Monash University

About: Monash University is a education organization based out in Melbourne, Victoria, Australia. It is known for research contribution in the topics: Population & Poison control. The organization has 35920 authors who have published 100681 publications receiving 3027002 citations.

DeepGauge: multi-granularity testing criteria for deep learning systems

Abstract: Deep learning (DL) defines a new data-driven programming paradigm that constructs the internal system logic of a crafted neuron network through a set of training data. We have seen wide adoption of DL in many safety-critical scenarios. However, a plethora of studies have shown that the state-of-the-art DL systems suffer from various vulnerabilities which can lead to severe consequences when applied to real-world applications. Currently, the testing adequacy of a DL system is usually measured by the accuracy of test data. Considering the limitation of accessible high quality test data, good accuracy performance on test data can hardly provide confidence to the testing adequacy and generality of DL systems. Unlike traditional software systems that have clear and controllable logic and functionality, the lack of interpretability in a DL system makes system analysis and defect detection difficult, which could potentially hinder its real-world deployment. In this paper, we propose DeepGauge, a set of multi-granularity testing criteria for DL systems, which aims at rendering a multi-faceted portrayal of the testbed. The in-depth evaluation of our proposed testing criteria is demonstrated on two well-known datasets, five DL systems, and with four state-of-the-art adversarial attack techniques against DL. The potential usefulness of DeepGauge sheds light on the construction of more generic and robust DL systems.

Observation of Attractive and Repulsive Polarons in a Bose-Einstein Condensate

Abstract: The problem of an impurity particle moving through a bosonic medium plays a fundamental role in physics. However, the canonical scenario of a mobile impurity immersed in a Bose-Einstein condensate (BEC) has not yet been realized. Here, we use radio frequency spectroscopy of ultracold bosonic ^{39}K atoms to experimentally demonstrate the existence of a well-defined quasiparticle state of an impurity interacting with a BEC. We measure the energy of the impurity both for attractive and repulsive interactions, and find excellent agreement with theories that incorporate three-body correlations, both in the weak-coupling limits and across unitarity. The spectral response consists of a well-defined quasiparticle peak at weak coupling, while for increasing interaction strength, the spectrum is strongly broadened and becomes dominated by the many-body continuum of excited states. Crucially, no significant effects of three-body decay are observed. Our results open up exciting prospects for studying mobile impurities in a bosonic environment and strongly interacting Bose systems in general.

Heart Failure Association of the European Society of Cardiology practical guidance on the use of natriuretic peptide concentrations

Abstract: Natriuretic peptide [NP; B-type NP (BNP), N-terminal proBNP (NT-proBNP), and midregional proANP (MR-proANP)] concentrations are quantitative plasma biomarkers for the presence and severity of haemodynamic cardiac stress and heart failure (HF). End-diastolic wall stress, intracardiac filling pressures, and intracardiac volumes seem to be the dominant triggers. This paper details the most important indications for NPs and highlights 11 key principles underlying their clinical use shown below. NPs should always be used in conjunction with all other clinical information. NPs are reasonable surrogates for intracardiac volumes and filling pressures. NPs should be measured in all patients presenting with symptoms suggestive of HF such as dyspnoea and/or fatigue, as their use facilitates the early diagnosis and risk stratification of HF. NPs have very high diagnostic accuracy in discriminating HF from other causes of dyspnoea: the higher the NP, the higher the likelihood that dyspnoea is caused by HF. Optimal NP cut-off concentrations for the diagnosis of acute HF (very high filling pressures) in patients presenting to the emergency department with acute dyspnoea are higher compared with those used in the diagnosis of chronic HF in patients with dyspnoea on exertion (mild increase in filling pressures at rest). Obese patients have lower NP concentrations, mandating the use of lower cut-off concentrations (about 50% lower). In stable HF patients, but also in patients with other cardiac disorders such as myocardial infarction, valvular heart disease, atrial fibrillation or pulmonary embolism, NP concentrations have high prognostic accuracy for death and HF hospitalization. Screening with NPs for the early detection of relevant cardiac disease including left ventricular systolic dysfunction in patients with cardiovascular risk factors may help to identify patients at increased risk, therefore allowing targeted preventive measures to prevent HF. BNP, NT-proBNP and MR-proANP have comparable diagnostic and prognostic accuracy. In patients with shock, NPs cannot be used to identify cause (e.g. cardiogenic vs. septic shock), but remain prognostic. NPs cannot identify the underlying cause of HF and, therefore, if elevated, must always be used in conjunction with cardiac imaging.

Meta-analysis: trophic level, habitat, and productivity shape the food web effects of resource subsidies

Abstract: Studies of the effects of cross-habitat resource subsidies have been a feature of food web ecology over the past decade. To date, most studies have focused on demonstrating the magnitude of a subsidy or documenting its effect in the recipient habitat. Ecologists have yet to develop a satisfactory framework for predicting the magnitude of these effects. We used 115 data sets from 32 studies to compare consumer responses to resource subsidies across recipient habitat type, trophic level, and functional group. Changes in consumer density or biomass in response to subsidies were inconsistent across habitats, trophic, and functional groups. Responses in stream cobble bar and coastline habitats were larger than in other habitats. Contrary to expectation, the magnitude of consumer response was not affected by recipient habitat productivity or the ratio of productivity between donor and recipient habitats. However, consumer response was significantly related to the ratio of subsidy resources to equivalent resources in the recipient habitat. Broad contrasts in productivity are modified by subsidy type, vector, and the physical and biotic characteristics of both donor and recipient habitats. For this reason, the ratio of subsidy to equivalent resources is a more useful tool for predicting the possible effect of a subsidy than coarser contrasts of in situ productivity. The commonness of subsidy effects suggests that many ecosystems need to be studied as open systems.

The link between total quality management practice and organisational performance

Abstract: Tests the strength of the relationship between TQM practice and organisational performance with and without the covariates, company size, industry type, and ISO 9000 certification status. Analysis is primarily of a large random sample of manufacturing companies in Australia and New Zealand. The central finding of the study is that TQM tends to have mixed results when covaried for company size and industry type. The relationship between TQM and organisational performance is not affected when covaried for ISO 9000 certification status. Concludes that there are significant differences in the relationship between TQM and organisational performance across industry sectors and different size companies, particularly on the effect of defect rates, warranty costs and innovation of new products. The main implication of the findings for managers is that a typical manufacturing organisation is more likely to achieve better performance in employee relations, customer satisfaction, operational performance and business performance, with TQM than without TQM.