An FPGA Implementation of Deep Spiking Neural Networks for Low-Power and Fast Classification

2020 ◽  
Vol 32 (1) ◽  
pp. 182-204 ◽  
Author(s):  
Xiping Ju ◽  
Biao Fang ◽  
Rui Yan ◽  
Xiaoliang Xu ◽  
Huajin Tang
Keyword(s):  
Neural Networks ◽  
High Performance ◽  
Large Scale ◽  
Hardware Architecture ◽  
Clock Frequency ◽  
Data Set ◽  
Speed Up ◽  
Fast Classification ◽  
Spike Signals ◽  
Gpu Implementation

A spiking neural network (SNN) is a type of biological plausibility model that performs information processing based on spikes. Training a deep SNN effectively is challenging due to the nondifferention of spike signals. Recent advances have shown that high-performance SNNs can be obtained by converting convolutional neural networks (CNNs). However, the large-scale SNNs are poorly served by conventional architectures due to the dynamic nature of spiking neurons. In this letter, we propose a hardware architecture to enable efficient implementation of SNNs. All layers in the network are mapped on one chip so that the computation of different time steps can be done in parallel to reduce latency. We propose new spiking max-pooling method to reduce computation complexity. In addition, we apply approaches based on shift register and coarsely grained parallels to accelerate convolution operation. We also investigate the effect of different encoding methods on SNN accuracy. Finally, we validate the hardware architecture on the Xilinx Zynq ZCU102. The experimental results on the MNIST data set show that it can achieve an accuracy of 98.94% with eight-bit quantized weights. Furthermore, it achieves 164 frames per second (FPS) under 150 MHz clock frequency and obtains 41[Formula: see text] speed-up compared to CPU implementation and 22 times lower power than GPU implementation.

Analyzing and Accelerating the Bottlenecks of Training Deep SNNs With Backpropagation

2020 ◽  
Vol 32 (12) ◽  
pp. 2557-2600
Author(s):  
Ruizhi Chen ◽  
Ling Li
Keyword(s):  
Neural Networks ◽  
Training Algorithms ◽  
Minimization Method ◽  
Data Set ◽  
Ultra Low Power ◽  
Training Performance ◽  
Speed Up ◽  
Pros And Cons ◽  
Accuracy Increase ◽  
Initialization Algorithm

Spiking neural networks (SNNs) with the event-driven manner of transmitting spikes consume ultra-low power on neuromorphic chips. However, training deep SNNs is still challenging compared to convolutional neural networks (CNNs). The SNN training algorithms have not achieved the same performance as CNNs. In this letter, we aim to understand the intrinsic limitations of SNN training to design better algorithms. First, the pros and cons of typical SNN training algorithms are analyzed. Then it is found that the spatiotemporal backpropagation algorithm (STBP) has potential in training deep SNNs due to its simplicity and fast convergence. Later, the main bottlenecks of the STBP algorithm are analyzed, and three conditions for training deep SNNs with the STBP algorithm are derived. By analyzing the connection between CNNs and SNNs, we propose a weight initialization algorithm to satisfy the three conditions. Moreover, we propose an error minimization method and a modified loss function to further improve the training performance. Experimental results show that the proposed method achieves 91.53% accuracy on the CIFAR10 data set with 1% accuracy increase over the STBP algorithm and decreases the training epochs on the MNIST data set to 15 epochs (over 13 times speed-up compared to the STBP algorithm). The proposed method also decreases classification latency by over 25 times compared to the CNN-SNN conversion algorithms. In addition, the proposed method works robustly for very deep SNNs, while the STBP algorithm fails in a 19-layer SNN.

scholarly journals RHPTree—Risk Hierarchical Pattern Tree for Scalable Long Pattern Mining

