Differentially private random forest with high utility

Privacy-preserving data mining has become an active focus of the research community in the domains where data are sensitive and personal in nature. For example, highly sensitive digital repositories of medical or financial records offer enormous values for risk prediction and decision making. However, prediction models derived from such repositories should maintain strict privacy of individuals. We propose a novel random forest algorithm under the framework of differential privacy. Unlike previous works that strictly follow differential privacy and keep the complete data distribution approximately invariant to change in one data instance, we only keep the necessary statistics (e.g. variance of the estimate) invariant. This relaxation results in significantly higher utility. To realize our approach, we propose a novel differentially private decision tree induction algorithm and use them to create an ensemble of decision trees. We also propose feasible adversary models to infer about the attribute and class label of unknown data in presence of the knowledge of all other data. Under these adversary models, we derive bounds on the maximum number of trees that are allowed in the ensemble while maintaining privacy. We focus on binary classification problem and demonstrate our approach on four real-world datasets. Compared to the existing privacy preserving approaches we achieve significantly higher utility.

The assessment of denial, cognitive distortions and victim empathy among pedophilic sex offenders. An evaluation of the utility of self-report measures.

The modification of denial, defensiveness, and cognitive distortions and the enhancement of victim empathy are central components in the treatment of pedophilic sex offenders (PSOs) and are thus important factors to evaluate. This review of the literature highlights three broad problems with self-report measures of these variables. First, the psychometric properties of measures vary enormously, with some having no established validity or reliability. Second, the purpose of the measure is generally quite transparent, enabling the respondent to easily pick the socially acceptable responses. Finally, it is difficult to determine which are the best measures to use in assessing PSOs. Measures range from those designed for the general public to those designed specifically for PSOs. Also, they range from those that assess broad processes (e.g., general empathy) to those that assess offensespecific variables (e.g., victim empathy). This article argues that these issues need to be addressed to improve both the assessment of these processes among PSOs and the evaluation of treatment programs for PSOs.

Generalized pseudo-likelihood estimates for markov random fields on lattice

In this paper we generalize Besag's pseudo-likelihood function for spatial statistical models on a region of a lattice. The correspondingly defined maximum generalized pseudo-likelihood estimates (MGPLEs) are natural extensions of Besag's maximum pseudo-likelihood estimate (MPLE). The MGPLEs connect the MPLE and the maximum likelihood estimate. We carry out experimental calculations of the MGPLEs for spatial processes on the lattice. These simulation results clearly show better performances of the MGPLEs than the MPLE, and the performances of differently defined MGPLEs are compared. These are also illustrated by the application to two real data sets.

Utility of mitochondrial DNA sequences from four gene regions for systematic studies of Australian freshwater crayfish of the Genus Cherax (Decapoda: Parastacidae)

One of the most important requirements for systematic and phylogenetic studies is the identification of gene regions with the appropriate level of variation for the question of interest. Molecular phylogenetic and systematic studies of freshwater crayfish have made use of DNA sequences mainly from ribosomal genes, especially the 16S rRNA gene region. Thus, little information is available on other potentially useful mitochondrial gene regions for systematic studies in these animals. In this study, we look at nucleotide variation and phylogenetic relations within and between four species of freshwater crayfish of the genus Cherax from the southwest of Western Australia using four fragments amplified from the 16S rRNA, 12S rRNA, Cytochrome Oxidase I (COI), and Cytochrome b (Cyt b) gene regions. Samples of Engaeus strictifrons, Euastacus bispinosus, and Geocharax falcata were also sequenced for comparative purposes. The size of the fragments varied from 358 bp to 600 bp. Across all samples, the four fragments showed significant phylogenetic signal and showed similar proportions of variable sites (28.81–37.33%). Average divergence within species for the mitochondrial gene regions varied from 1.18% to 4.91%, with the 16S rRNA being the least variable and Cyt b the most variable. Average divergence between species ranged 7.63–15.53%, with 16S rRNA being the least variable and COI the most variable. At the generic level, average divergence ranged 17.21–23.82%. Phylogenetic analyses of the 16S rRNA, 12S rRNA, and COI regions generated four clades consistent with the presence of four species previously identified on the basis of allozyme and morphological studies. The relationships among samples were largely congruent across the data set, although some relationships remained unresolved. Not all samples could be amplified using the Cyt b primers, and some of those that were showed quite anomalous relationships, suggesting that one or more Cyt b pseudogenes were being amplified.

