关键词： 专利质量   机器学习   区块链   专利质量分析   分类预测  

摘要： [目的/意义]本研究在一定程度上可以减少专利审查员的时间成本和主观性并提高业务素质,也可以为专利申请者的专利布局提供参考。[方法/过程]本研究在专利质量界定与指标体系构建的基础上,提出基于机器学习的组合模型用于专利质量的分析与分类预测,并以新兴产业的区块链技术专利为例展开研究。该模型由自组织映射、核主成分分析以及支持向量机3种方法组成,其实现过程包括两阶段。第一阶段通过自组织映射(SOM)对从国家知识产权局专利数据库收集的21496项区块链技术专利数据进行分析并界定专利质量类别;第二阶段,通过核主成分分析(KPCA)对专利数据进行降维降噪处理,经过处理的专利数据再由支持向量机(SVM)得出分类结果。[结果/结论]在对区块链技术专利质量分类模型进行训练后,基于3306项区块链专利历史数据来验证训练模型的性能,实验结果的匹配度达到87.26%。因此,本研究提出的组合模型能够有效地对专利质量进行分类与预测。

关键词： 通货紧缩   债务—通缩   结构性去杠杆   供给侧结构性改革   机器学习  

摘要： 最近30年间中国已经发生过五轮通货紧缩(简称通缩),考虑到通缩对经济的冲击往往较大,因此有必要识别其背后的影响因素。已有文献主要使用结构向量自回归等传统线性方法开展研究,未能全面考察各个因素对通缩的非线性影响,而且研究的变量个数较少。为了弥补已有研究的不足,本文使用多种非线性机器学习方法识别通缩的影响因素,尤其是对2012年之前发生的三轮以消费价格指数和生产价格指数双双下跌为表征的全局性通缩和2012年之后发生的两轮以生产价格指数下跌但是消费价格指数上涨为表征的生产部门通缩的影响因素进行了对比分析。结果表明,生产部门通缩和全局性通缩的影响因素存在显著差异:债务因素对全局性通缩的影响较小,但是2012年以来各部门债务负担不断加重,债务因素已经成为生产部门通缩最主要的影响因素;供给侧因素对全局性通缩的影响同样较小,但是对生产部门通缩的影响明显提高;需求侧因素和货币政策因素对全局性通缩的影响较大,但是对生产部门通缩的影响大幅下降。鉴于此,本文认为,中央正在推进的结构性去杠杆和供给侧结构性改革对于2012年以来频繁发生的生产部门通缩具有很好的针对性,未来应该加强两类措施以防范债务因素和供给侧因素再度触发通缩风险。

关键词： 流线型箱梁   机器学习   颤振  

摘要： 为在工程设计初步阶段快速评价流线型箱梁颤振性能,基于4种机器学习算法(支持向量机回归、神经网络、随机森林回归和高斯过程回归),利用自由振动风洞试验获取的15种不同流线型箱梁断面在5种不同风攻角下的颤振临界风速,建立了从气动外形与动力参数到颤振临界风速的颤振临界风速预测模型。结果表明:支持向量机回归模型预测结果整体精度最高,且在实际桥梁上表现性能较好;神经网络模型较差,但其相对误差仍远低于JTG/T 3360-01-2018《公路桥梁抗风设计规范》中提供的3种不同颤振临界风速简化计算公式的结果。该研究的结果符合预期要求,未来可进一步扩充数据集,以期成为对抗风设计工作者提供高精度颤振性能评估的有力工具。

关键词： 财务管理风险   机器学习   WOE-IV信息量  

摘要： 复杂的国际贸易环境下,上市公司往往面临着巨大财务管理风险,建立有效的财务管理风险评估模型,对于企业发展和相关利益方进行决策,监管和投资都有着重要意义。基于此,本文首先结合企业财务和非财务方面指标,构建了财务管理风险预警评估指标体系,然后运用多种机器学习模型进行风险预测,并使用交叉验证对各模型的准确率和稳定性进行评估,最后通过最优模型解释机制筛选出重要影响指标。实证结果表明:XGBoost集成学习模型综合评估结果最优,对于该模型输出的各指标影响程度,本文通过使用WOE-IV信息量佐证了其合理性。

关键词： 机器学习   差分隐私   图像分类   推理攻击   影子模型  

摘要： 图像分类中的差分隐私算法在通过添加噪声的方式提高机器学习模型的隐私保护能力的同时,容易造成模型分类准确度的下降。针对以上问题,提出了一种基于特征映射的差分隐私保护机器学习方法,该方法结合预训练神经网络和影子模型训练技术,以差分向量的形式将原数据样本的特征向量映射到高维向量空间,缩短样本在高维向量空间的距离,以减小模型更新造成的隐私信息泄露风险,同时提高机器学习模型的隐私保护能力和分类能力。由MNIST和CIFAR-10数据集上的实验结果表明,ε分别等于0.01和0.11的ε0-差分隐私的模型的分类准确度分别提高到了99%和96%,说明所提方法训练的模型相比DP-SGD等现有多种常用差分隐私算法,能在更低隐私预算下保持更强的分类能力;且在两个数据集上针对该模型的推理攻击成功率降低为10%,其对推理攻击的防御能力相比传统图像分类的CNN模型有较大幅度的提升。

关键词： 金属玻璃   各向异性度   机器学习   动力学   不均匀性   结构特征  

摘要： In crystalline solids,plastic deformation is determined by the motion of structural defects,particularly *** for glassy liquids and disordered solids,it has long been assumed that localized structural rearrangements induced by mechanical load or thermal stimulus also occur at localized defects[1-3].

