关键词： 交通工程   现代统计模型   机器学习方法   案例教学法   多元化教学  

摘要： 相较于飞速发展的信息化、大数据浪潮和要求日益提升的交通安全目标,传统的交通统计分析课程内容设置和教学形式都不能满足日益增长的交通安全研究的需求。通过介绍统计分析方法在宏观交通安全管理、道路设施安全影响因素判别和事故多发道路判别与改善等领域的具体应用,说明交通统计分析课程内容在新时代下的要求。在此基础上介绍了此课程在主要教学内容方面的完善,以及案例分析与理论相结合、多元化教学的教学方法方面的改革。

关键词： 区间组合预测   机器学习   损失函数   梯度下降法  

摘要： 随着数据量的不断增多,对于预测精度的要求也越来越高。引入了机器学习中的损失函数和梯度下降算法,与传统的线性规划求解组合系数方法相比,简化了计算步骤,在保证精度的同时也提升了算法对大量数据的处理能力。最后通过实例表明,基于机器学习方法的区间组合预测模型能够有效地提升预测精度。

关键词： Data-driven model   Artificial intelligence   Multi-objective optimization   ROP   Drilling optimization  

摘要： Drilling wells in challenging oil/gas environments implies in large capital expenditure on wellbore's construction. In order to optimize the drilling related operation, real-time decisions making have been put in place, so that prediction of rate of penetration (ROP) with accuracy is essential. Despite many efforts (theoretical and experimental) throughout the years, modeling the ROP as a mathematical function of some key variables is not so trivial, due to the highly non-linearity behavior experienced. Therefore, several researches in the recent years have been proposing to use data-driven models from artificial intelligence field for ROP prediction and optimization. This paper presents an extensive review of the literature on ROP prediction, especially, with machine learning techniques, as well as how these models can be used to optimize the drilling activities. The ROP models are classified as traditional models (based on physics-models), statistical models (e.g. multiple regression), or machine learning methods. This review enables to see that machine learning techniques can potentially outperform in terms of ROP-prediction accuracy on top of traditional or statistical models. Throughout this work, an extensive analysis of different ways of obtaining ROP models is carried out, concluding with different strategies adopted in literature to perform data-driven model optimization. Despite the saving potential which can be achieved with real-time optimization based on data-driven ROP models, it is noticeable that there is a lack of implementation of those techniques in the industry, as per literature review. To take a step forward in real implementations, the petroleum industry must be aware that yet no rule of thumb already exists on this specific area, but still, good and very reasonable results can be achieved by following the best practices identified in this review. In addition, the modern practices of machine learning provide promising guidelines for implementi

关键词： Image matching   Machine learning   Nearest neighbor search   Light field imaging  

摘要： At present, there are many methods to realize the matching of specified images with features, and the basic components include image feature point detection, feature description, and image matching. Based on this background, this article has done different research and exploration around these three aspects. The image feature point detection method is firstly studied, which commonly include image edge information-based feature detection method, corner information-based detection method, and various interest operators. However, all of the traditional detection methods are involved in problems of large computation burden and time consumption. In order to solve this problem, a feature detection method based on image grayscale information-FAST operator is used in this paper, which is combined with decision tree theory to effectively improve the speed of extracting image feature points. Then, the feature point description method BRIEF operator is studied, which is a local expression of detected image feature points based on descriptors. Since the descriptor does not have rotation invariance, the detection operator is endowed by a direction that is proposed in this paper, and then the local feature description is conducted on the feature descriptor to generate a binary string array containing direction information. Finally, the feature matching machine learning method is analyzed, and the nearest search method is used to find the nearest feature point pair in Euclidean distance, of which the calculation burden is small. The simulation results show that the proposed nearest neighbor search and matching machine learning algorithm has higher matching accuracy and faster calculation speed compared with the classical feature matching algorithm, which has great advantages in processing a large number of array images captured by the light field camera.

