Cooperative learning for multiview analysis.
Cooperative learning for multiview analysis.
复制标题
多视点分析的合作学习。
DOI：
10.1073/pnas.2202113119
复制
发表时间：
2022-09-20
影响因子：
11.1
通讯作者：
中科院分区：
文献类型：
--
作者：
多视图分析与“组学”的数据，如基因组学和蛋白质组学，测量一组共同的样本代表了越来越重要的挑战，在生物学和医学。根据“融合”发生的时间，常用的方法可以大致分为早期融合和晚期融合。我们介绍了一种监督学习算法-“合作学习”-它包括早期和晚期的融合和混合版本的这些方法。该算法鼓励来自不同观点的预测达成一致，并以数据自适应的方式选择一致程度。通过利用多组学中的对齐信号，它可以对疾病分类和治疗反应预测等任务进行更好的预测，并对改善诊断和治疗具有重要意义。我们提出了一种具有多组特征（“视图”）的监督学习方法。多视图问题在生物学和医学中尤其重要，其中“组学”数据，例如基因组学、蛋白质组学和放射组学，是在一组共同的样本上测量的。“合作学习”将通常的预测平方误差损失与“一致性”惩罚相结合，以鼓励来自不同数据视图的预测达成一致。通过改变协议惩罚的权重，我们得到了一个连续的解决方案，包括众所周知的早期和晚期融合方法。协同学习以自适应的方式选择一致度（或融合度），使用验证集或交叉验证来估计测试集预测误差。我们的拟合过程的一个版本是模块化的，其中可以选择不同的拟合机制（例如，lasso、随机森林、boosting或神经网络），适用于不同的数据视图。在合作正则化线性回归的设置中，该方法结合了套索惩罚和协议惩罚，产生特征稀疏性。当不同的数据视图在它们的信号中共享一些潜在的关系时，该方法可以特别强大，这些关系可以用来增强信号。我们表明，合作学习实现了更高的预测精度上的模拟数据和真实的多组学的例子，分娩预测。通过利用对齐的信号并允许不同模态的灵活拟合机制，合作学习为多组学数据融合提供了一种强大的方法。
Multiview analysis with “-omics” data, such as genomics and proteomics, measured on a common set of samples represents an increasingly important challenge in biology and medicine. Commonly used approaches can be broadly categorized into early and late fusion, depending on when “fusion” occurs. We introduce a supervised learning algorithm—“cooperative learning”—that encompasses both early and late fusion and blended versions of these methods. This algorithm encourages the predictions from different views to agree and chooses the degree of agreement in a data-adaptive manner. By leveraging aligned signals in multiomics, it can yield better predictions on tasks such as disease classification and treatment response prediction and has implications for improving diagnostics and therapeutics. We propose a method for supervised learning with multiple sets of features (“views”). The multiview problem is especially important in biology and medicine, where “-omics” data, such as genomics, proteomics, and radiomics, are measured on a common set of samples. “Cooperative learning” combines the usual squared-error loss of predictions with an “agreement” penalty to encourage the predictions from different data views to agree. By varying the weight of the agreement penalty, we get a continuum of solutions that include the well-known early and late fusion approaches. Cooperative learning chooses the degree of agreement (or fusion) in an adaptive manner, using a validation set or cross-validation to estimate test set prediction error. One version of our fitting procedure is modular, where one can choose different fitting mechanisms (e.g., lasso, random forests, boosting, or neural networks) appropriate for different data views. In the setting of cooperative regularized linear regression, the method combines the lasso penalty with the agreement penalty, yielding feature sparsity. The method can be especially powerful when the different data views share some underlying relationship in their signals that can be exploited to boost the signals. We show that cooperative learning achieves higher predictive accuracy on simulated data and real multiomics examples of labor-onset prediction. By leveraging aligned signals and allowing flexible fitting mechanisms for different modalities, cooperative learning offers a powerful approach to multiomics data fusion.
登录
查看更多内容
影响因子：
8.8
作者：
Ponzetti M;Capulli M;Angelucci A;Ventura L;Monache SD;Mercurio C;Calgani A;Sanità P;Teti A;Rucci N
通讯作者：
Rucci N
影响因子：
2.1
作者：
Gross, Samuel M.;Tibshirani, Robert
通讯作者：
Tibshirani, Robert
DOI：
10.1038/nrg.2018.4
发表时间：
2018-05
期刊：
Nature reviews. Genetics
影响因子：
--
作者：
Karczewski KJ;Snyder MP
通讯作者：
Snyder MP
影响因子：
17.1
作者：
Stelzer IA;Ghaemi MS;Han X;Ando K;Hédou JJ;Feyaerts D;Peterson LS;Rumer KK;Tsai ES;Ganio EA;Gaudillière DK;Tsai AS;Choisy B;Gaigne LP;Verdonk F;Jacobsen D;Gavasso S;Traber GM;Ellenberger M;Stanley N;Becker M;Culos A;Fallahzadeh R;Wong RJ;Darmstadt GL;Druzin ML;Winn VD;Gibbs RS;Ling XB;Sylvester K;Carvalho B;Snyder MP;Shaw GM;Stevenson DK;Contrepois K;Angst MS;Aghaeepour N;Gaudillière B
通讯作者：
Gaudillière B
影响因子：
10.3
作者：
Gentles, Andrew J.;Bratman, Scott V.;Diehn, Maximilian
通讯作者：
Diehn, Maximilian