Clean label food is a rising consumer trend in the food industry.Milk protein concentrate(MPC)and reduced-calcium milk protein concentrate(RCMPC)could serve as natural emulsifiers and increase the total protein conten...Clean label food is a rising consumer trend in the food industry.Milk protein concentrate(MPC)and reduced-calcium milk protein concentrate(RCMPC)could serve as natural emulsifiers and increase the total protein content of ice cream products.The objectives of this study were to determine and compare the effects of MPC and RCMPC on ice cream composition,mix viscosity,storage stability,meltdown rate,and texture.A base formulation with 3%non-fat dry milk(NFDM)and no added emulsifiers was set as the control.Three levels of MPC or RCMPC(each powder containing 85%protein)at 1%,2%,and 3%were incorporated by replacing equivalent amounts of NFDM and keeping other ingredients unchanged.All ice cream treatments were processed with a target overrun of 70%and hardened at−25℃in a blast freezer.Additions of MPC or RCMPC at 1%,2%,and 3%corresponded to increases in protein content of ice cream by 15%,30%,and 45%,respectively.The viscosity of the ice-cream mix increased with increasing levels of MPC or RCMPC.In general,higher protein samples had slower meltdown rate and higher values of hardness and adhesiveness,but the trends were inconsistent.No shrinkage in volume was observed in any ice cream stored at−25℃after 180 days.However,an additional storage stability study revealed that the control showed significant shrinkage after 60 days(−6.5%±1.5%),90 days(−7.1%±1.8%),and 180 days(−7.9%±1.1%)in a typical household-style freezer at−13℃.MPC at 1%also showed significant shrinkage after 180 days,while samples with RCMPC at any levels showed no shrinkage at all.Ice cream manufacturers may consider MPC and RCMPC natural alternatives to synthetic emulsifiers,with RCMPC being more effective than MPC in terms of ice cream storage stability.展开更多
心电图(electrocardiogram,ECG)异常的自动检测是一个典型的多标签分类问题,训练分类器需要大量有高质量标签的样本.但心电数据集异常标签经常缺失或错误,如何清洗弱标签得到干净的心电数据集是一个亟待解决的问题.在一个标签完整且准...心电图(electrocardiogram,ECG)异常的自动检测是一个典型的多标签分类问题,训练分类器需要大量有高质量标签的样本.但心电数据集异常标签经常缺失或错误,如何清洗弱标签得到干净的心电数据集是一个亟待解决的问题.在一个标签完整且准确的示例数据集辅助下,提出一种基于异常特征模式(abnormality-feature pattern,AFP)的方法对弱标签心电数据进行标签清洗,以获取所有正确的异常标签.清洗分2个阶段,即基于聚类的规则构造和基于迭代的标签清洗.在第1阶段,通过狄利克雷过程混合模型(Dirichlet process mixture model,DPMM)聚类,识别每个异常标签对应的不同特征模式,进而构建异常发现规则、排除规则和1组二分类器.在第2阶段,根据发现和排除规则辨识初始相关标签集,然后根据二分类器迭代扩展相关标签并排除不相关标签.AFP方法捕捉了示例数据集和弱标签数据集的共享特征模式,既应用了人的知识,又充分利用了正确标记的标签;同时,渐进地去除错误标签和填补缺失标签,保证了标签清洗的可靠性.真实和模拟数据集上的实验证明了AFP方法的有效性.展开更多
该研究以猪皮蛋白粉(porcine skin protein powder,PSPP)为原料,以大豆油为油相制备乳液,研究了不同质量分数的PSPP(0.5%~3.0%)对乳液的理化性质、流变特性及稳定性的影响。结果表明,随着PSPP添加量的增加,更多的蛋白颗粒参与界面吸附...该研究以猪皮蛋白粉(porcine skin protein powder,PSPP)为原料,以大豆油为油相制备乳液,研究了不同质量分数的PSPP(0.5%~3.0%)对乳液的理化性质、流变特性及稳定性的影响。结果表明,随着PSPP添加量的增加,更多的蛋白颗粒参与界面吸附和连续相网络形成,导致乳液液滴尺寸降低(300~40μm),乳液持水力和流变学特性(表观黏度、模量)增加,进而提高了乳液的贮藏、离心、冻融稳定性。研究结果可为猪皮的高值利用拓展新途径,并为PSPP作为乳化型功能性组分在食品中应用提供理论依据和指导意义。展开更多