关键词： Machine learning   Energy consumption   Crosstalk   Extreme learning machine   Elastic optical networks   Space division multiplexing  

摘要： Elastic optical networks (EONs) and space division multiplexing (SDM) are promising technologies for future core optical networks with high transmission capacity. However, energy consumption and crosstalk are two inevitable problems of practical application of space division multiplexing elastic optical networks (SDM-EONs). In this paper, we leverage machine learning techniques to decrease the energy consumption and crosstalk of SDM-EONs. Specifically, we use an online sequence extreme learning machine to forecast traffic load on the lightpath, and reconstruct an energy model containing energy-consumption components in the whole network so as to reduce energy consumption. We also present a suitable scheme to decrease crosstalk. Numerical results show that the proposed algorithm is effective in achieving substantial savings in energy consumption while maintaining blocking probability and crosstalk at levels comparable to those of earlier algorithms.

关键词： Fungicide   Reproduction   Spermatozoa   Rat   Raman spectroscopy   Artificial intelligence  

摘要： Difenoconazole is a fungicide extensively used in agriculture. The aim of this study was to evaluate the effects of difenoconazole fungicide on the sperm quality of rats. Wistar rats were divided into four groups: control and exposed to 5 (D5), 10 (D10), or 50 mg(-1) kg bw(-1) day (D50) of difenoconazole for 30 days, by gavage. Classical sperm parameters and surface-enhanced Raman scattering (SERS) were performed. Progressive motility, acrosomal integrity, and percentage of morphologically normal spermatozoa were reduced in the D10 and D50 groups in comparison with the control group. Sperm viability was reduced only in the D50 group. Sperm number in the testis and caput/corpus epididymis and daily sperm production were reduced in the three exposed groups. SERS measurements showed changes in the spectra of spermatozoa from D50 group, suggesting DNA damage. In addition, machine learning (ML) methods were used to evaluate the performance of three classification algorithms (artificial neural network-ANN, K-nearest neighbors-K-NN, and support vector machine-SVM) in the identification task of the groups exposed to difenoconazole. The results obtained by ML algorithms were very promising with accuracy >= 90% and validated the hypothesis of the exposure to difenoconazole reduces sperm quality. In conclusion, exposure of rats to different doses of the fungicide difenoconazole may impair sperm quality, with a recognizable classification pattern of exposure groups.

关键词： CT lung cancer screening   CT dose   Machine learning   Convolutional neural network   Artificial neural network  

摘要： Low-dose computed tomography (CT) lung cancer screening is recommended by the US Preventive Services Task Force for high lung cancer-risk populations. In this study, we investigated an important factor affecting the CT dose-the scan length, for this CT exam. A neural network model based on the "UNET" framework was established to segment the lung region in the CT scout images. It was trained initially with 247 chest X-ray images and then with 40 CT scout images. The mean Intersection over Union (IOU) and Dice coefficient were reported to be 0.954 and 0.976, respectively. Lung scan boundaries were determined from this segmentation and compared with the boundaries marked by an expert for 150 validation images, resulting an average 4.7% difference. Seven hundred seventy CT low-dose lung screening exams were retrospectively analyzed with the validated model. The average "desired" scan length was 252 mm with a standard deviation of 28 mm. The average "over-range" was 58.5 mm or 24%. The upper boundary (superior) on average had an "over-range" of 17 mm, and the lower boundary (inferior) on average had an "over-range" of 41 mm. Further analysis of this data showed that the extent of "over-range" was independent of acquisition date, acquisition time, acquisition station, and patient age, but dependent on technologist and patient weight. We concluded that this machine learning method could effectively support quality control on the scan length for CT low-dose screening scans, enabling the eliminations of unnecessary patient dose.