2022 ◽  
Vol 16 (4) ◽  
pp. 1-33
Author(s):  
Danlu Liu ◽  
Yu Li ◽  
William Baskett ◽  
Dan Lin ◽  
Chi-Ren Shyu
Keyword(s):  
High Performance ◽  
Large Scale ◽  
Pattern Mining ◽  
Tree Structure ◽  
Research Initiative ◽  
Speed Up ◽  
Dynamic Tree ◽  
Significant Patterns ◽  
Search Approach ◽  
Hierarchical Pattern

Risk patterns are crucial in biomedical research and have served as an important factor in precision health and disease prevention. Despite recent development in parallel and high-performance computing, existing risk pattern mining methods still struggle with problems caused by large-scale datasets, such as redundant candidate generation, inability to discover long significant patterns, and prolonged post pattern filtering. In this article, we propose a novel dynamic tree structure, Risk Hierarchical Pattern Tree (RHPTree), and a top-down search method, RHPSearch, which are capable of efficiently analyzing a large volume of data and overcoming the limitations of previous works. The dynamic nature of the RHPTree avoids costly tree reconstruction for the iterative search process and dataset updates. We also introduce two specialized search methods, the extended target search (RHPSearch-TS) and the parallel search approach (RHPSearch-SD), to further speed up the retrieval of certain items of interest. Experiments on both UCI machine learning datasets and sampled datasets of the Simons Foundation Autism Research Initiative (SFARI)—Simon’s Simplex Collection (SSC) datasets demonstrate that our method is not only faster but also more effective in identifying comprehensive long risk patterns than existing works. Moreover, the proposed new tree structure is generic and applicable to other pattern mining problems.

scholarly journals The Old and the New: Can Physics-Informed Deep-Learning Replace Traditional Linear Solvers?

2021 ◽  
Vol 4 ◽  
Author(s):  
Stefano Markidis
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
High Performance ◽  
Critical Role ◽  
Low Frequency ◽  
Accurate Solution ◽  
Limiting Factor ◽  
Data Set ◽  
Linear Solvers ◽  
Computational Performance

Physics-Informed Neural Networks (PINN) are neural networks encoding the problem governing equations, such as Partial Differential Equations (PDE), as a part of the neural network. PINNs have emerged as a new essential tool to solve various challenging problems, including computing linear systems arising from PDEs, a task for which several traditional methods exist. In this work, we focus first on evaluating the potential of PINNs as linear solvers in the case of the Poisson equation, an omnipresent equation in scientific computing. We characterize PINN linear solvers in terms of accuracy and performance under different network configurations (depth, activation functions, input data set distribution). We highlight the critical role of transfer learning. Our results show that low-frequency components of the solution converge quickly as an effect of the F-principle. In contrast, an accurate solution of the high frequencies requires an exceedingly long time. To address this limitation, we propose integrating PINNs into traditional linear solvers. We show that this integration leads to the development of new solvers whose performance is on par with other high-performance solvers, such as PETSc conjugate gradient linear solvers, in terms of performance and accuracy. Overall, while the accuracy and computational performance are still a limiting factor for the direct use of PINN linear solvers, hybrid strategies combining old traditional linear solver approaches with new emerging deep-learning techniques are among the most promising methods for developing a new class of linear solvers.

scholarly journals A parallel computational framework for ultra-large-scale sequence clustering analysis

2018 ◽  
Vol 35 (3) ◽  
pp. 380-388 ◽  
Author(s):  
Wei Zheng ◽  
Qi Mao ◽  
Robert J Genco ◽  
Jean Wactawski-Wende ◽  
Michael Buck ◽  
...  
Keyword(s):  
Parallel Computing ◽  
High Performance ◽  
Large Scale ◽  
De Novo ◽  
Rapid Development ◽  
Operational Taxonomic Unit ◽  
Supplementary Information ◽  
Computational Framework ◽  
Speed Up ◽  
Scale Sequence

