关键词： Retina   Blood vessel   Image processing   Vessel extraction   Classification   Clustering  

摘要： In medicine, retinal vessel analysis of fundus images is a prominent task for the screening and diagnosis of various ophthalmological and cardiovascular diseases. In this research, a method is proposed for extracting the retinal blood vessels employing a set of effective image features and combination of supervised and unsupervised machine learning techniques. Further to the common features used in extracting blood vessels, three strong features having a significant influence on the accuracy of the vessel extraction are utilized. The selected combination of the different types of individually efficient features results in a rich local information with better discrimination for vessel and non-vessel pixels. The proposed method first extracts the thick and clear vessels in an unsupervised manner, and then, it extracts the thin vessels in a supervised way. The goal of the combination of the supervised and unsupervised methods is to deal with the problem of intra-class high variance of image features calculated from various vessel pixels. The proposed method is evaluated on three publicly available databases DRIVE, STARE and CHASE_DB1. The obtained results (DRIVE: Acc = 0.9531, AUC = 0.9752;STARE: Acc = 0.9691, AUC = 0.9853;CHASE_DB1: Acc = 0.9623, AUC = 0.9789) demonstrate the better performance of the proposed method compared to the state-of-the-art methods.

关键词： Statistics  

摘要： The aim of this work is to investigate the possibility of using machine learning (ML) methods in order to generate novel, safety-relevant knowledge from existing flight data. Airlines routinely generate vast amounts of flight data from routine monitoring, but the concept of extracting safety knowledge from this data is still based on detecting exceedances of expert-defined thresholds. This system is conceptually unable to detect novel occurrences for which no such filters exist. ML techniques are able to close this gap. This paper first reviews the literature to select an appropriate ML method. A form of unsupervised learning called "Local Outlier Probability" is selected. Next, an appropriate feature space is developed and implemented in the flight data monitoring system of a supporting airline to generate the dataset. This dataset is cleaned and the outlier calculation performed. The results are statistically analysed. Furthermore, the top outliers are reviewed by the airline's review pilots in the same way as the traditional exceedance events. Last, the severities and safety relevance of both types of events are compared. This work successfully shows that the chosen approach is able to reduce the number of undetected safety-relevant occurrences by finding novel occurrence types which were undetected by a contemporary and mature flight data monitoring system. This research builds on recent literature by developing a novel method which can be scaled to work in an airline production environment with large datasets, as demonstrated by the efficient analysis of 1.2 million flights.

关键词： energy efficiency   GPGPU   machine learning   mixed precision   MPDATA   random forest  

摘要： In this work, we propose a method that allows us to reduce energy consumption of an application executed on supercomputing centers. The proposed method is based on a mixed precision arithmetic where the precision of data is calibrated at runtime. For this reason, we develop a modified version of the random forest algorithm. The effectiveness of the proposed approach is validated with a real-life scientific application called MPDATA, which is part of the numerical model used in weather forecasting. The energy efficiency of the proposed method is examined using two GPU-based clusters. The first of them is the Piz Daint supercomputer, currently ranked 3rd at the TOP500 list (November 2017). It is equipped with NVIDIA Tesla P100 GPU accelerators based on the Pascal architecture. The second is the MICLAB cluster containing NVIDIA Tesla K80 based on the Kepler architecture. The achieved results show that the proposed machine learning method allows us to provide the accuracy of computation comparable with that achieved double precision and reduce the energy consumption up to 36% compared to the double precision version of MPDATA.

关键词： Machine learning   anomaly detection   deep learning   extreme learning machine   smart water grids   water quality  

摘要： Traditional machine learning (ML) techniques such as support vector machine, logistic regression, and artificial neural network have been applied most frequently in water quality anomaly detection tasks. This paper presents a review of progress and advances made in detecting anomalies in water quality data using ML techniques. The review encompasses both traditional ML and deep learning (DL) approaches. Our findings indicate that: 1) Generally, DL approaches outperform traditional ML techniques in terms of feature learning accuracy and fewer false positive rates. However, it is difficult to make a fair comparison between studies because of different datasets, models and parameters employed. 2) We notice that despite advances made and the advantages of the extreme learning machine (ELM), its application is sparsely exploited in this domain. This study also proposes a hybrid DL-ELM framework as a possible solution that could be investigated further and used to detect anomalies in water quality data.

