BitStream

Zhao, Tianli; He, Xiaoliang; Cheng, Jian; Hu, Jing

doi:10.1145/3240508.3240673

Cited by 7 publications

(5 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experimental results in Figure 2 illustrate that our upgraded bit-packing scheme is ∼4× as fast as the naive way. Note that our method is also compatible with fused-BN-binarization layer in [15].…”

Section: Dabnnmentioning

confidence: 97%

“…As such, xnor can be directly performed between these binarized vectors. Previous works [7][14] [15] often perform bit-packing in a naive way. For example, [7] compares every 32-bit element with zero and sets the corresponding bit sequentially.…”

Section: Dabnnmentioning

confidence: 99%

“…In the literature, there are several inference frameworks for BNNs: BitStream [15], BitFlow [7], and BMXNet [14]. Among them, BitStream and BitFlow are not open-source and not available to public.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

daBNN

Zhang

Pan

Yao

et al. 2019

Proceedings of the 27th ACM International Conference on Multimedia

View full text Add to dashboard Cite

It is always well believed that Binary Neural Networks (BNNs) could drastically accelerate the inference efficiency by replacing the arithmetic operations in float-valued Deep Neural Networks (DNNs) with bit-wise operations. Nevertheless, there has not been open-source implementation in support of this idea on low-end ARM devices (e.g., mobile phones and embedded devices). In this work, we propose daBNN -a super fast inference framework that implements BNNs on ARM devices. Several speed-up and memory refinement strategies for bit-packing, binarized convolution, and memory layout are uniquely devised to enhance inference efficiency. Compared to the recent open-source BNN inference framework, BMXNet, our daBNN is 7×∼23× faster on a single binary convolution, and about 6× faster on Bi-Real Net 18 (a BNN variant of ResNet-18). The daBNN is a BSD-licensed inference framework, and its source code, sample projects and pre-trained models are available on-line: https://github.com/JDAI-CV/dabnn. CCS CONCEPTS• Software and its engineering → Software libraries and repositories; • Computer systems organization → Neural networks. KEYWORDS

show abstract

Section: Dabnnmentioning

confidence: 97%

Section: Dabnnmentioning

confidence: 99%

See 1 more Smart Citation

daBNN

Zhang

Pan

Yao

et al. 2019

Proceedings of the 27th ACM International Conference on Multimedia

View full text Add to dashboard Cite

show abstract

“…There are some inference frameworks for BNN, some of them are open-source software frameworks like BMXNet [222], BMXNet 2 [223], daBNN [224], Riptide [225], and Larq [226], and the others are not available to the public like BitStream [227]. This section discusses only the open-source software frameworks.…”

Section: ) Other Frameworkmentioning

confidence: 99%

A Systematic Literature Review on Binary Neural Networks

et al. 2023

View full text Add to dashboard Cite

This paper presents an extensive literature review on Binary Neural Network (BNN). BNN utilizes binary weights and activation function parameters to substitute the full-precision values. In digital implementations, BNN replaces the complex calculations of Convolutional Neural Networks (CNNs) with simple bitwise operations. BNN optimizes large computation and memory storage requirements, which leads to less area and power consumption compared to full-precision models. Although there are many advantages of BNN, the binarization process has a significant impact on the performance and accuracy of the generated models. To reflect the state-of-the-art in BNN and explore how to develop and improve BNNbased models, we conduct a systematic literature review on BNN with data extracted from 239 research studies. Our review discusses various BNN architectures and the optimization approaches developed to improve their performance. There are three main research directions in BNN: accuracy optimization, compression optimization, and acceleration optimization. The accuracy optimization approaches include quantization error reduction, special regularization, gradient error minimization, and network structure. The compression optimization approaches combine fractional BNN and pruning. The acceleration optimization approaches comprise computing in-memory, FPGA-based implementations, and ASIC-based implementations. At the end of our review, we present a comprehensive analysis of BNN applications and their evaluation metrics. Also, we shed some light on the most common BNN challenges and the future research trends of BNN.

show abstract

“…Owing to the sheer scale and complexity of convolutional neural networks (CNNs), lightweight methods (quantization [27]- [30], pruning [31], [32], and low-rank approximation [33]) are required. Xiao et al [34] proposed layer freezing training based on the gradient magnitude, which aims to shorten the training time of CNNs.…”

Section: Weight Transfer and Freezingmentioning

confidence: 99%

SFAO: Sign-Flipping-Aware Optimization for Early-Stopping of Binarized Neural Networks

Kang,

Ryu,

Kang

et al. 2023

IEEE Access

View full text Add to dashboard Cite

One of the vital challenges for the binary neural networks (BNNs) is improving their inference performance by expanding their data representation capabilities for figuring out delicate patterns and nuances in the data. Addressing the explosive computational demands on neural network training is essential to guarantee sustainable development and scalable deployment. However, mitigating the increase in the computational cost during the training phase is critical for ensuring sustainability and scalability during deployment. In this study, an advanced sign-flipping-aware optimizer (SFAO) that focuses on BNNs was introduced to diminish the computational burden. SFAO balanced the model performance and computational cost through sign-flipping-aware updating rules throughout the training of BNNs. SFAO optimizer, tailored for BNNs with binary weight-specific updating rules, considerably reduced the computing resources needed for training on the CIFAR-10 dataset. Specifically, it surpassed the conventional full-precision updating rule by reducing the total instruction count by 21.89%. In contrast, SFAO showed a marginal 0.44% decline in the image classification accuracy relative to the updating rules for the full-precision parameters. Furthermore, the implementation of early stopping using the sign flip rate led to a notable reduction of 9.37% in the average computation time per network for the ImageNet dataset.INDEX TERMS artificial intelligence, model compression, optimizer, efficient machine learning, binarized neural networks, layer freezing.

show abstract

BitStream

Cited by 7 publications

References 5 publications

daBNN

daBNN

A Systematic Literature Review on Binary Neural Networks

SFAO: Sign-Flipping-Aware Optimization for Early-Stopping of Binarized Neural Networks

Contact Info

Product

Resources

About