The assessment of motivation for behaviour change among sex offenders against children: an investigation of the utility of the Stages of Change Questionnaire

A better understanding of motivation for behaviour change among sex offenders against children would improve treatment programmes designed to modify sexual offending behaviour. However, investigation of this issue is limited by lack of theoretically and empirically sound measures of motivation for behaviour change among sex offenders. This paper reports on two studies that were conducted to investigate the psychometric properties (validity, reliability, and social desirability) of the Stages of Change Questionnaire, adapted to measure motivation for behaviour change among sex offenders against children. In Study 1, the psychometric properties of the questionnaire were assessed for sex offenders against children (n=36) who were on a treatment waiting list. In Study 2, the psychometric properties of the questionnaire were assessed for sex offenders against children (n=47) at pre-treatment, mid-treatment, and post-treatment. Both studies supported the validity and reliability of the adapted Stages of Change Questionnaire, and the influence of social desirability upon responding was less than expected. The results of this investigation supported the potential utility of the Stages of Change Questionnaire as a measure of motivation for behaviour change for sex offenders against children.

Chromatic and luminance losses with multiple sclerosis and optic neuritis measured using dynamic random luminance contrast noise

We measured thresholds for detecting changes in colour and in luminance contrast in observers with multiple sclerosis (MS) and/or optic neuritis (ON) to determine whether reduced sensitivity occurs principally in red-green or blue-yellow second-stage chromatic channels or in an achromatic channel. Colour thresholds for the observers with MS/ON were higher in the red-green direction than in the blue-yellow direction, indicating greater levels of red-green loss than blue-yellow loss. Achromatic thresholds were raised less than either red-green or blue-yellow thresholds, showing less luminance-contrast loss than chromatic loss. With the MS/ON observers, blue-yellow and red-green thresholds were positively correlated but increasing impairment was associated with more rapid changes in red-green thresholds than blue-yellow thresholds. These findings indicate that demyelinating disease selectively reduces sensitivity to colour vision over luminance vision and red-green colours over blue-yellow colours.

Class library ranlip for multivariate nonuniform random variate generation

This paper describes generation of nonuniform random variates from Lipschitz-continuous densities using acceptance/rejection, and the class library _ranlip which implements this method. It is assumed that the required distribution has Lipschitz-continuous density, which is either given analytically or as a black box. The algorithm builds a piecewise constant upper approximation to the density (the hat function), using a large number of its values and subdivision of the domain into hyperrectangles. The class library _ranlip provides very competitive preprocessing and generation times, and yields small rejection constant, which is a measure of efficiency of the generation step. It exhibits good performance for up to five variables, and provides the user with a black box nonuniform random variate generator for a large class of distributions, in particular, multimodal distributions. It will be valuable for researchers who frequently face the task of sampling from unusual distributions, for which specialized random variate generators are not available.

The utility of the trans-theoretical model of behavior change in the treatment of sex offenders

This paper discusses the use of the Trans-theoretical Model of Behavior Change in the treatment of sex offenders. Constructs within this theory are the Stages of Change, Processes of Change and Decisional Balance. The first section of this paper provides a brief description of these constructs. The second section provides a brief review of research related to these constructs and discusses the implications of this research in relation to the treatment of sex offenders. The third section of this paper provides a practical description of the use of the constructs of the Trans-theoretical Model of Behaviour Change in the treatment of sex offenders. Although the validity of this model among sex offenders requires further investigation, the Trans-theoretical Model of Behavior Change appears to have considerable utility as an overarching theoretical model to conceptualize and facilitate behavior change among sex offenders.

Universal nonuniform random vector generator based on acceptance-rejection

The acceptance/rejection approach is widely used in universal nonuniform random number generators. Its key part is an accurate approximation of a given probability density from above by a hat function. This article uses a piecewise constant hat function, whose values are overestimates of the density on the elements of the partition of the domain. It uses a sawtooth overestimate of Lipschitz continuous densities, and then examines all local maximizers of such an overestimate. The method is applicable to multivariate multimodal distributions. It exhibits relatively short preprocessing time and fast generation of random variates from a very large class of distributions

The high court and the utility of multiple judgments

Recently, the High Court has been criticised for its supposed increasing tendency to deliver multiple majority judgments. Ostensibly this impairs the capacity for the Court to clarify and unify the law, thereby making it more difficult for citizens to plan and coordinate their affairs. This criticism of the High Court is unsound. First, there is no evidence to suggest that the High Court is now more fragmented than it has been during other periods of its history. Secondly, the precise reasoning process (and the underlying jurisprudence reflected by this) is a cardinal aspect of the development of precedent and legal principle. Convergence in conclusion only is of little utility and does not promote certainty and clarity in the law. One cannot make an informed assessment of the impact and breadth of a decision without an understanding of the (actual) premise underpinning the decision. It is for this reason that legislation is such a poor vehicle for declaring the law and why in recent decades there has been an increasing degree of reliance on extraneous material to assist in the interpretation of legislation. Conclusion without (genuine) reasons is not highly instructive. Coerced agreement, no matter how subtle, is undesirable. The High Court should resist calls to deliver more single majority judgments.

