindrance to cybersecurity, along with a lack of knowledge on where to start implementing good security practices. SMBs also lacked familiarity with relevant regulations and frameworks. The study highlights areas for improvement, such as access control mechanisms, individual user accounts, formalised policies and procedures, and dedicated budgets. SMBs heavily rely on Google search for cybersecurity information, emphasising the need for optimised search results from authoritative sources. IT service providers and informal networks also emerge as important sources of cybersecurity guidance, while local universities could assist SMBs but remain underutilised in this regard. Interestingly, factors such as organisational size, industry sector, and revenue level did not significantly impact SMBs' perception of vulnerability to cyber threats. However, further investigation is needed to evaluate the effectiveness of different IT service models for SMBs' cybersecurity needs. Overall, the research provides valuable insights into the specific gaps and challenges faced by SMBs in the cybersecurity domain, as well as their preferred methods of seeking and consuming cybersecurity assistance. The findings can guide the development of targeted strategies and policies to enhance the cybersecurity posture of SMBs.

Luck (2024) argues that being “rightly dominated” by a benevolent super-intelligent AI (super-AI) that would optimise our freedom by allowing as much as we need but not more, may be a good thing and provides analogies between God and super-AI playing this role. Our view is that such an approach is highly problematic if we have a grasp of how super-AI works (we will call this option one) and even more problematic if we don’t (option two)...

Deep learning (DL) has achieved superior classification in many applications due to its capability of extracting features from the data. However, the success of DL comes with the tradeoff of possible overfitting. The bias towards the data it has seen during the training process leads to poor generalisation. One way of solving this issue is by having enough training data so that the classifier is invariant to many data patterns. In the literature, data augmentation has been used as a type of regularisation method to reduce the chance for the model to overfit. However, most of the relevant works focus on image, sound or text data. There is not much work on numerical data augmentation, although many real-world problems deal with numerical data. In this paper, we propose using a technique based on Fuzzy C-Means clustering and fuzzy membership grades. Fuzzy-related techniques are used to address the variance problem by generating new data items based on fuzzy numbers and each data item's belongings to different fuzzy clusters. This data augmentation technique is used to improve the generalisation of a Deep Neural Network that is suitable for numerical data. By combining the proposed fuzzy data augmentation technique with the Dropout regularisation technique, we manage to balance the classification model's bias-variance tradeoff. Our proposed technique is evaluated using four popular data sets and is shown to provide better regularisation and higher classification accuracy compared with popular regularisation approaches.

The constant increase in users’ bandwidth needs, through a large variety of multimedia applications, creates the need for highly effective network traffic control. This need is imperative in wireless networks, where the available bandwidth is limited, but is very important for wired networks as well. In this work we focus on the problem of policing video traffic from sources encoded with H.264 and H.265, given that these are the major state-of-the-art standards currently in the market. Building on work that has shown that classic traffic policing schemes can lead to unnecessarily strict policing for conforming video sources, we propose the use of Artificial Intelligence (AI) – based traffic policing schemes for video traffic. We conduct a performance evaluation of several AI – based schemes with the classic token bucket and we show that our proposed Frame Size Predictor and Policer scheme improve the performance of the classic token bucket by around 90% for conforming users, while providing only slightly worse policing results for non-conforming users.

Small-to-medium sized businesses (SMBs) constitute a large fraction of many countries’ economies but according to the literature SMBs are not adequately implementing cyber security which leaves them susceptible to cyber-attacks. Furthermore, research in cyber security is rarely focused on SMBs, despite them representing a large proportion of businesses. In this paper we review recent research on the cyber security of SMBs, with a focus on the alignment of this research to the popular NIST Cyber Security Framework (CSF). From the literature we also summarise the key challenges SMBs face in implementing good cyber security and conclude with key recommendations on how to implement good cyber security. We find that research in SMB cyber security is mainly qualitative analysis and narrowly focused on the Identify and Protect functions of the NIST CSF with very little work on the other existing functions. SMBs should have the ability to detect, respond and recover from cyber-attacks, and if research lacks in those areas, then SMBs may have little guidance on how to act. Future research in SMB cyber security should be more balanced and researchers should adopt well-established powerful quantitative research approaches to refine and test research whilst governments and academia are urged to invest in incentivising researchers to expand their research focus.

As video has become the dominant type of traffic over wired and wireless networks, the efficient transmission of video streams is of paramount importance. Hence, especially for wireless networks, the optimum utilization of the available bandwidth while preserving the users’ Quality of Service and Quality of Experience requirements is crucial. Towards this goal, the accurate prediction of upcoming video frame sizes can play a significant role. This work focuses on achieving such an accurate prediction for videos encoded with H.264 and H.265, which are the major state-of-the-art standards based on their current market share. Unlike previous studies, we use single-step and multi-step approaches to capture the long-range dependence and short-range dependence properties of variable bit rate video traces through neural networks-based modeling. We evaluate the accuracy of Long Short Term Memory, Convolutional Neural Networks and Sequence-to-Sequence models and compare them with existing approaches. Our models show significantly higher accuracy for a variety of videos. We also provide a case study on how our model can be used for traffic policing purposes.

