Limits...
Robustness of spiking Deep Belief Networks to noise and reduced bit precision of neuro-inspired hardware platforms.

Stromatias E, Neil D, Pfeiffer M, Galluppi F, Furber SB, Liu SC - Front Neurosci (2015)

Bottom Line: 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.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.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.

View Article: PubMed Central - PubMed

Affiliation: Advanced Processor Technologies Group, School of Computer Science, University of Manchester Manchester, UK.

ABSTRACT
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.

No MeSH data available.


Impact of different rounding methods during learning on learned weight representations. Comparison of first-layer weights in networks trained with the dual-copy rounding method (left) and the post-learning rounding method (right). The weights shown here are representative samples from 16 clusters of weight vectors in the learned dual-copy rounding weight matrix. On the right, the weights from the post-learning rounding weight matrix that are most similar to these chosen weights are displayed. The dual-copy rounding method is able to preserve much more fine structure, compared to simply rounding the network weights after training, and is thus more suitable for training networks that will be executed with lower bit precision weights.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4496577&req=5

Figure 8: Impact of different rounding methods during learning on learned weight representations. Comparison of first-layer weights in networks trained with the dual-copy rounding method (left) and the post-learning rounding method (right). The weights shown here are representative samples from 16 clusters of weight vectors in the learned dual-copy rounding weight matrix. On the right, the weights from the post-learning rounding weight matrix that are most similar to these chosen weights are displayed. The dual-copy rounding method is able to preserve much more fine structure, compared to simply rounding the network weights after training, and is thus more suitable for training networks that will be executed with lower bit precision weights.

Mentions: For qualitative differences, observe the weights shown in Figure 8. In order to show representative samples, the learned weights in the first layer from the dual-copy rounding method were clustered into 16 categories, and the post-learning rounding method weights with the closest Euclidean distance to these cluster exemplars were identified and plotted on the right. The dual-copy rounding method preserves significantly more fine-grained structure, which would be lost with other rounding methods.


Robustness of spiking Deep Belief Networks to noise and reduced bit precision of neuro-inspired hardware platforms.

Stromatias E, Neil D, Pfeiffer M, Galluppi F, Furber SB, Liu SC - Front Neurosci (2015)

Impact of different rounding methods during learning on learned weight representations. Comparison of first-layer weights in networks trained with the dual-copy rounding method (left) and the post-learning rounding method (right). The weights shown here are representative samples from 16 clusters of weight vectors in the learned dual-copy rounding weight matrix. On the right, the weights from the post-learning rounding weight matrix that are most similar to these chosen weights are displayed. The dual-copy rounding method is able to preserve much more fine structure, compared to simply rounding the network weights after training, and is thus more suitable for training networks that will be executed with lower bit precision weights.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4496577&req=5

Figure 8: Impact of different rounding methods during learning on learned weight representations. Comparison of first-layer weights in networks trained with the dual-copy rounding method (left) and the post-learning rounding method (right). The weights shown here are representative samples from 16 clusters of weight vectors in the learned dual-copy rounding weight matrix. On the right, the weights from the post-learning rounding weight matrix that are most similar to these chosen weights are displayed. The dual-copy rounding method is able to preserve much more fine structure, compared to simply rounding the network weights after training, and is thus more suitable for training networks that will be executed with lower bit precision weights.
Mentions: For qualitative differences, observe the weights shown in Figure 8. In order to show representative samples, the learned weights in the first layer from the dual-copy rounding method were clustered into 16 categories, and the post-learning rounding method weights with the closest Euclidean distance to these cluster exemplars were identified and plotted on the right. The dual-copy rounding method preserves significantly more fine-grained structure, which would be lost with other rounding methods.

Bottom Line: 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.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.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.

View Article: PubMed Central - PubMed

Affiliation: Advanced Processor Technologies Group, School of Computer Science, University of Manchester Manchester, UK.

ABSTRACT
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.

No MeSH data available.