关键词： 医学图像   大数据   分割诊断   机器学习   深度学习  

摘要： 随着人工智能技术的发展与应用和医学图像数据爆炸式增长,传统依靠医生对医学图像进行人工分割诊断,不仅工作效率低下、工作量大,还容易产生误诊、漏诊。机器学习,尤其是深度学习在医学图像非结构化大数据领域发挥着越来越重要的作用。为了进一步了解机器学习在医学图像自动分割和识别诊断中的研究,本文对机器学习及其在医学图像分析领域的研究进展进行综述,为机器学习方法解决医学图像非结构化大数据提供方法学参考。

关键词： 环形RNA   编码蛋白   机器学习  

摘要： 环形RNA是一类广泛存在于真核细胞中的内源性RNA分子,没有5’末端帽子和3’末端polyA尾巴,以共价键连接形成封闭环状结构.一直以来,环形RNA被认为是一类不能翻译的非编码RNA.然而近年来有研究报道,环形RNA能够编码蛋白质从而调控重要的生命活动,引起了研究者们的注意.运用机器学习的方法,基于环形RNA的序列与结构特征,使用XGBoost、随机森林和支持向量机组合而成的综合分类模型,预测环形RNA编码蛋白的潜能,平均预测准确率达到86.66%,为实验研究人员提供可靠的参考,有助于发现更多可编码蛋白的环形RNA.

关键词： 设备   时间序列   预测   数据闭环   AIaaS  

摘要： 主要介绍了基于机器学习方法的设备监控指标的模型设计与实现。基于时间序列算法,从生产实际出发,在不同应用场景下对不同厂商,甚至是同一厂商内不同设备,不同版本,设计和构建了“冷启动”的指标预测模型。使得指标的预测“千人千面”,并同时达成数据闭环和模型更新的效果。在应用上主要用于监控不同维度的网络业务恶化的问题。

关键词： inter-well connectivity   bottom-hole pressure   machine learning   artificial neural network   time lags   oil   gas reservoirs   petroleum engineering   petroleum wells-drilling   production   construction   wells-injection   oil   gas  

摘要： In the production and development of oil fields, production wells generally produce at a constant rate since the fixed production is easier to control than the fixed pressure. Thus, it is more feasible to use bottom-hole pressure data for connectivity analysis than historical injection and production data when producers are set in fixed rates. In this work, a practical procedure is proposed to infer inter-well connectivity based on the bottom-hole pressure data of injectors and producers. The procedure first preprocesses the bottom-hole pressure based on nonlinear diffusion filters to constitute the dataset for machine learning. An artificial neural network (ANN) is then generated and trained to simulate the connection relationship between the producer and its adjacent injectors. The genetic algorithm (GA) is also introduced to avoid the tedious process of determining time lags and other hyper-parameters of ANN. In particular, the time lag is normally determined by subjective judgment, which is optimized by GA for the first time. After optimizing the parameters, the sensitivity analysis is performed on the well-trained ANN to quantify inter-well connectivity. For the evaluation and verification purposes, the proposed GA and sensitivity analysis based ANN were applied to two synthetic reservoirs and one actual case from JD oilfields, China. The results show that the calculated connectivity conforms to known geological characteristics and tracer test results. And it demonstrates that the presented approach is an effective alternative way to characterize the reservoir connectivity and determine the flow direction of injected water.

关键词： 信用风险评估   xgboost-LR算法   随机森林   支持向量机  

摘要： 互联网金融在国内的兴起,使得个人信贷风险成为许多企业关注的热点。本文通过对LendingClub网站中个人信用贷款数据集的探索,构建基于xgboost和logistic regression组合算法xgboost-LR模型、随机森林和支持向量机算法三种机器学习方法,对个人信用风险进行全面的评估。通过实证数据分析,其中新构建的xgboost-LR算法评价效果最好,能够更加准确地预测个人信用风险。

关键词： RNA structure prediction   RNA scoring function   knowledge-based potentials   machine learning   convolutional neural networks  

摘要： RNAs play crucial and versatile roles in biological processes. Computational prediction approaches can help to understand RNA structures and their stabilizing factors, thus providing information on their functions, and facilitating the design of new RNAs. Machine learning (ML) techniques have made tremendous progress in many fields in the past few years. Although their usage in protein-related fields has a long history, the use of ML methods in predicting RNA tertiary structures is new and rare. Here, we review the recent advances of using ML methods on RNA structure predictions and discuss the advantages and limitation, the difficulties and potentials of these approaches when applied in the field.