Class imbalance is a serious issue in classification as a traditional classifier is generally biased towards the majority class. The accuracy of the classifier could be further impacted in cases where additionally to the class imbalance, there are overlapped data instances. Further, data sparsity has shown to be a possible issue that may lead to non- invariance and poor generalisation. Data augmentation is a technique that can handle the generalisation issue and improve the regularisation of the Deep Neural Network (DNN). A method to handle both class overlap and class imbalance while also incorporating regularisation is proposed in this paper. In our work, the imbalanced dataset is balanced using SMOTETomek, and then the non-categorical attributes are fuzzified. The purpose of fuzzifying the attributes is to handle the overlapping in the data and provide some form of data augmentation that can be used as a regularisation technique. Therefore, in this paper, the invariance is achieved as the augmented data are generated based on the fuzzy concept. The balanced augmented dataset is then trained using a DNN classifier. The datasets used in the experiments were selected from UCI and KEEL data repositories. The experiments show that the proposed Fuzzy data augmentation for handling overlapped and imbalanced data can address the overlapped and imbalanced data issues, and provide regularisation using data augmentation for numerical data to improve the performance of a DNN classifier.

Background and Aims Substantial scientific evidence supports the effectiveness of a Mediterranean diet (MedDiet) in managing type 2 diabetes mellitus (T2DM). Potential benefits of time restricted feeding (TRF) in T2DM are unknown. The MedDietFast trial aims to investigate the efficacy of a MedDiet with or without TRF compared to standard care diet in managing T2DM. Methods and Results 120 adults aged 20 -75 with a body mass index (BMI) of 20-35 kg/m2 and T2DM will be randomised in a 3-arm parallel design to follow an ad libitum MedDiet with or without 12-hours TRF or the standard Australian Dietary Guidelines (ADG) for 24 weeks. All groups will receive dietary counselling fortnightly for 12 weeks and monthly thereafter. The primary outcome is changes in HbA1c from baseline to 12 and 24 weeks. Secondary outcomes include fasting blood glucose, insulin, blood lipids, weight loss, insulin resistance index (HOMA), Glucagon-like peptide 1 (GLP-1) and high-sensitivity C- reactive protein (hs-CRP). Data on medical history, anthropometry, wellbeing, MedDiet adherence and satiety will be measured at a private clinic via self-report questionnaires at baseline, 6, 12 and 24 weeks. Additionally, specimens (blood, urine and stool) will be collected at all time points for future omics analysis. Conclusion The MedDietFast trial will examine the feasibility and effectiveness of a MedDiet with/without TRF in T2DM patients. Potential synergistic effects of a MedDiet with TRF will be evaluated. Future studies will generate microbiomic and metabolomic data for translation of findings into simple and effective management plans for T2DM patients.

Wifi-based localisation systems have gained significant interest with many researchers proposing different localisation techniques using publicly available datasets. However, these datasets are limited because they only contain Wifi fingerprints collected and labelled by users, and they are restricted to indoor locations. We have generated the first Wifi-based localisation datasets for a GPS-denied open area. We selected a busy open area at Murdoch University to generate the datasets using so-called “smart bins”, which are rubbish bins that we enabled to work as access points. The data gathered consists of two different datasets. In the first, four users generated labelled WiFi fingerprints for all available Reference Points using four different smartphones. The second dataset includes 2450865 auto-generated rows received from more than 1000 devices. We have developed a light-weight algorithm to label the second dataset from the first and we proposed a localisation approach that converts the second dataset from asynchronous format to synchronous, applies feature engineering and a deep learning classifier. Finally, we have demonstrated via simulations that by using this approach we achieve higher prediction accuracy, with up to 19% average improvement, compared with using only the fingerprint dataset.

Analysis of time series data has been a challenging research subject for decades. Email traffic has recently been modelled as a time series function using a Recurrent Neural Network (RNN) and RNNs were shown to provide higher prediction accuracy than previous probabilistic models from the literature. Given the exponential rise of email workloads which need to be handled by email servers, in this paper we first present and discuss the literature on modelling email traffic. We then explain the advantages and limitations of different approaches as well as their points of agreement and disagreement. Finally, we present a comprehensive comparison between the performance of RNN and Long Short Term Memory (LSTM) models. Our experimental results demonstrate that both approaches can achieve high accuracy over four large datasets acquired from different universities’ servers, outperforming existing work, and show that the use of LSTM and RNN is very promising for modelling email traffic.