Search

Modeling cyber risks has been an important but challenging task in the domain of cyber security. It is mainly because of the high dimensionality and heavy tails of risk patterns. Those obstacles have hindered the development of statistical modeling of the multivariate cyber risks. In this work, we propose a novel approach for modeling the multivariate cyber risks which relies on the deep learning and extreme value theory. The proposed model not only enjoys the high accurate point predictions via deep learning but also can provide the satisfactory high quantile prediction via extreme value theory. The simulation study shows that the proposed model can model the multivariate cyber risks very well and provide satisfactory prediction performances. The empirical evidence based on real honeypot attack data also shows that the proposed model has very satisfactory prediction performances., 25 pages

Causality is an intriguing concept that once tamed, can have many applications. While having been widely investigated in other domains, its relevance and usefulness in the cybersecurity domain has received little attention. In this paper, we present a systematic investigation of a particular approach to causality, known as Granger causality (G-causality), in cybersecurity. We propose a framework, dubbed Cybersecurity Granger Causality (CGC), for characterizing the presence of G-causality in cyber attack rate time series and for leveraging G-causality to predict (i.e., forecast) cyber attack rates. The framework offers a range of research questions, which can be adopted or adapted to study G-causality in other kinds of cybersecurity time series data. In order to demonstrate the usefulness of CGC, we present a case study by applying it to a particular cyber attack dataset collected at a honeypot. From this case study, we draw a number of insights into the usefulness and limitations of G-causality in the cybersecurity domain.

It is important to understand to what extent, and in what perspectives, cyber attacks can be predicted. Despite its evident importance, this problem was not investigated until very recently, when we proposed using the innovative methodology of {\em gray-box prediction}. This methodology advocates the use of gray-box models, which accommodate the statistical properties/phenomena exhibited by the data. Specifically, we showed that gray-box models that accommodate the Long-Range Dependence (LRD) phenomenon can predict the attack rate (i.e., the number of attacks per unit time) 1-hour ahead-of-time with an accuracy of 70.2-82.1\%. To the best of our knowledge, this is the first result showing the feasibility of prediction in this domain. We observe that the prediction errors are partly caused by the models' incapability in predicting the large attack rates, which are called {\em extreme values} in statistics. This motivates us to analyze the {\em extreme-value phenomenon}, by using two complementary approaches: the Extreme Value Theory (EVT) and the Time Series Theory (TST). In this paper, we show that EVT can offer long-term predictions (e.g., 24-hour ahead-of-time), while gray-box TST models can predict attack rates 1-hour ahead-of-time with an accuracy of 86.0-87.9\%. We explore connections between the two approaches, and point out future research directions. Although our prediction study is based on specific cyber attack data, our methodology can be equally applied to analyze any cyber attack data of its kind.

Background Pharmacy databases are commonly used to assess medication usage, and a number of measures have been developed to measure patients’ adherence to medication. An extensive literature now supports these measures, although few studies have systematically compared the properties of different adherence measures. Methods As part of an 18-month randomized clinical trial to assess the impact of automated telephone reminders on adherence to inhaled corticosteroids (ICS) among 6903 adult members of a managed care organization, we computed eight pharmacy-based measures of ICS adherence using outpatient pharmacy dispensing records obtained from the health plan’s electronic medical record. We used simple descriptive statistics to compare the relative performance characteristics of these measures. Results Comparative analysis found a relative upward bias in adherence estimates for those measures that require at least one dispensing event to be calculated. Measurement strategies that require a second dispensing event evidence even greater upward bias. These biases are greatest with shorter observation times. Furthermore, requiring a dispensing to be calculated meant that these measures could not be defined for large numbers of individuals (17-32 % of participants in this study). Measurement strategies that do not require a dispensing event to be calculated appear least vulnerable to these biases and can be calculated for everyone. However they do require additional assumptions and data (e.g., pre-intervention dispensing data) to support their validity. Conclusions Many adherence measures require one, or sometimes two, dispensings in order to be defined. Since such measures assume all dispensed medication is used as directed, they have a built in upward bias that is especially pronounced when they are calculated over relatively short timeframes (< 9 months). Less biased measurement strategies that do not require a dispensing event are available, but require additional data to support their validity. Trial registration The study was funded by grant R01HL83433 from the National Heart, Lung and Blood Institute (NHLBI) and is filed as study NCT00414817 in the clinicaltrials.gov database.

Quantitative security analysis of networked computer systems is one of the decades-long open problems in computer security. Recently, a promising approach was proposed in \cite{XuTDSC11}, which however made some strong assumptions including the exponential distribution of, and the independence between, the relevant random variables. In this paper, we substantially weaken these assumptions while offering, in addition to the same types of analytical results as in \cite{XuTDSC11}, methods for obtaining the desired security quantities in practice. Moreover, we investigate the problem from a higher-level abstraction, which also leads to both analytical results and practical methods for obtaining the desired security quantities. These would represent a significant step toward ultimately solving the problem of quantitative security analysis of networked computer systems.

In this paper, some ordering properties of convolutions of heterogeneous Bernoulli random variables are discussed. It is shown that, under some suitable conditions, the likelihood ratio order and the reversed hazard rate order hold between convolutions of two heterogeneous Bernoulli sequences. The results established here extend and strengthen the previous results of Pledger and Proschan (1971) and Boland, Singh and Cukic (2002).

Kochar and Xu (2009) proved that a parallel system with heterogeneous exponential component lifetimes is more skewed (according to the convex transform order) than the system with independent and identically distributed exponential components. In this paper we extend this study to the general k-out-of-n systems for the case when there are only two types of component in the system. An open problem proposed in Pǎltǎnea (2008) is partially solved.