关键词： Slow-wave sleep   Slow waves   Automatic methods   Artificial neural network  

摘要： Background: During slow-wave sleep the electroencephalographic (EEG) and local field potential (LFP) recordings reveal the presence of large amplitude slow waves. Systematic extraction of individual slow waves is not trivial. New method: In this study, we used the neural network pattern recognition to detect individual slow waves in LFP recorded from mice as well as other commonly used methods that are based on fast frequencies modulation, amplitude, or duration. Results: The number and quality of events detected as slow waves depended on the chosen method of detection, level of thresholds, or on combination of methods. Each individual method yields some false-positive and false-negative detections. Typically, the fast frequency-method has a higher false discovery rate, but almost no missing waves;amplitude-based method has relatively high false-positive and false-negative rates;duration-based method has low false-negative rates;neural network pattern recognition approach has the lowest false-positive rate among individual methods, often rejecting waves that were falsely detected by other approaches. Combining all 4 detection methods practically eliminated false-positive errors, but a large number of slow waves remained undetected. Conclusions: The use of a particular method of slow wave detection needs to be adjusted to the objectives of a given study: to detect all slow waves, but also numerous false positives can be achieved using the fast frequency approach. Neural network pattern recognition method alone can detect slow waves with the lowest false-positive rate, that can be further minimized with the use of combination of other methods.

关键词： 设计模式   设计模式挖掘   子图同构   机器学习  

摘要： 从源码中挖掘设计模式实例具有重要意义,它能够提升软件系统的可理解性和可维护性。基于机器学习方法,提出了一个设计模式挖掘方法。其中,使用类关系图表示软件系统和设计模式,采用子图匹配的方法从软件系统中提取出候选设计模式实例,过滤了大量非设计模式实例组合,提高了设计模式挖掘效率;提出了度量指标,将软件系统和设计模式向量化,采用机器学习的方法对候选设计模式实例进行分类,提高了设计模式挖掘的准确率。实验结果证明此方法效率和有效性上得到了一定的提升。

关键词： 机器学习   原发性肝癌   logistic回归   支持向量机   预后预测  

摘要： 目的:使用机器学习方法对原发性肝癌患者生存和复发时间进行预测,并比较不同机器学习方法的预测效果。方法:使用Logistic回归、随机森林、SVM、C5.0决策树、神经网络、Bagging算法和Adaboost算法构建患者肝切除手术后3年无瘤生存时间和3年总生存时间的预后预测模型。结果:通过Logistic回归模型找到了影响患者生存和复发的影响因素,并且支持向量机和随机森林有较好的效果,测试集上的AUC为0.7122和0.7103。结论:机器学习方法相较于Logistic回归模型有较好的预测效果,但是Logistic回归模型有较好的可解释性。在分析实际问题时,可以采用二者结合的方法。

关键词： 中医治未病   贝叶斯算法   病例大数据   机器学习  

摘要： 本文采用贝叶斯网络对"未病人群"健康检查数据进行及时准确的分析,依据病例大数据,构建机器学习模型,推进健康医疗大数据的应用,从而科学的评估自身健康状况及存在风险,合理的管理健康指标,并及时进行健康干预。本研究提出的方法在一定程度上提高了量化评估结果,缩短建模时间,使临床医生针对特定人群制定相应策略;提高医生工作效率,使新型中医诊疗方法不断继承、创新、发展。

关键词： 慢性阻塞性肺疾病   k-最近相邻法   SVM   准确度  

摘要： 目的:采用基于机器学习的分类判断算法,建立慢性阻塞性肺疾病(以下简称"慢阻肺")分期模型,提高慢阻肺诊断和分期的准确度。方法:选择确诊慢阻肺住院患者2 504例,以国际GOLD分期为依据,收集与慢阻肺分期密切相关的临床特征参数指标,对参数进行筛选,参照医院的临床确诊结果,采用机器学习方法(k-最近相邻法、SVM)训练并测试慢阻肺的分期模型。结果与结论:数据为不平衡数据,虽采用分层比例抽样,但针对此类数据SVM的准确率更高为85.26%。说明机器学习提供的模型能为慢阻肺分期提供较准确的分类依据。

关键词： 机器学习   脑电图   癫痫分类预测   特征提取  

摘要： 目前,癫痫作为世界上最为常见的神经系统疾病之一,迫切需要有效的方法来进行诊治。脑电图检查对于诊断癫痫病有着决定性的作用,最近几十年来,一些新模型新算法的引入给癫痫分类带来了较高的准确性,尤其是深度学习神经网络算法,如CNN、LSTM、RNN等。总体来说算法大致分两大类,一类是手动提取特征并用分类器进行分类,一类是让机器自行学习特征并分类,也就是神经网络算法。介绍了一些机器学习方法在癫痫分类预测上的实现和应用,以及针对脑电波的一些预处理和特征提取方法,如脑电波的特征提取和选择,最后对面临的问题和发展状况进行总结和展望。