Results 21 to 30 of about 1,721,209 (309)
Forward and Backward Information Retention for Accurate Binary Neural Networks [PDF]
Computer Vision and Pattern Recognition, 2019 Weight and activation binarization is an effective approach to deep neural network compression and can accelerate the inference by leveraging bitwise operations. Although many binarization methods have improved the accuracy of the model by minimizing the
Haotong Qin +6 more
semanticscholar +1 more source
Optimization design of binary VGG convolutional neural network accelerator
Dianzi Jishu Yingyong, 2021 Most of the existing researches on accelerators of binary convolutional neural networks based on FPGA are aimed at small-scale image input, while the applications mainly take large-scale convolutional neural networks such as YOLO and VGG as backbone ...
Zhang Xuxin +3 more
doaj +1 more source
Communication-Efficient Federated Learning with Binary Neural Networks [PDF]
IEEE Journal on Selected Areas in Communications, 2021 Federated learning (FL) is a privacy-preserving machine learning setting that enables many devices to jointly train a shared global model without the need to reveal their data to a central server.
Yuzhi Yang +2 more
semanticscholar +1 more source
FracBNN: Accurate and FPGA-Efficient Binary Neural Networks with Fractional Activations [PDF]
Symposium on Field Programmable Gate Arrays, 2020 Binary neural networks (BNNs) have 1-bit weights and activations. Such networks are well suited for FPGAs, as their dominant computations are bitwise arithmetic and the memory requirement is also significantly reduced.
Yichi Zhang +5 more
semanticscholar +1 more source
Exploring the Connection Between Binary and Spiking Neural Networks
Frontiers in Neuroscience, 2020 On-chip edge intelligence has necessitated the exploration of algorithmic techniques to reduce the compute requirements of current machine learning frameworks.
Sen Lu, Abhronil Sengupta
doaj +1 more source
STT-BNN: A Novel STT-MRAM In-Memory Computing Macro for Binary Neural Networks
IEEE Journal on Emerging and Selected Topics in Circuits and Systems, 2022 This paper presents a novel architecture for in-memory computation of binary neural network (BNN) workloads based on STT-MRAM arrays. In the proposed architecture, BNN inputs are fed through bitlines, then, a BNN vector multiplication can be done by ...
Thi-Nhan Pham +3 more
semanticscholar +1 more source
AdaSTE: An Adaptive Straight-Through Estimator to Train Binary Neural Networks [PDF]
Computer Vision and Pattern Recognition, 2021 We propose a new algorithm for training deep neural networks (DNNs) with binary weights. In particular, we first cast the problem of training binary neural networks (BiNNs) as a bilevel optimization instance and subsequently construct flexible ...
Huu Le +3 more
semanticscholar +1 more source
Sub-bit Neural Networks: Learning to Compress and Accelerate Binary Neural Networks [PDF]
IEEE International Conference on Computer Vision, 2021 In the low-bit quantization field, training Binarized Neural Networks (BNNs) is the extreme solution to ease the deployment of deep models on resource-constrained devices, having the lowest storage cost and significantly cheaper bit-wise operations ...
Yikai Wang +3 more
semanticscholar +1 more source
An Information-Reserved and Deviation-Controllable Binary Neural Network for Object Detection
Mathematics, 2022 Object detection is a fundamental task in computer vision, which is usually based on convolutional neural networks (CNNs). While it is difficult to be deployed in embedded devices due to the huge storage and computing consumptions, binary neural networks
Ganlin Zhu +4 more
doaj +1 more source
Quantum Annealing Formulation for Binary Neural Networks [PDF]
International Conference on Digital Image Computing: Techniques and Applications, 2021 Quantum annealing is a promising paradigm for building practical quantum computers. Compared to other approaches, quantum annealing technology has been scaled up to a larger number of qubits.
M. Sasdelli, Tat-Jun Chin
semanticscholar +1 more source