关键词： Machine learning   Hybrid organic-inorganic perovskites   First principle calculations   Photovoltaics  

摘要： Traditional trial-and-error methods seriously restrict and hinder the searching of high-performance functional materials, especially when the search space is large. Rapid searching for advanced functional materials has always been a hot research topic, and attracted a lot of experimental and theoretical research attention. Here, by combining machine learning method with density functional theory (DFT) calculations, a target-driven method is proposed here to speed up the discovery of hidden hybrid organic-inorganic perovskites (HOIPs) for photovoltaic applications from 230808 HOIPs candidates which is almost two orders larger than previous studied. After imposing two criterions, i.e., charge neutrality condition and stability condition, on potential HOIPs candidates, followed by a machine learning (ML) screening, 686 orthorhombic-like HOIPs with proper bandgap are selected. In machine learning screening, ensemble learning using three ML models, including gradient boosting regression (GBR), supporting vector regression (SVR) and kernel ridge regression (KRR), are applied to predict the bandgap of 38086 HOIPs candidates. 132 stable and non-toxic (Cd-, Pb- and Hg-free) orthorhombiclike HOIPs are finally verified by DFT calculations with appropriate band gap for solar cells. In the present study, not only a series of unexplored stable and non-toxic HOIPs are discovered for further experimental synthesis, a new HOIPs database is constructed as well, thus beneficial to future functional material design.

关键词： Deformation  

摘要： Twinning is an important deformation mechanism for Mg and other hexagonal metals. While twin nucleation is known to depend on the size and crystal orientation of the parent grain, so far there is no satisfactory criterion to predict twin nucleation in a polycrystalline Mg alloy prior to its deformation. In this work, machine learning is employed to tackle this problem. From a single-phase, polycrystalline Mg-0.47 wt pct Ca extruded alloy, three micro-tensile specimens, E-0, E-45, and E-90, were fabricated with their tensile axis being 0, 45, and 90 deg from the extrusion direction. Each specimen was deformed by 4 pct tensile strain in a scanning electron microscope. Six hundred thirty-six grains from E-0, 572 grains from E-45, and 840 grains from E-90 were characterized by electron backscattered diffraction (EBSD) before and after deformation. Twin nucleation was identified in 27, 150, and 220 grains in E-0, E-45, and E-90, respectively. Eighteen attributes that can influence twin nucleation, such as grain diameter, c-axis direction, and Schmid factors, were computed for each grain. Five machine learning algorithms, including decision tree, tree ensemble (XGBoost), artificial neural network (ANN), support vector machine (SVM), and naive Bayes, were used to build models to predict twin nucleation according to a grain's 18 attribute values, with E-45 as the training set and the other two specimens as the test sets. The ANN and SVM models show the best performance, both achieving similar to 87 pct prediction accuracy for specimens E-45 and E-90. None of the models perform well for E-0 because of the imbalanced class distribution in this specimen. The unpredicted twin nucleation events in E-90 mostly originate from triple junctions or twin-twin transmission at grain boundaries.

关键词： EEG   BCI   Mental state detection   Drowsiness detection   Support vector machine   Passive control task  

摘要： Recent advances in technology bring about novel operating environments where the role of human participants is reduced to passive observation. While opening new frontiers in productivity and lifestyle, such environments also create hazards related to the inability of human individuals to maintain focus and concentration during passive control tasks. A passive brain-computer interface for monitoring mental attention states of human individuals (focused, unfocused, and drowsy) by using electroencephalographic (EEG) brain activity imaging and machine learning data analysis methods is developed in this work. An EEG data processing pipeline and a machine learning mental state detection algorithm using the Support Vector Machine (SVM) method were designed and compared with k-Nearest Neighbor and Adaptive Neuro-Fuzzy System methods. To collect 25 h of EEG data from 5 participants, a classic EEG headset was modified. We found that the changes in EEG activity in frontal and parietal lobes occurring at 1-5 Hz and 10-15 Hz frequency bands were associated with the changes in individuals' attention state. We demonstrated the ability to use such changes to identify individuals' attention state with 96.70% (best) and 91.72% (avg.) accuracy in experimental settings using a version of continuous performance task with SVM-based mental state detector. The findings help guide the design of future systems for monitoring the state of human individuals by means of EEG brain activity data. (C) 2019 Elsevier Ltd. All rights reserved.