Leveraging Prior Concept Learning Improves Generalization From Few Examples in Computational Models of Human Object Recognition.
Leveraging Prior Concept Learning Improves Generalization From Few Examples in Computational Models of Human Object Recognition.
复制标题
DOI：
10.3389/fncom.2020.586671
复制
发表时间：
2020
影响因子：
3.2
通讯作者：
Riesenhuber M
中科院分区：
文献类型：
--
作者：
Rule JS;Riesenhuber M
人类可以从稀疏的数据中快速准确地学习新的视觉概念，有时只是一个例子。人工神经网络的令人印象深刻的性能，分层池传入跨尺度和位置表明，人类视觉系统的分层组织是其准确性的关键。然而，这些方法需要比人类学习者多数量级的例子。我们使用了一个基准深度学习模型来表明层次结构也可以被用来极大地提高学习速度。我们特别展示了如何使用之前学习过的但经过广泛调整的概念表征来从两个积极的例子中学习视觉概念；重用视觉层次结构中早期的视觉表示，就像在之前的方法中一样，需要更多的示例来进行比较。这些结果提出了更有效的学习技术，并提供了一种从少数例子中学习新视觉概念的生物学上可行的方法。
Humans quickly and accurately learn new visual concepts from sparse data, sometimes just a single example. The impressive performance of artificial neural networks which hierarchically pool afferents across scales and positions suggests that the hierarchical organization of the human visual system is critical to its accuracy. These approaches, however, require magnitudes of order more examples than human learners. We used a benchmark deep learning model to show that the hierarchy can also be leveraged to vastly improve the speed of learning. We specifically show how previously learned but broadly tuned conceptual representations can be used to learn visual concepts from as few as two positive examples; reusing visual representations from earlier in the visual hierarchy, as in prior approaches, requires significantly more examples to perform comparably. These results suggest techniques for learning even more efficiently and provide a biologically plausible way to learn new visual concepts from few examples.
登录
查看更多内容
影响因子：
1.8
作者：
Brady TF;Konkle T;Alvarez GA
通讯作者：
Alvarez GA
影响因子：
16.2
作者：
Jiang, Xiong;Bradley, Evan;Riesenhuber, Maximilian
通讯作者：
Riesenhuber, Maximilian
影响因子：
19.9
作者：
Kravitz DJ;Saleem KS;Baker CI;Ungerleider LG;Mishkin M
通讯作者：
Mishkin M
影响因子：
5.3
作者：
Chan, Alexander M.;Baker, Janet M.;Halgren, Eric
通讯作者：
Halgren, Eric
影响因子：
3.7
作者：
Coutanche, Marc N.;Thompson-Schill, Sharon L.
通讯作者：
Thompson-Schill, Sharon L.