Curriculum Learning of Visual Attribute Clusters for Multi-Task Classification
Curriculum Learning of Visual Attribute Clusters for Multi-Task Classification
复制标题
DOI：
10.1016/j.patcog.2018.02.028
复制
发表时间：
2017-09
期刊：
影响因子：
--
通讯作者：
N. Sarafianos;Theodore Giannakopoulos;Christophoros Nikou;I. Kakadiaris
中科院分区：
文献类型：
--
作者：
N. Sarafianos;Theodore Giannakopoulos;Christophoros Nikou;I. Kakadiaris
视觉属性，来自简单对象（例如，背包，帽子）到软生物测定（例如，性别、身高、服装）已被证明是用于诸如图像描述和人类识别的许多应用的强大的表示方法。在本文中，我们介绍了一种新的方法，联合收割机的优点，多任务和课程学习的视觉属性分类框架。单个任务在基于它们的相关性执行层次聚类之后被分组。通过在集群之间转移知识，在课程学习设置中学习任务集群。每个集群内的学习过程在多任务分类设置中执行。通过利用获得的知识，我们加快了流程并提高了性能。我们通过消融研究和协变量的详细分析证明了我们的方法的有效性，在各种公开可用的人体站立数据集上，全身可见。大量的实验证明，所提出的方法提高了4%-10%的性能。
Visual attributes, from simple objects (e.g., backpacks, hats) to soft-biometrics (e.g., gender, height, clothing) have proven to be a powerful representational approach for many applications such as image description and human identification. In this paper, we introduce a novel method to combine the advantages of both multi-task and curriculum learning in a visual attribute classification framework. Individual tasks are grouped after performing hierarchical clustering based on their correlation. The clusters of tasks are learned in a curriculum learning setup by transferring knowledge between clusters. The learning process within each cluster is performed in a multi-task classification setup. By leveraging the acquired knowledge, we speed-up the process and improve performance. We demonstrate the effectiveness of our method via ablation studies and a detailed analysis of the covariates, on a variety of publicly available datasets of humans standing with their full-body visible. Extensive experimentation has proven that the proposed approach boosts the performance by 4%–10%.