A forecasting system of patent application counts is studied in this paper. The optimization model proposed in the research is based on support vector machines （SVM）, in which cross-validation algorithm is used for ...A forecasting system of patent application counts is studied in this paper. The optimization model proposed in the research is based on support vector machines （SVM）, in which cross-validation algorithm is used for preferences selection. Results of data simulation show that the proposed method has higher forecasting precision power and stronger generalization ability than BP neural network and RBF neural network. In addi- tion, it is feasible and effective in forecasting patent application counts.展开更多
利用可见/近红外光谱技术对冷却肉菌落总数和颜色进行快速、无损检测。采用400~1 100 nm可见/近红外光谱成像系统,获取54个冷却肉样本表面的光谱图像,采用主成分分析结合马氏距离方法对异常光谱进行判别及剔除。通过Gompertz分布函数对...利用可见/近红外光谱技术对冷却肉菌落总数和颜色进行快速、无损检测。采用400~1 100 nm可见/近红外光谱成像系统,获取54个冷却肉样本表面的光谱图像,采用主成分分析结合马氏距离方法对异常光谱进行判别及剔除。通过Gompertz分布函数对散射特征曲线进行拟合,得到表征光谱信息的Gompertz参数,结合支持向量机算法建立冷却肉菌落总数和肉色L*的预测模型。α、β、θ、δ组合和α、β、δ组合建模对细菌总数预测效果最好,预测相关系数分别为0.937和0.935,预测标准差为0.600 lg CFU/g和0.702 lg CFU/g。β、δ组合建模对肉色L*预测效果较好,预测相关系数达到0.930,预测标准差为1.515。研究结果表明利用Vis/NIR光谱散射特征结合支持向量机可以实现冷却肉品质的快速、高效、无损伤检测。展开更多
基金Sponsored by "985" Philosophy and Social Science Innovation Base of the Ministry of Education of China (107008200400024)
文摘A forecasting system of patent application counts is studied in this paper. The optimization model proposed in the research is based on support vector machines （SVM）, in which cross-validation algorithm is used for preferences selection. Results of data simulation show that the proposed method has higher forecasting precision power and stronger generalization ability than BP neural network and RBF neural network. In addi- tion, it is feasible and effective in forecasting patent application counts.
文摘利用可见/近红外光谱技术对冷却肉菌落总数和颜色进行快速、无损检测。采用400~1 100 nm可见/近红外光谱成像系统,获取54个冷却肉样本表面的光谱图像,采用主成分分析结合马氏距离方法对异常光谱进行判别及剔除。通过Gompertz分布函数对散射特征曲线进行拟合,得到表征光谱信息的Gompertz参数,结合支持向量机算法建立冷却肉菌落总数和肉色L*的预测模型。α、β、θ、δ组合和α、β、δ组合建模对细菌总数预测效果最好,预测相关系数分别为0.937和0.935,预测标准差为0.600 lg CFU/g和0.702 lg CFU/g。β、δ组合建模对肉色L*预测效果较好,预测相关系数达到0.930,预测标准差为1.515。研究结果表明利用Vis/NIR光谱散射特征结合支持向量机可以实现冷却肉品质的快速、高效、无损伤检测。