关键词： heat-resistant nickel alloys   high-temperature strength   phase stability   simulation  

摘要： Data on the nature of changing the heat resistance of nickel alloys, which are used for making the most critical parts, is of great topicality for the design of gas turbine engines of high resource. A model of changing the heat resistance and an analytical expression that makes it possible to determine the thermal stability parameter for each alloy composition are obtained using the machine learning method. The long-term strength limit was estimated and extrapolated using the Larson-Miller temperature-time dependence. The adequacy of the obtained model is confirmed by the satisfactory convergence of the experimental and calculated results.

关键词： Support vector machine   Logistic regression   Artificial neural network   Pattern recognition   sEMG   Shoulder motions  

摘要： Machine learning (ML) methods have been previously applied and compared in pattern recognition of hand and elbow motions based on surface electromyographic (sEMG) signals. However, there are only a few studies that have investigated the ML methods for shoulder motion pattern recognition. This study compared the efficiency of ML algorithms, including support vector machine (SVM), logistic regression (LR), and artificial neural network (ANN) in processing sEMG signals for shoulder motion pattern recognition. This study also investigated the the effects of sliding time window epoch on the recognition accuracy. Eighteen healthy subjects were recruited for this study, their EMG signals were collected from twelve muscles during performing activities of daily living (ADL) motions including drinking, pushing forward/pulling backward, and abduction/adduction. The 80 % of recoded sEMG datasets were used for model training to build the ML models and 20 % were used for model validation and determination of the accuracy of ML algorithms in motion pattern recognition. The influence of sliding time window sizes was studied for algorithm optimization. Statistical analysis was performed to determine the difference in the accuracy of ML methods. Results showed that there was a significant difference among the three machine learning methods and different sliding time window sizes. There was not a significant difference in overlapping time. The highest accuracy was 97.41 +/- 1.8 % using the SVM method with a sliding time window of 270 ms. Machine learning techniques provided a quick approach for shoulder motion pattern recognition. The better classifier for pattern recognition of shoulder motion was SVM.

关键词： Receivers   Machine learning   Training   Inverse problems   Electromagnetics   Data models   Transmitters   Convolutional neural network (CNN)   electromagnetic (EM) inversion   extreme learning machine (ELM)   machine learning  

摘要： In practical applications, electromagnetic measurements often contain damaged data due to malfunctioning receivers, which can severely influence the inversion performance. Thus, in this letter, a new receiver approximation machine learning (RAML) method is proposed to repair the data from the damaged receivers and the repaired data are the input to the dual-module nonlinear mapping module-image enhancing module machine learning scheme for the 2-D inverse scattering problem. The RAML method has been tested for the inversion of both measured transverse electric and transverse magnetic datasets provided by Institute Fresnel. Excellent inversion performance has been obtained when 10 out of 241 (4.15%) receivers are damaged. Furthermore, in an extreme situation, the RAML method can also have good inversion performance even though half of the receivers are damaged.

关键词： Youth in foster care   Permanency   Transitional support programs   Machine learning   Predictive modeling   Algorithmic fairness  

摘要： Background: Youth who exit the nation's foster care system without permanency are at high risk of experiencing difficulties during the transition to adulthood. Objective: To present an illustrative test of whether an algorithmic decision aid could be used to identify youth at risk of existing foster care without permanency. Methods: For youth placed in foster care between ages 12 and 14, we assessed the risk of exiting care without permanency by age 18 based on their child welfare service involvement history. To develop predictive risk models, 28 years (1991-2018) of child welfare service records from California were used. Performances were evaluated using F1, AUC, and precision and recall scores at k %. Algorithmic racial bias and fairness was also examined. Results: The gradient boosting decision tree and random forest showed the best performance (F1 score = .54-.55, precision score = .62, recall score = .49). Among the top 30 % of youth the model identified as high risk, half of all youth who exited care without permanency were accurately identified four to six years prior to their exit, with a 39 % error rate. Although racial disparities between Black and White youth were observed in imbalanced error rates, calibration and predictive parity were satisfied. Conclusions: Our study illustrates the manner in which potential applications of predictive analytics, including those designed to achieve universal goals of permanency through more targeted allocations of resources, can be tested. It also assesses the model using metrics of fairness.