用磁阻RAM实现二值化神经网络
Implementing Binarized Neural Networks with Magnetoresistive RAM without Error Correction
摘要:
自旋力矩磁阻随机存取存储器(ST-MRAM)的最令人兴奋的应用之一是深度神经网络的存储器内实现,其可以允许将人工智能的能量效率提高几个数量级,相对于其在计算机和图形卡上的实现。特别是,ST-MRAM可能是实现二进制神经网络(BNN)的理想选择,BNN是2016年发现的一种深度神经网络,它可以实现最先进的性能,同时大大减少了传统人工智能方法的内存占用。然而,ST-MRAM的挑战在于它容易出现写入错误,并且通常需要使用纠错。在这项工作中,我们表明,这些位错误可以容忍BNN到一个突出的水平,基于图像识别任务的例子(MNIST,CIFAR-10和ImageNet):位出错率的ST-MRAM高达0.1%的识别精度几乎没有影响。因此,相对于传统应用,BNN对ST-MRAM的要求相当宽松。因此,我们表明,对于BNN,ST-MRAM可以编程弱(低能量)的编程条件,没有纠错码。我们表明,这一结果可以允许使用低能量和低面积的ST-MRAM单元,并表明,在系统级的能量节省可以达到一个因素2。
英文摘要:
One of the most exciting applications of Spin Torque Magnetoresistive Random Access Memory (ST-MRAM) is the in-memory implementation of deep neural networks, which could allow improving the energy efficiency of Artificial Intelligence by orders of magnitude with regards to its implementation on computers and graphics cards. In particular, ST-MRAM could be ideal for implementing Binarized Neural Networks (BNNs), a type of deep neural networks discovered in 2016, which can achieve state-of-the-art performance with a highly reduced memory footprint with regards to conventional artificial intelligence approaches. The challenge of ST-MRAM, however, is that it is prone to write errors and usually requires the use of error correction. In this work, we show that these bit errors can be tolerated by BNNs to an outstanding level, based on examples of image recognition tasks (MNIST, CIFAR-10 and ImageNet): bit error rates of ST-MRAM up to 0.1% have little impact on recognition accuracy. The requirements for ST-MRAM are therefore considerably relaxed for BNNs with regards to traditional applications. By consequence, we show that for BNNs, ST-MRAMs can be programmed with weak (low-energy) programming conditions, without error correcting codes. We show that this result can allow the use of low energy and low area ST-MRAM cells, and show that the energy savings at the system level can reach a factor two.
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。