关键词： Machine learning   Time series prediction   Predictive maintenance   Catalyst degradation  

摘要： Accurately predicting industrial aging processes makes it possible to schedule maintenance events further in advance, ensuring a cost-efficient and reliable operation of the plant. So far, these degradation processes were usually described by mechanistic or simple empirical prediction models. In this paper, we evaluate a wider range of data-driven models, comparing some traditional stateless models (linear and kernel ridge regression, feed-forward neural networks) to more complex recurrent neural networks (echo state networks and LSTMs). We first examine how much historical data is needed to train each of the models on a synthetic dataset with known dynamics. Next, the models are tested on real-world data from a large scale chemical plant. Our results show that recurrent models produce near perfect predictions when trained on larger datasets, and maintain a good performance even when trained on smaller datasets with domain shifts, while the simpler models only performed comparably on the smaller datasets. (C) 2020 Elsevier Ltd. All rights reserved.

关键词： 共享单车   数据分析   极端随机树   需求量预测   机器学习  

摘要： 由于共享单车的流动性强,随机性很高,因此快速精确地预测出城市共享单车的短时需求量具有十分重要的意义。采用随机森林、极端随机树、支持向量机、人工神经网络、XGBoost这5种机器学习方法,基于美国华盛顿共享单车项目数据,分析时间因子、气象因子等对单车需求量的影响,实现对共享单车短时需求量的预测。仿真结果表明,影响单车需求量的主要因素包括温度、节假日、季节以及早晚高峰时间段等因素;极端随机树的预测效果最优,MAE和RMSE最小,为22.93和36.84,训练集得分和验证集得分最高,为1.0和0.941,与随机森林和其它算法相比,鲁棒性高,泛化能力强,且预测结果曲线与真实结果曲线相吻合,预测精度高,可为实际的车辆预测和调度提供参考依据。

关键词： 随机临床试验   肉毒杆菌   人工智能   偏头痛   机器学习方法   特定疾病   科研工作   偶然性  

摘要： 科学家们开发了一种可以处理大量数据以帮助现有药物找到其新用途的机器学习方法。这项科研工作的目的是为了加快药物再利用的步伐,药物再利用并不稀奇,例如最初被批准用于治疗眼疾的肉毒杆菌注射液,现在被用于治疗偏头痛和美容减皱。但是,要获得这些新用途,通常需要进行偶然性、耗时且昂贵的随机临床试验,以确保被认为对特定疾病有效的药物可用于治疗其他疾病。

关键词： Phasor measurement units  

摘要： As Phasor Measurement Units (PMUs) become widely deployed, power systems can take advantage of the large amount of data provided by PMUs and leverage the advances in big data analytics to improve real-time monitoring and diagnosis. In this paper, we develop practical analytics that are not tightly coupled with the power flow analysis and state estimation, as these tasks require detailed and accurate information about the power system. We focus on power line outage identification, and use a machine learning framework to locate line outages. The same framework is used for the prediction of both single line and multiple line outages. We investigate a range of machine learning algorithms and feature extraction methods. The algorithms are designed to capture the essential dynamic characteristics of the power system when the topology change occurs abruptly. The proposed methods use only voltage phasor angles obtained by continuously monitoring the buses. We tested the proposed methods on their prediction performance under different levels of noise and missingness. It is shown that the proposed methods have better tolerance for noisy data and incomplete data when compared to the previous work that involves solving power flow equations or state estimation equations. © 2020 Elsevier Inc.

关键词： suspicious transactions   non-typical transactions   anti-fraud system   machine learning   classification methods.  

摘要： The growing popularity of payment cards has led to the emergence of new types of illegal transactions with money. In particular, the widespread use of non-cash payments has allowed fraud to reach the international level. Therefore, financial institutions are interested in the development and implementation of new effective fraud monitoring systems that will minimize the risk of approving illegal transactions. The article presents the results of applying machine learning methods to detect fraudulent transactions with bank cards. The use of various classification methods in modeling the specified problem is investigated. Generalized algorithm for detecting fraudulent transactions has been developed, which makes it possible to detect atypical international money transfers in real time. Generalized algorithm for detecting atypical international transfers will allow timely detection of potential fraud cases, thereby reducing the total volume of losses from illegal transactions and minimizing the reputation damage caused to the organization.

关键词： public procurement   government procurement   associative rules   cluster analysis methods   machine learning.  