Utility-driven evidence for healthy cities : problems with evidence generation and application

The question whether the WHO Healthy Cities project ‘works’ has been asked ever since a number of novel ideas and actions related to community health, health promotion and healthy public policy in the mid 1980s came together in the Healthy Cities Movement initiated by the World Health Organization. The question, however, has become more urgent since we have entered an era in which the drive for ‘evidence’ seems all-pervasive.

The article explores the nature of evidence, review available evidence on Healthy Cities accomplishments, and discusses whether enough evidence has been accumulated on different performances within the realm of Healthy Cities. A main point of reference is the European Healthy Cities Project (E-HCP).

Building on the information gathered through documentary research on the topic, it is concluded that there is fair evidence that Healthy Cities works. However, the future holds great challenges for further development and evidence-oriented evaluations of Healthy Cities. There are problems with (1) the communication of evidence, (2) the tension between the original intention of the Healthy Cities Movement and its current operations, and (3) the complex nature of Healthy Cities and the methodological tools currently available.

Reproducibility and clinical utility of tendon palpation to detect patellar tendinopathy in young basketball players

Background—Palpation is an important clinical test for jumper's knee.

Objectives—To (a) test the reproducibility of palpation tenderness, (b) evaluate the sensitivity and specificity of palpation in subjects with clinical symptoms of jumper's knee, and (c) determine whether tenderness to palpation may serve as a useful screening test for patellar tendinopathy. The yardstick for diagnosis of patellar tendinopathy was ultrasonographic abnormality.

Methods—In 326 junior symptomatic and asymptomatic athletes' tendons, palpation was performed by a single examiner before ultrasonographic examination by a certified ultrasound radiologist. In 58 tendons, palpation was performed twice to test reliability. Tenderness to palpation was scored on a scale from 0 to 3 where 0 represented no pain, and 1, 2, and 3 represented mild, moderate, and severe tenderness respectively.

Results—Patellar tendon palpation was a reliable examination for a single examiner (Pearson r = 0.82). In symptomatic tendons, the positive predictive value of palpation was 68%. As a screening examination in asymptomatic subjects, the positive predictive value of tendon palpation was 36–38%. Moderate and severe palpation tenderness were better predictors of ultrasonographic tendon pathology than absent or mild tenderness (p<0.001). Tender and symptomatic tendons were more likely to have ultrasound abnormality than tenderness alone (p<0.01).

Conclusions—In this age group, palpation is a reliable test but it is not cost effective in detecting patellar tendinopathy in a preparticipation examination. In symptomatic tendons, palpation is a moderately sensitive but not specific test. Mild tenderness in the patellar tendons in asymptomatic jumping athletes should be considered normal.

Evaluating public risk preferences in forest land-use choices using multi-attribute utility theory

Forest policy decisions inherently involve multiple attributes and risk and uncertainty as they largely deal with complex biological, ecological, and socio-political systems. Identifying risk preferences and quantifying their inter-relationships and tradeoffs are useful in formulating better forest policy. Often, technocrats and experts deal with risky decisions, but ideally, stakeholder risk characteristics should be explicitly considered in making policy decisions. This paper analysed societal risk preferences on public forest land-use attributes using multi-attribute utility theory (MAUT). The results indicate significant risk-averse behaviour towards old-growth forest conservation and forest-based recreation but less risk-averse behaviour towards native timber extraction. Overall, the respondents preferred a more conservative forest land-use option, which is consistent with their risk attitudes. The method provides insights into risk preferences of forest stakeholders, which could lead to better understanding of forest management conflicts. Moreover, the method explicitly distinguishes the technical and value components of the decision and is useful in unravelling public risk preferences in multiple-use forest planning situations.

The exchange rate disconnect puzzle: a resolution?

Early empirical studies of exchange rate determinants demonstrated that fundamentals-based monetary models were unable to outperform the benchmark random walk model in out-of-sample forecasts while later papers found evidence in favor of long-run exchange rate predictability. More recent theoretical works have adopted a microeconomic structure; a utility-based new open economy macroeconomic framework and a rational expectations present value model. Some recent empirical work argues that if the models are adjusted for parameter instability, it is a good predictor of nominal exchange rates while others use aggregate idiosyncratic volatility to generate good predictions. This latest research supports the idea that fundamental economic variables are likely to influence exchange rates especially in the long run and further that the emphasis should change to the economic value or utility based value to assess these macroeconomic models.