High accuracy android malware detection using ensemble learning

With over 50 billion downloads and more than 1.3 million apps in Google’s official market, Android has continued to gain popularity amongst smartphone users worldwide. At the same time there has been a rise in malware targeting the platform, with more recent strains employing highly sophisticated detection avoidance techniques. As traditional signature based methods become less potent in detecting unknown malware, alternatives are needed for timely zero-day discovery. Thus this paper proposes an approach that utilizes ensemble learning for Android malware detection. It combines advantages of static analysis with the efficiency and performance of ensemble machine learning to improve Android malware detection accuracy. The machine learning models are built using a large repository of malware samples and benign apps from a leading antivirus vendor. Experimental results and analysis presented shows that the proposed method which uses a large feature space to leverage the power of ensemble learning is capable of 97.3 % to 99% detection accuracy with very low false positive rates.

Modeling Linkage Disequilibrium Increases Accuracy of Polygenic Risk Scores

Polygenic risk scores have shown great promise in predicting complex disease risk and will become more accurate as training sample sizes increase. The standard approach for calculating risk scores involves linkage disequilibrium (LD)-based marker pruning and applying a p value threshold to association statistics, but this discards information and can reduce predictive accuracy. We introduce LDpred, a method that infers the posterior mean effect size of each marker by using a prior on effect sizes and LD information from an external reference panel. Theory and simulations show that LDpred outperforms the approach of pruning followed by thresholding, particularly at large sample sizes. Accordingly, predicted R(2) increased from 20.1% to 25.3% in a large schizophrenia dataset and from 9.8% to 12.0% in a large multiple sclerosis dataset. A similar relative improvement in accuracy was observed for three additional large disease datasets and for non-European schizophrenia samples. The advantage of LDpred over existing methods will grow as sample sizes increase.

Accuracy of rural refractionists in western China

Purpose. We assessed the prevalence and predictors of inaccurate refractive error among rural refractionists in western China. Methods. A subset of primary school children with visual acuity (VA) ≤6/12 in ≥1 eye, undergoing subjective refinement by local refractionists after cycloplegic autorefraction in an ongoing population-based study, received repeat refraction by university optometrists for quality control. Results. Among 502 children (mean age 10.5 years, 53.2% girls), independent predictors of poor (inaccurate by ≥1.0 diopter [D]) refraction by 21 rural practitioners (66.7% with high school or lower education) included hyperopia (odds ratio [OR], 4.2; 95% confidence interval [CI], 2.4-7.3, P < 0.001), astigmatism (OR = 3.8; 95% CI, 2.5-5.6; P < 0.001) and VA uncorrectable to >6/12 by the rural refractionist (OR = 4.7; 95% CI, 3.1-7.3; P = < 0.001). Among 201 children whose vision was uncorrectable in ≥1 eye by the rural refractionists, vision could be improved to >6/12 by the university optometrist in 110 (54.7%). We estimate vision could be so improved in 9.1% of all children refracted by these rural refractionists. A reason for inaccuracy in this setting is the erroneous tendency of rural refractionists to adjust instrument values for accommodation, even under cycloplegia. Conclusions. Rural refractionists in western China have little formal training and frequently fail to optimize VA among children, even when autorefractors are used. Training is needed emphasizing better use of automated refraction, particularly in children with astigmatism and hyperopia. © 2014 The Association for Research in Vision and Ophthalmology, Inc.

Strategies to improve the accuracy of vision measurement by teachers in rural Chinese secondary schoolchildren: Xichang Pediatric Refractive Error Study (X-PRES) report no. 6.

OBJECTIVE: To assess and improve the accuracy of lay screeners compared with vision professionals in detecting visual impairment in secondary schoolchildren in rural China. METHODS: After brief training, 32 teachers and a team of vision professionals independently measured vision in 1892 children in Xichang. The children also underwent vision measurement by health technicians in a concurrent government screening program. RESULTS: Of 32 teachers, 28 (87.5%) believed that teacher screening was worthwhile. Sensitivity (93.5%) and specificity (91.2%) of teachers detecting uncorrected presenting visual acuity of 20/40 or less were better than for presenting visual acuity (sensitivity, 85.2%; specificity, 84.8%). Failure of teachers to identify children owning but not wearing glasses and teacher bias toward better vision in children wearing glasses explain the worse results for initial vision. Wearing glasses was the student factor most strongly predictive of inaccurate teacher screening (P < .001). The sensitivity and specificity of the government screening program detecting low presenting visual acuity were 86.7% and 28.7%, respectively. CONCLUSIONS: Teacher vision screening after brief training can achieve accurate results in this setting, and there is support among teachers for screening. Screening of uncorrected rather than presenting visual acuity is recommended in settings with a high prevalence of corrected and uncorrected refractive error. Low specificity in the government program renders it ineffective.

Pupil constriction can alter the accuracy of dark room provocative test.