摘要： The article provides the results of applying machine learning methods for prediction of unfavorable outcome of the public procurement procedure in the building and grounds maintenance domain. Based on a comprehensive analysis of the domain it was decided to investigate the following risks: the risk of collusion among suppliers; the risk of conspiracy between customers and suppliers; the risk associated with inaccurate data in the Unified Information System. Usage of various classification techniques has been researched while modeling the problem in the domain. In order to form sustainable groups of suppliers, the association rule mining was done using the "Apriori" algorithm. While searching for representative characteristics of the groups of similar objects, the solution to the clustering problem was found using the Ward and K-means++ methods. The Cluster models, which were defined to analyze each of the collusion risks, were built on the feature space. The models make it possible to identify the most typical behavioral patterns of two suppliers with each other as well as the customer with the supplier.

关键词： COVID-19   Masked face   Deep transfer learning   Classical machine learning  

摘要： The coronavirus COVID-19 pandemic is causing a global health crisis. One of the effective protection methods is wearing a face mask in public areas according to the World Health Organization (WHO). In this paper, a hybrid model using deep and classical machine learning for face mask detection will be presented. The proposed model consists of two components. The first component is designed for feature extraction using Resnet50. While the second component is designed for the classification process of face masks using decision trees, Support Vector Machine (SVM), and ensemble algorithm. Three face masked datasets have been selected for investigation. The Three datasets are the Real-World Masked Face Dataset (RMFD), the Simulated Masked Face Dataset (SMFD), and the Labeled Faces in the Wild (LFW). The SVM classifier achieved 99.64% testing accuracy in RMFD. In SMFD, it achieved 99.49%, while in LFW, it achieved 100% testing accuracy.

关键词： Copy number alteration   gene expression   cancer   machine learning   bioinformatics  

摘要： Detection of DNA copy number alteration in cancer cells is critical to understanding cancer initiation and progression. Widely used methods, such as DNA arrays and genomic DNA sequencing, are relatively expensive and require DNA samples at a microgram level, which are not avaiblable in certain situations like clinical biopsies or single-cell genomes. Here, we developed an alternative method-CNAPE to computationally infer copy number alterations from gene expression data. A prior knowledge-aided machine learning model was proposed, trained and tested on 9,740 cancer samples from The Cancer Genome Atlas. We then applied CNAPE to study gliomas, the most common and aggressive brain cancer in adult. Particularly, using RNA sequencing data, CNAPE respectively predicted DNA copy number of chromosomes, chromosomal arms, and 12 commonly altered genes, and achieved over 80 percent accuracy in almost all broad regions and some focal regions. CNAPE was developed as an easy-to-use tool at https://***/WangLabHKUST/CNAPE.

关键词： 机器学习   流感预测   公共卫生风险   风险预测  

摘要： 【目的】探索应用机器学习预测流感这类公共卫生风险的可行性和有效性。【方法】首先,收集2009-2016年兰州市的流感和气象数据,拆分成2009-2015年和2016年两组,分别作为训练和验证数据;然后,分别基于SARIMA、Kalman Filter和VAR建立三种机器学习预测方法,并设计两种多方法联合预测策略;最后,评估、比较上述方法(策略)的预测性能。【结果】在设定的全期、爆发期和稳定期三种场景下,SARIMA、VAR和Kalman Filter方法的预测效果分别为最佳(RMSE分别为11.68、19.23和1.60;R^(2)分别为0.932、0.923和0.956);多方法联合策略可进一步提升三种场景下的预测效果,其中联合策略Comb_2的表现更好(RMSE分别为10.82、14.68和1.38;R^(2)分别为0.942、0.934和0.963)。【局限】相关数据限制,主要考虑了气象一类外部相关因素。【结论】应用机器学习预测流感等公共卫生风险具有可行性和有效性,且潜力巨大。但目前面临的主要困境是多源数据缺乏,需要从技术、组织和制度层面打破数据壁垒,推动数据共享与开放。

关键词： Spam   Spam Filtering Method   Naive Bayes   K-Nearest Neighbors   SVM   Logistic regression   Decision tree   Random forest  

摘要： Technological advances are accelerating the dissemination of information. Today, millions of devices and their users are connected to the Internet, allowing businesses to interact with consumers regardless of geography. People all over the world send and receive emails every day. Email is an effective, simple, fast, and cheap way to communicate. It can be divided into two types of emails: spam and ham. More than half of the letters received by the user – spam. To use Email efficiently without the threat of losing personal information, you should develop a spam filtering system. The aim of this work is to reduce the amount of spam using a classifier to detect it. The most accurate spam classification can be achieved using machine learning methods. A natural language processing approach was chosen to analyze the text of an email in order to detect spam. For comparison, the following machine learning algorithms were selected: Naive Bayes, K-Nearest Neighbors, SVM, Logistic regression, Decision tree, Random forest. Training took place on a ready-made dataset. Logistic regression and NB give the highest level of accuracy – up to 99%. The results can be used to create a more intelligent spam detection classifier by combining algorithms or filtering methods.