以猪皮蛋白粉(porcine skin protein powder,PSPP)为原料,大豆油为油相制备乳液,再经热诱导形成猪皮蛋白粉基乳液凝胶,研究PSPP的浓度和加热处理时间与次数对乳液稳定性、乳液凝胶质构及持水力的影响。结果表明,随着PSPP浓度的增加,加...以猪皮蛋白粉(porcine skin protein powder,PSPP)为原料,大豆油为油相制备乳液,再经热诱导形成猪皮蛋白粉基乳液凝胶,研究PSPP的浓度和加热处理时间与次数对乳液稳定性、乳液凝胶质构及持水力的影响。结果表明,随着PSPP浓度的增加,加热后乳液析出水层逐渐减小,乳液结构保持更加完整,液滴尺寸变化显著减小,乳液的表观黏度增大。质构和持水力分析显示,随着PSPP浓度的增加,乳液凝胶的硬度、黏聚性及持水力均显著增强。激光共聚焦显微镜显示,随着PSPP浓度的增加,乳液凝胶颗粒的直径明显减小,油水界面的蛋白覆盖率增大。当PSPP质量浓度为30 g/L时,随着热处理时间延长和次数的增加,乳液凝胶质构特性变化较小,这表明PSPP乳液凝胶可耐受长时间和反复加热处理。展开更多
基于主动学习的标签噪声清洗方法(Active label noise cleaning,ALNC)是一种通过主动学习筛选疑似噪声样本,进而交给人工专家进行再标记的标签噪声清洗方法.虽然该方法既有很好的噪声识别效果又能保持原有数据的完整性,但仍存在人工额...基于主动学习的标签噪声清洗方法(Active label noise cleaning,ALNC)是一种通过主动学习筛选疑似噪声样本,进而交给人工专家进行再标记的标签噪声清洗方法.虽然该方法既有很好的噪声识别效果又能保持原有数据的完整性,但仍存在人工额外标记代价较高的问题,即筛选出的疑似噪声样本中存在一定比例的正常样本.为了解决这一问题,降低标签噪声清洗过程中的人工额外检验代价,本文提出了一种基于SPXY(Sample Set Partitioning based on Joint X-Y Distance Sampling)采样的标签噪声主动清洗方法(Active label noise cleaning based on SPXY,SPXYALNC),该方法在主动学习筛选疑似噪声样本的过程中结合了SPXY采样方法,这样既考虑了样本的不确定性,又考虑了样本的代表性,并且在原有标准数据集上针对分类问题进行了实验,实验结果表明该方法在保持原有噪声识别效果的同时可以明显降低人工额外检验代价.展开更多
文摘Clean label food is a rising consumer trend in the food industry.Milk protein concentrate(MPC)and reduced-calcium milk protein concentrate(RCMPC)could serve as natural emulsifiers and increase the total protein content of ice cream products.The objectives of this study were to determine and compare the effects of MPC and RCMPC on ice cream composition,mix viscosity,storage stability,meltdown rate,and texture.A base formulation with 3%non-fat dry milk(NFDM)and no added emulsifiers was set as the control.Three levels of MPC or RCMPC(each powder containing 85%protein)at 1%,2%,and 3%were incorporated by replacing equivalent amounts of NFDM and keeping other ingredients unchanged.All ice cream treatments were processed with a target overrun of 70%and hardened at−25℃in a blast freezer.Additions of MPC or RCMPC at 1%,2%,and 3%corresponded to increases in protein content of ice cream by 15%,30%,and 45%,respectively.The viscosity of the ice-cream mix increased with increasing levels of MPC or RCMPC.In general,higher protein samples had slower meltdown rate and higher values of hardness and adhesiveness,but the trends were inconsistent.No shrinkage in volume was observed in any ice cream stored at−25℃after 180 days.However,an additional storage stability study revealed that the control showed significant shrinkage after 60 days(−6.5%±1.5%),90 days(−7.1%±1.8%),and 180 days(−7.9%±1.1%)in a typical household-style freezer at−13℃.MPC at 1%also showed significant shrinkage after 180 days,while samples with RCMPC at any levels showed no shrinkage at all.Ice cream manufacturers may consider MPC and RCMPC natural alternatives to synthetic emulsifiers,with RCMPC being more effective than MPC in terms of ice cream storage stability.
