arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2006.12433cs.LGstat.ML
收藏

什么形状特征表示?探索数据集、架构和培训

What shapes feature representations? Exploring datasets, architectures, and training

Stanford University(斯坦福大学)

Katherine L. Hermann, Andrew K. Lampinen

摘要:

在自然主义学习问题中,模型的输入包含各种各样的特征,有些对手头的任务有用,有些则没有。在这些有用的特性中,模型使用了哪些?在与任务无关的特征中,模型代表了哪些特征?这些问题的答案对于理解模型决策的基础,以及构建能够学习超出原始训练任务的通用、适应性强的表示的模型非常重要。我们使用合成数据集研究这些问题,在合成数据集中,输入特征的任务相关性可以直接控制。我们发现,当两个特征冗余地预测标签时,模型优先表示其中一个,并且它的偏好反映了从未训练过的模型中最线性解码的内容。在训练过程中,任务相关的特征被增强,而任务无关的特征被部分抑制。有趣的是,在某些情况下,一个更容易、预测性较弱的特征可以抑制一个预测性更强、但更困难的特征。此外,经过训练以识别简单和困难特征的模型学习与仅使用简单特征的模型最相似的表示。此外,容易的功能导致更一致的表示跨模型运行比硬功能。最后,模型与未经训练的模型比在不同任务上训练的模型具有更大的代表性相似性。我们的研究结果突出了决定模型代表哪些特征的复杂过程。

英文摘要:

In naturalistic learning problems, a model's input contains a wide range of features, some useful for the task at hand, and others not. Of the useful features, which ones does the model use? Of the task-irrelevant features, which ones does the model represent? Answers to these questions are important for understanding the basis of models' decisions, as well as for building models that learn versatile, adaptable representations useful beyond the original training task. We study these questions using synthetic datasets in which the task-relevance of input features can be controlled directly. We find that when two features redundantly predict the labels, the model preferentially represents one, and its preference reflects what was most linearly decodable from the untrained model. Over training, task-relevant features are enhanced, and task-irrelevant features are partially suppressed. Interestingly, in some cases, an easier, weakly predictive feature can suppress a more strongly predictive, but more difficult one. Additionally, models trained to recognize both easy and hard features learn representations most similar to models that use only the easy feature. Further, easy features lead to more consistent representations across model runs than do hard features. Finally, models have greater representational similarity to an untrained model than to models trained on a different task. Our results highlight the complex processes that determine which features a model represents.

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。

↑