arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.12377cs.LGstat.ML
收藏

动态标度深度典型相关分析

Dynamically-Scaled Deep Canonical Correlation Analysis

Tel Aviv University(特拉维夫大学)

Tomer Friedlander, Lior Wolf

摘要:

典型相关分析(CCA)是一种通过寻找两个视图的最大相关线性投影来进行特征提取的方法。在文献中已经介绍了CCA的几种变体,特别是基于深度神经网络的变体,用于学习两个视图的高度相关的非线性变换。由于这些模型通常是参数化的,因此它们的可学习参数在训练过程之后仍然独立于输入,这可能会限制它们学习高度相关表示的能力。本文提出了一种新的动态标度方法来训练输入依赖的典型相关模型。在我们的深度CCA模型中,最后一层的参数由第二个神经网络缩放,该网络以模型的输入为条件,从而产生依赖于输入样本的参数化。我们在多个数据集上评估了我们的模型,并证明了与传统的基于参数化CCA的模型相比,学习到的表示更相关,并且也获得了更好的检索结果。我们的代码可在https://github.com/tomerfr/DynamicallyScaledDeepCCA上获得。

英文摘要:

Canonical Correlation Analysis (CCA) is a method for feature extraction of two views by finding maximally correlated linear projections of them. Several variants of CCA have been introduced in the literature, in particular, variants based on deep neural networks for learning highly correlated nonlinear transformations of two views. As these models are parameterized conventionally, their learnable parameters remain independent of the inputs after the training process, which may limit their capacity for learning highly correlated representations. We introduce a novel dynamic scaling method for training an input-dependent canonical correlation model. In our deep-CCA models, the parameters of the last layer are scaled by a second neural network that is conditioned on the model's input, resulting in a parameterization that is dependent on the input samples. We evaluate our model on multiple datasets and demonstrate that the learned representations are more correlated in comparison to the conventionally-parameterized CCA-based models and also obtain preferable retrieval results. Our code is available at https://github.com/tomerfr/DynamicallyScaledDeepCCA.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