arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1906.08862cs.NEcs.LGstat.ML
收藏

神经存储程序记忆

Neural Stored-program Memory

Applied AI Institute, Deakin University(迪肯大学应用人工智能研究所)

Hung Le, Truyen Tran, Svetha Venkatesh

摘要:

由外部存储器驱动的神经网络模拟计算机行为。这些模型使用存储器为神经控制器存储数据,可以学习算法和其他复杂的任务。在本文中,我们介绍了一种新的存储器来存储控制器的权重,类似于现代计算机体系结构中的存储程序存储器。该模型被称为神经存储程序记忆,增强了当前的记忆增强神经网络,创建了可微分机器,可以通过时间切换程序,适应可变的上下文,从而类似于通用图灵机。大量的实验表明,由此产生的机器不仅在经典的算法问题中表现出色,而且在组合、连续、少样本学习和问答任务中也有潜力。

英文摘要:

Neural networks powered with external memory simulate computer behaviors. These models, which use the memory to store data for a neural controller, can learn algorithms and other complex tasks. In this paper, we introduce a new memory to store weights for the controller, analogous to the stored-program memory in modern computer architectures. The proposed model, dubbed Neural Stored-program Memory, augments current memory-augmented neural networks, creating differentiable machines that can switch programs through time, adapt to variable contexts and thus resemble the Universal Turing Machine. A wide range of experiments demonstrate that the resulting machines not only excel in classical algorithmic problems, but also have potential for compositional, continual, few-shot learning and question-answering tasks.

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。

↑