BACKGROUND: Primary angle-closure glaucoma (PACG) is a major cause of visual morbidity in East Asia. Dark-room provocative test (DRPT) has been used to determine which narrow angles have the risk to develop angle closure. However, the accuracy of DRPT might be altered because that after emerging from the dark room, the configuration of the angle is affected by the light of the slit-lamp and the appositionally closed angle reopens. The aim of this study was to examine the pupillary diameter in different light conditions and use it as a parameter to assess the accuracy of dark-room provocative test. METHODS: Patients with suspected primary angle-closure glaucoma undergoing DRPT were recruited. The anterior chamber angle was examined by anterior segment optical coherence tomography under the following conditions: (1) in standard room illumination; (2) after short-term dark-adaptation and (3) after DRPT. Mean values of pupil size and numbers of appositionally closed angle under different conditions were compared. RESULTS: A total of 47 eyes of 47 patients were analyzed. The pupil size after DRPT was smaller than that after short-term dark-adaptation (P < 0.001) and smaller than that in standard room illumination (P = 0.026). The numbers of appositionally closed angles after short-term dark-adaptation were significantly larger than those after DRPT (P < 0.001). There was no significant difference between the numbers of appositionally closed angles in standard room illumination and after DRPT (P = 0.157). CONCLUSIONS: Constriction of pupil size immediately after prolonged dark room provocative test may lead to change in the angle configuration, which may lead to false negative results. We suggest a modified protocol of recording intraocular pressure immediately after DRPT and performing gonioscopy following short-term dark adaptation to improve the accuracy of angle closure assessment.

Validating the accuracy of a model to predict the onset of myopia in children.

PURPOSE: To assess the sensitivity and specificity of models predicting myopia onset among ethnically Chinese children. METHODS: Visual acuity, height, weight, biometry (A-scan, keratometry), and refractive error were assessed at baseline and 3 years later using the same equipment and protocol in primary schools in Xiamen (China) and Singapore. A regression model predicting the onset of myopia < -0.75 diopters (D) after 3 years in either eye among Xiamen children was validated with Singapore data. RESULTS: Baseline data were collected from 236 Xiamen children (mean age, 7.82 ± 0.63 years) and from 1979 predominantly Chinese children in Singapore (7.83 ± 0.84 years). Singapore children were significantly taller and heavier, and had more myopia (31.4% vs. 6.36% < -0.75 D in either eye, P < 0.001) and longer mean axial length. Three-year follow-up was available for 80.0% of Xiamen children and 83.1% in Singapore. For Xiamen, the area under the receiver-operator curve (AUC) in a model including ocular biometry, height, weight, and presenting visual acuity was 0.974 (95% confidence interval [CI], 0.945-0.997). In Singapore, the same model achieved sensitivity, specificity, and positive predictive value of 0.844, 0.650, and 0.669, with an AUC of 0.815 (95% CI, 0.791-0.839). CONCLUSIONS: Accuracy in predicting myopia onset based on simple measurements may be sufficient to make targeted early intervention practical in settings such as Singapore with high myopia prevalence. Models based on cohorts with a greater prevalence of high myopia than that in Xiamen could be used to assess accuracy of models predicting more severe forms of myopia.

People and places: Understanding geographical accuracy in administrative data from the census and healthcare systems

Administrative systems such as health care registration are of increasing importance in providing information for statistical, research, and policy purposes. There is thus a pressing need to understand better the detailed relationship between population characteristics as recorded in such systems and conventional censuses. This paper explores these issues using the unique Northern Ireland Longitudinal Study (NILS). It takes the 2001 Census enumeration as a benchmark and analyses the social, demographic and spatial patterns of mismatch with the health register at individual level. Descriptive comparison is followed by multivariate and multilevel analyses which show that approximately 25% of individuals are reported to be in different addresses and that age, rurality, education, and housing type are all important factors. This level of mismatch appears to be maintained over time, as earlier migrants who update their address details are replaced by others who have not yet done so. In some cases, apparent mismatches seem likely to reflect complex multi-address living arrangements rather than data error.

Functionality, Accuracy, and Feasibility: Talking with Modelers

Models are an important part of many policy development processes, but meeting policy objectives relies on policy analysts engaging effectively with the modeling process and modelers understanding the policy issues. Furthermore, there are many different modeling methods, each with characteristics that potentially make it more or less suitable for analyzing a particular policy issue.
This paper presents a novel framework to assist policy analysts to engage with modelers so as to make the best use of models. The framework has three dimensions: Functionality, Accuracy and Feasibility. Functionality concerns ways in which modeling can be used to support broader policy objectives, such as promoting negotiation or comparing options. Accuracy concerns how to best represent the fundamental features of the system being modeled, and relies on selecting an appropriate technique. Feasibility concerns practical issues such as access to data and modeling skills.

