AX-DBN: An Approximate Computing Framework for the Design of Low-Power Discriminative Deep Belief Networks

Abstract

The power budget for embedded hardware implementations of Deep Learning algorithms can be extremely tight. To address implementation challenges in such domains, new design paradigms, like Approximate Computing, have drawn significant attention. Approximate Computing exploits the innate error-resilience of Deep Learning algorithms, a property that makes them amenable for deployment on low-power computing platforms. This paper describes an Approximate Computing design methodology, AX-DBN, for an architecture belonging to the class of stochastic Deep Learning algorithms known as Deep Belief Networks (DBNs). Specifically, we consider procedures for efficiently implementing the Discriminative Deep Belief Network (DDBN), a stochastic neural network which is used for classification tasks, extending Approximation Computing from the analysis of deterministic to stochastic neural networks. For the purpose of optimizing the DDBN for hardware implementations, we explore the use of: (a) Limited precision of neurons and functional approximations of activation functions; (b) Criticality analysis to identify the nodes in the network which can operate at reduced precision while allowing the network to maintain target accuracy levels; and (c) A greedy search methodology with incremental retraining to determine the optimal reduction in precision for all neurons to maximize power savings. Using the AX-DBN methodology proposed in this paper, we present experimental results across several network architectures that show significant power savings under a user-specified accuracy loss constraint with respect to ideal full precision implementations.

An RFML Ecosystem: Considerations for the Application of Deep Learning to Spectrum Situational Awareness

“…In an effort to further reduce processing requirements, some RFML implementations have also embedded traditional signal processing techniques such as Fourier and wavelet transforms, cyclostationary feature estimators, and other expert features directly into the NN [170], [174], [175]. Meanwhile, other research has focused on reduced precision implementations of NNs, enabling a path towards real-time implementation [176]- [178]. However, reducing real-time computational resources to mobile systems remains a challenge that must be overcome, especially if online learning techniques are to be developed for future RFML systems [179], [180].…”

The RFML Ecosystem: A Look at the Unique Challenges of Applying Deep Learning to Radio Frequency Applications

“…Further, some RFML implementations incorporate pre-calculated traditional signal processing techniques such as Fourier and wavelet transforms, cyclostationary feature estimators, and other expert features to serve as a more efficient feature that may be merged with machine learned behaviors [241], [245], [246]. Other research has focused on reduced precision implementations of machine learning structures as a method to gain computational efficiency [247]- [249]. However, the use of online learning techniques in RF scenarios requires real-time computational resources that are currently difficult to reduce to a mobile system [250], [251], in addition to the challenges discussed in Section III.…”