TY - JOUR
T1 - Robustness of spiking Deep Belief Networks to noise and reduced bit precision of neuro-inspired hardware platforms
AU - Stromatias, E
AU - Neil, D
AU - Pfeiffer, M
AU - Galluppi, Francesco
AU - Furber, Stephen
AU - Liu, SC
N1 - We acknowledge the following grants: SNF grant 200021_146608, SNF grant 200021_135066, EPSRC grant BIMPA (EP/G015740/1), EU grant SeeBetter (FP7-ICT-270324), EU grant HBP (FP7-604102), EU grant BrainScales-Extension (FP7-287701), and the Samsung Advanced Institute of Technology. We also acknowledge the Telluride Neuromorphic Cognition Workshop where the initial discussions and work were carried out.
PY - 2015/7/9
Y1 - 2015/7/9
N2 - Increasingly large deep learning architectures, such as Deep Belief Networks (DBNs) are the focus of current machine learning research and achieve state-of-the-art results in different domains. However, both training and execution of large-scale Deep Networks require vast computing resources, leading to high power requirements and communication overheads. The on-going work on design and construction of spike-based hardware platforms offers an alternative for running deep neural networks with significantly lower power consumption, but has to overcome hardware limitations in terms of noise and limited weight precision, as well as noise inherent in the sensor signal. This article investigates how such hardware constraints impact the performance of spiking neural network implementations of DBNs. In particular, the influence of limited bit precision during execution and training, and the impact of silicon mismatch in the synaptic weight parameters of custom hybrid VLSI implementations is studied. Furthermore, the network performance of spiking DBNs is characterized with regard to noise in the spiking input signal. Our results demonstrate that spiking DBNs can tolerate very low levels of hardware bit precision down to almost two bits, and show that their performance can be improved by at least 30% through an adapted training mechanism that takes the bit precision of the target platform into account. Spiking DBNs thus present an important use-case for large-scale hybrid analog-digital or digital neuromorphic platforms such as SpiNNaker, which can execute large but precision-constrained deep networks in real time.
AB - Increasingly large deep learning architectures, such as Deep Belief Networks (DBNs) are the focus of current machine learning research and achieve state-of-the-art results in different domains. However, both training and execution of large-scale Deep Networks require vast computing resources, leading to high power requirements and communication overheads. The on-going work on design and construction of spike-based hardware platforms offers an alternative for running deep neural networks with significantly lower power consumption, but has to overcome hardware limitations in terms of noise and limited weight precision, as well as noise inherent in the sensor signal. This article investigates how such hardware constraints impact the performance of spiking neural network implementations of DBNs. In particular, the influence of limited bit precision during execution and training, and the impact of silicon mismatch in the synaptic weight parameters of custom hybrid VLSI implementations is studied. Furthermore, the network performance of spiking DBNs is characterized with regard to noise in the spiking input signal. Our results demonstrate that spiking DBNs can tolerate very low levels of hardware bit precision down to almost two bits, and show that their performance can be improved by at least 30% through an adapted training mechanism that takes the bit precision of the target platform into account. Spiking DBNs thus present an important use-case for large-scale hybrid analog-digital or digital neuromorphic platforms such as SpiNNaker, which can execute large but precision-constrained deep networks in real time.
KW - Deep Belief Networks, spiking neural networks, SpiNNaker, noise robustness, neuro-inspired hardware
U2 - 10.3389/fnins.2015.00222
DO - 10.3389/fnins.2015.00222
M3 - Article
SN - 1662-4548
VL - 9
JO - Frontiers in Neuroscience
JF - Frontiers in Neuroscience
M1 - 00222
ER -