An evaluation of the accuracy of models for the determination of the weighted mean temperature of the atmosphere

The estimates of the zenith wet delay resulting from the analysis of data from space techniques, such as GPS and VLBI, have a strong potential in climate modeling and weather forecast applications. In order to be useful to meteorology, these estimates have to be converted to precipitable water vapor, a process that requires the knowledge of the weighted mean temperature of the atmosphere, which varies both in space and time. In recent years, several models have been proposed to predict this quantity. Using a database of mean temperature values obtained by ray-tracing radiosonde profiles of more than 100 stations covering the globe, and about 2.5 year’s worth of data, we have analyzed several of these models. Based on data from the European region, we have concluded that the models provide identical levels of precision, but different levels of accuracy. Our results indicate that regionally-optimized models do not provide superior performance compared to the global models.

Improvement in the Accuracy of Back Trajectories Using WRF to Identify Pollen Sources in Southern Iberian Peninsula

Airborne pollen transport at micro-, meso-gamma and meso-beta scales must be studied by atmospheric models, having special relevance in complex terrain. In these cases, the accuracy of these models is mainly determined by the spatial resolution of the underlying meteorological dataset. This work examines how meteorological datasets determine the results obtained from atmospheric transport models used to describe pollen transport in the atmosphere. We investigate the effect of the spatial resolution when computing backward trajectories with the HYSPLIT model. We have used meteorological datasets from the WRF model with 27, 9 and 3 km resolutions and from the GDAS files with 1 ° resolution. This work allows characterizing atmospheric transport of Olea pollen in a region with complex flows. The results show that the complex terrain affects the trajectories and this effect varies with the different meteorological datasets. Overall, the change from GDAS to WRF-ARW inputs improves the analyses with the HYSPLIT model, thereby increasing the understanding the pollen episode. The results indicate that a spatial resolution of at least 9 km is needed to simulate atmospheric flows that are considerable affected by the relief of the landscape. The results suggest that the appropriate meteorological files should be considered when atmospheric models are used to characterize the atmospheric transport of pollen on micro-, meso-gamma and meso-beta scales. Furthermore, at these scales, the results are believed to be generally applicable for related areas such as the description of atmospheric transport of radionuclides or in the definition of nuclear-radioactivity emergency preparedness.

Accuracy assessment of depth estimates from variation in pre- and post-processing of multi-beam SONAR surveys in a fjord environment

Senior thesis written for Oceanography 445

Validation of the needle targeting accuracy of a MRI/TRUS- image-guided system for transperineal prostate cancer biopsy

Multi-parametric magnetic resonance imaging (mp-MRI) has become an increasingly important method for detecting and treating prostate cancer. Transrectal ultrasound (TRUS) is the most commonly used method for guiding prostate needle biopsy and remains the gold standard for diagnosis of prostate cancer. MRI-to-TRUS image reg- istration is an important technology for enabling computer-assisted targeting of the majority of prostate lesions that are visible in MRI but not independently distinguishable in TRUS images. The aim of this study was to estimate the needle placement accuracy of an image guidance system (SmartTargetÒ), developed by our research group, using a surgical training phantom.

Post awakening salivary cortisol secretion and trait well-being: The importance of sample timing accuracy

Indices of post awakening cortisol secretion (PACS), include the rise in cortisol(cortisol awakening response: CAR) and overall cortisol concentrations (e.g. area under the curve with reference to ground: AUCg) in the first 30—45 min. Both are commonly investigated in relation to psychosocial variables. Although sampling within the domestic setting is ecologically valid, participant non-adherence to the required timing protocol results in erroneous measurement of PACS and this may explain discrepancies in the literature linking these measures to trait well-being (TWB). We have previously shown that delays of little over 5 min(between awakening and the start of sampling) to result in erroneous CAR estimates. In this study, we report for the first time on the negative impact of sample timing inaccuracy (verified by electronic-monitoring) on the efficacy to detect significant relationships between PACS and TWB when measured in the domestic setting.Healthy females (N = 49, 20.5 ± 2.8 years) selected for differences in TWB collected saliva samples (S1—4) on 4 days at 0, 15, 30, 45 min post awakening, to determine PACS. Adherence to the sampling protocol was objectively monitored using a combination of electronic estimates of awakening (actigraphy) and sampling times (track caps).Relationships between PACS and TWB were found to depend on sample timing accuracy. Lower TWB was associated with higher post awakening cortisol AUCg in proportion to the mean sample timing accuracy (p < .005). There was no association between TWB and the CAR even taking into account sample timing accuracy. These results highlight the importance of careful electronic monitoring of participant adherence for measurement of PACS in the domestic setting. Mean sample timing inaccuracy, mainly associated with delays of >5 min between awakening and collection of sample 1 (median = 8 min delay), negatively impacts on the sensitivity of analysis to detect associations between PACS and TWB.