关键词： Imaging spectroscopy   Inversion   Gaussian processes   Radiative transfer modeling   Agricultural monitoring   EnMAP  

摘要： Hyperspectral acquisitions have proven to be the most informative Earth observation data source for the estimation of nitrogen (N) content, which is the main limiting nutrient for plant growth and thus agricultural production. In the past, empirical algorithms have been widely employed to retrieve information on this biochemical plant component from canopy reflectance. However, these approaches do not seek for a cause-effect relationship based on physical laws. Moreover, most studies solely relied on the correlation of chlorophyll content with nitrogen, and thus neglected the fact that most N is bound in proteins. Our study presents a hybrid retrieval method using a physically-based approach combined with machine learning regression to estimate crop N content. Within the workflow, the leaf optical properties model PROSPECT-PRO including the newly calibrated specific absorption coefficients (SAC) of proteins, was coupled with the canopy reflectance model 4SAIL to PROSAIL-PRO. The latter was then employed to generate a training database to be used for advanced probabilistic machine learning methods: a standard homoscedastic Gaussian process (GP) and a heteroscedastic GP regression that accounts for signal-to-noise relations. Both GP models have the property of providing confidence intervals for the estimates, which sets them apart from other machine learners. Moreover, a GP-based sequential backward band removal algorithm was employed to analyze the band-specific information content of PROSAIL-PRO simulated spectra for the estimation of aboveground N. Data from multiple hyperspectral field campaigns, carried out in the framework of the future satellite mission Environmental Mapping and Analysis Program (EnMAP), were exploited for validation. In these campaigns, corn and winter wheat spectra were acquired to simulate spectral EnMAP data. Moreover, destructive N measurements from leaves, stalks and fruits were collected separately to enable plant-organ-specific validati

关键词： Runoff model   Nature-inspired optimizers   Malaysia  

摘要： Rainfall and runoff are considered the main components in the hydrological cycle. Developing an accurate model to capture the dynamic connection between rainfall and runoff remains a problematic task for engineers. Several studies have been carried out to develop models to accurately predict the changes in runoff from rainfall. However, these models have limitations in terms of accuracy and complexity when large numbers of parameters are needed. Therefore, recently, with the advancement of data-driven techniques, a vast number of hydrologists have adopted models to predict changes in runoff. However, data-driven models still encounter several limitations related to hyperparameter optimization and overfitting. Hence, there is a need to improve these models to overcome these limitations. In this study, data-driven techniques such as a Multi-Layer Perceptron (MLP) neural network and Least Squares Support Vector Machine (LSSVM) are integrated with an advanced nature-inspired optimizer, namely, Harris Hawks Optimization (HHO) to model the rainfall-runoff relationship. Five different scenarios will be examined based on the autocorrelation function (ACF), cross-correlation function (CCF) and partial autocorrelation function (PACF). Finally, for comprehensive analysis, the performance of the proposed model will then be compared with integrated data-driven techniques with particle swarm optimization (PSO). The results revealed that all the augmented models with HHO outperformed other integrated models with PSO in predicting the changes in runoff. In addition, a high level of accuracy in predicting runoff values was achieved when HHO was integrated with LSSVM.

关键词： Metaheuristic evolutionary   Classification   Landslide perdition  

摘要： The present study aims to assess the superiority of the metaheuristic evolutionary when compared to the conventional machine learning classification techniques for landslide occurrence estimation. To evaluate and compare the applicability of these metaheuristic algorithms, a real-world problem of landslide assessment (i.e., including 266 records and fifteen landslide conditioning factors) is selected. In the first step, seven of the most common traditional classification techniques are applied. Then, after introducing the elite model, it is optimized using six state-of-the-art metaheuristic evolutionary techniques. The results show that applying the proposed evolutionary algorithms effectively increases the prediction accuracy from 81.6 to the range (87.8-98.3%) and the classification ratio from 58.3% to the range (60.1-85.0%).

关键词： concurrent   interleaved   RNN   BiLSTM   SCCRF   activity recognition   Smart Home  

摘要： Human activity recognition has become an important research topic within the field of pervasive computing, ambient assistive living (AAL), robotics, health-care monitoring, and many more. Techniques for recognizing simple and single activities are typical for now, but recognizing complex activities such as concurrent and interleaving activity is still a major challenging issue. In this paper, we propose a two-phase hybrid deep machine learning approach using bi-directional Long-Short Term Memory (BiLSTM) and Skip-Chain Conditional random field (SCCRF) to recognize the complex activity. BiLSTM is a sequential generative deep learning inherited from Recurrent Neural Network (RNN). SCCRFs is a distinctive feature of conditional random field (CRF) that can represent long term dependencies. In the first phase of the proposed approach, we recognized the concurrent activities using the BiLSTM technique, and in the second phase, SCCRF identifies the interleaved activity. Accuracy of the proposed framework against the counterpart state-of-art methods using the publicly available datasets in a smart home environment is analyzed. Our experiment's result surpasses the previously proposed approaches with an average accuracy of more than 93%.