Binarized Convolutional Neural Networks with Separable Filters for Efficient Hardware Acceleration

Lin, Jeng-Hau; Xing, Tianwei; Zhao, Ritchie; Zhang, Zhiru; Srivastava, Mani; Tu, Zhuowen; Gupta, Rajesh K.

doi:10.1109/cvprw.2017.48

Cited by 19 publications

(8 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For MNIST, the CNN(lite) model contains only one convolutional layer with 40 kernels. The CNN(lite) model for SVHN has 2 convolutional layers (8)(9)(10)(11)(12)(13)(14)(15)(16)(17). The CNN(lite) model for CIFAR-10 also has 2 convolutional layers (8)(9).…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Local Binary Pattern Networks

Lin¹,

Yang²,

Gupta³

et al. 2018

Preprint

Self Cite

View full text Add to dashboard Cite

Memory and computation efficient deep learning architectures are crucial to continued proliferation of machine learning capabilities to new platforms and systems. Binarization of operations in convolutional neural networks has shown promising results in reducing model size and computing efficiency. In this paper, we tackle the problem using a strategy different from the existing literature by proposing local binary pattern networks or LBPNet, that is able to learn and perform binary operations in an end-to-end fashion. LBPNet 1 uses local binary comparisons and random projection in place of conventional convolution (or approximation of convolution) operations. These operations can be implemented efficiently on different platforms including direct hardware implementation. We applied LBPNet and its variants on standard benchmarks. The results are promising across benchmarks while providing an important means to improve memory and speed efficiency that is particularly suited for small footprint devices and hardware accelerators.

show abstract

Section: Resultsmentioning

confidence: 99%

“…In the BNN [10] paper, the classification on MNIST is done with a binarized multilayer perceptron network (MLP). We adopt the binarized convolutional neural network (BCNN) in [10] for SVHN to perform the classification and reproduce the same accuracy as shown in [17] on MNIST.…”

Section: Resultsmentioning

confidence: 99%

Local Binary Pattern Networks

Lin¹,

Yang²,

Gupta³

et al. 2018

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Hubara et al [59] proposed a method to train the Quantized Neural Network (QNN). Lin et al [67] proposed a Binarized CNN (BCNN) with a separable filter in binary quantization network, which applies Singular Value Decomposition (SVD) on BCNN kernels. Unlike simple matrix approximation, Hou et al [68] proposed a proximal Newton algorithm with diagonal Hessian approximation that directly minimizes the loss w.r.t.…”

Section: Referencesmentioning

confidence: 99%

Compacting Deep Neural Networks for Internet of Things: Methods and Applications

Zhang

Ying

Dai

et al. 2021

IEEE Internet Things J.

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) have shown great success in completing complex tasks. However, DNNs inevitably bring high computational cost and storage consumption due to the complexity of hierarchical structures, thereby hindering their wide deployment in Internet-of-Things (IoT) devices, which have limited computational capability and storage capacity. Therefore, it is a necessity to investigate the technologies to compact DNNs. Despite tremendous advances in compacting DNNs, few surveys summarize compacting-DNNs technologies, especially for IoT applications. Hence, this paper presents a comprehensive study on compacting-DNNs technologies. We categorize compacting-DNNs technologies into three major types: 1) network model compression, 2) Knowledge Distillation (KD), 3) modification of network structures. We also elaborate on the diversity of these approaches and make side-by-side comparisons. Moreover, we discuss the applications of compacted DNNs in various IoT applications and outline future directions.

show abstract

“…Thus, we would like to make a distinction between explicitly Ternary Networks, using 2-bit representation, and Binary-Sparse Networks, leveraging XNOR multiplication and size compression, like ours. As an example of the latter, Lin et al [53], exploited Singular Value Decomposition to reduce kernel sizes in Binarized Neural Networks (BNNs). Certain software and hardware implementations rely on operand-gating XNOR multiplication [21,31]; however, they still require 2-bits of information per weight: value and mask.…”

Section: Sparse Binary Networkmentioning

confidence: 99%

3PXNet

Romaszkan

Gupta

2020

ACM Trans. Embed. Comput. Syst.

View full text Add to dashboard Cite

As the adoption of Neural Networks continues to proliferate different classes of applications and systems, edge devices have been left behind. Their strict energy and storage limitations make them unable to cope with the sizes of common network models. While many compression methods such as precision reduction and sparsity have been proposed to alleviate this, they don't go quite far enough. To push size reduction to its absolute limits, we combine binarization with sparsity in Pruned-Permuted-Packed XNOR Networks (3PXNet), which can be efficiently implemented on even the smallest of embedded microcontrollers. 3PXNets can reduce model sizes by up to 38X and reduce runtime by up to 3X compared with already compact conventional binarized implementations with less than 3% accuracy reduction. We have created the first software implementation of sparse-binarized Neural Networks, released as open source library targeting edge devices. Our library is complete with training methodology and model generating scripts, making it easy and fast to deploy.

show abstract

Binarized Convolutional Neural Networks with Separable Filters for Efficient Hardware Acceleration

Cited by 19 publications

References 23 publications

Local Binary Pattern Networks

Local Binary Pattern Networks

Compacting Deep Neural Networks for Internet of Things: Methods and Applications

3PXNet

Contact Info

Product

Resources

About