文摘心电图(electrocardiogram,ECG)异常的自动检测是一个典型的多标签分类问题,训练分类器需要大量有高质量标签的样本.但心电数据集异常标签经常缺失或错误,如何清洗弱标签得到干净的心电数据集是一个亟待解决的问题.在一个标签完整且准确的示例数据集辅助下,提出一种基于异常特征模式(abnormality-feature pattern,AFP)的方法对弱标签心电数据进行标签清洗,以获取所有正确的异常标签.清洗分2个阶段,即基于聚类的规则构造和基于迭代的标签清洗.在第1阶段,通过狄利克雷过程混合模型(Dirichlet process mixture model,DPMM)聚类,识别每个异常标签对应的不同特征模式,进而构建异常发现规则、排除规则和1组二分类器.在第2阶段,根据发现和排除规则辨识初始相关标签集,然后根据二分类器迭代扩展相关标签并排除不相关标签.AFP方法捕捉了示例数据集和弱标签数据集的共享特征模式,既应用了人的知识,又充分利用了正确标记的标签;同时,渐进地去除错误标签和填补缺失标签,保证了标签清洗的可靠性.真实和模拟数据集上的实验证明了AFP方法的有效性.
文摘该研究以猪皮蛋白粉(porcine skin protein powder,PSPP)为原料,以大豆油为油相制备乳液,研究了不同质量分数的PSPP(0.5%~3.0%)对乳液的理化性质、流变特性及稳定性的影响。结果表明,随着PSPP添加量的增加,更多的蛋白颗粒参与界面吸附和连续相网络形成,导致乳液液滴尺寸降低(300~40μm),乳液持水力和流变学特性(表观黏度、模量)增加,进而提高了乳液的贮藏、离心、冻融稳定性。研究结果可为猪皮的高值利用拓展新途径,并为PSPP作为乳化型功能性组分在食品中应用提供理论依据和指导意义。
文摘以猪皮蛋白粉(porcine skin protein powder,PSPP)为原料,大豆油为油相制备乳液,再经热诱导形成猪皮蛋白粉基乳液凝胶,研究PSPP的浓度和加热处理时间与次数对乳液稳定性、乳液凝胶质构及持水力的影响。结果表明,随着PSPP浓度的增加,加热后乳液析出水层逐渐减小,乳液结构保持更加完整,液滴尺寸变化显著减小,乳液的表观黏度增大。质构和持水力分析显示,随着PSPP浓度的增加,乳液凝胶的硬度、黏聚性及持水力均显著增强。激光共聚焦显微镜显示,随着PSPP浓度的增加,乳液凝胶颗粒的直径明显减小,油水界面的蛋白覆盖率增大。当PSPP质量浓度为30 g/L时,随着热处理时间延长和次数的增加,乳液凝胶质构特性变化较小,这表明PSPP乳液凝胶可耐受长时间和反复加热处理。
文摘基于主动学习的标签噪声清洗方法(Active label noise cleaning,ALNC)是一种通过主动学习筛选疑似噪声样本,进而交给人工专家进行再标记的标签噪声清洗方法.虽然该方法既有很好的噪声识别效果又能保持原有数据的完整性,但仍存在人工额外标记代价较高的问题,即筛选出的疑似噪声样本中存在一定比例的正常样本.为了解决这一问题,降低标签噪声清洗过程中的人工额外检验代价,本文提出了一种基于SPXY(Sample Set Partitioning based on Joint X-Y Distance Sampling)采样的标签噪声主动清洗方法(Active label noise cleaning based on SPXY,SPXYALNC),该方法在主动学习筛选疑似噪声样本的过程中结合了SPXY采样方法,这样既考虑了样本的不确定性,又考虑了样本的代表性,并且在原有标准数据集上针对分类问题进行了实验,实验结果表明该方法在保持原有噪声识别效果的同时可以明显降低人工额外检验代价.