Abstract Motivation The rapid development of sequencing technology has led to an explosive accumulation of genomic data. Clustering is often the first step to be performed in sequence analysis. However, existing methods scale poorly with respect to the unprecedented growth of input data size. As high-performance computing systems are becoming widely accessible, it is highly desired that a clustering method can easily scale to handle large-scale sequence datasets by leveraging the power of parallel computing. Results In this paper, we introduce SLAD (Separation via Landmark-based Active Divisive clustering), a generic computational framework that can be used to parallelize various de novo operational taxonomic unit (OTU) picking methods and comes with theoretical guarantees on both accuracy and efficiency. The proposed framework was implemented on Apache Spark, which allows for easy and efficient utilization of parallel computing resources. Experiments performed on various datasets demonstrated that SLAD can significantly speed up a number of popular de novo OTU picking methods and meanwhile maintains the same level of accuracy. In particular, the experiment on the Earth Microbiome Project dataset (∼2.2B reads, 437 GB) demonstrated the excellent scalability of the proposed method. Availability and implementation Open-source software for the proposed method is freely available at https://www.acsu.buffalo.edu/~yijunsun/lab/SLAD.html. Supplementary information Supplementary data are available at Bioinformatics online.

scholarly journals Automatic Gully Detection: Neural Networks and Computer Vision

2020 ◽  
Vol 12 (11) ◽  
pp. 1743
Author(s):  
Artur M. Gafurov ◽  
Oleg P. Yermolayev
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Large Scale ◽  
Satellite Images ◽  
Gully Erosion ◽  
Training Data ◽  
Russian Plain ◽  
Data Set ◽  
High Resolution Satellite Images

Transition from manual (visual) interpretation to fully automated gully detection is an important task for quantitative assessment of modern gully erosion, especially when it comes to large mapping areas. Existing approaches to semi-automated gully detection are based on either object-oriented selection based on multispectral images or gully selection based on a probabilistic model obtained using digital elevation models (DEMs). These approaches cannot be used for the assessment of gully erosion on the territory of the European part of Russia most affected by gully erosion due to the lack of national large-scale DEM and limited resolution of open source multispectral satellite images. An approach based on the use of convolutional neural networks for automated gully detection on the RGB-synthesis of ultra-high resolution satellite images publicly available for the test region of the east of the Russian Plain with intensive basin erosion has been proposed and developed. The Keras library and U-Net architecture of convolutional neural networks were used for training. Preliminary results of application of the trained gully erosion convolutional neural network (GECNN) allow asserting that the algorithm performs well in detecting active gullies, well differentiates gullies from other linear forms of slope erosion — rills and balkas, but so far has errors in detecting complex gully systems. Also, GECNN does not identify a gully in 10% of cases and in another 10% of cases it identifies not a gully. To solve these problems, it is necessary to additionally train the neural network on the enlarged training data set.

Visualising large-scale geodynamic simulations: How to Dive into Earth's Mantle with Virtual Reality

2020 ◽  
Author(s):  
Markus Wiedemann ◽  
Bernhard S.A. Schuberth ◽  
Lorenzo Colli ◽  
Hans-Peter Bunge ◽  
Dieter Kranzlmüller
Keyword(s):  
Virtual Reality ◽  
Real Time ◽  
High Performance ◽  
Large Scale ◽  
Large Data ◽  
Optimization Techniques ◽  
Physical Parameters ◽  
Mantle Flow ◽  
Data Set ◽  
Precise Knowledge

<p>Precise knowledge of the forces acting at the base of tectonic plates is of fundamental importance, but models of mantle dynamics are still often qualitative in nature to date. One particular problem is that we cannot access the deep interior of our planet and can therefore not make direct in situ measurements of the relevant physical parameters. Fortunately, modern software and powerful high-performance computing infrastructures allow us to generate complex three-dimensional models of the time evolution of mantle flow through large-scale numerical simulations.</p><p>In this project, we aim at visualizing the resulting convective patterns that occur thousands of kilometres below our feet and to make them "accessible" using high-end virtual reality techniques.</p><p>Models with several hundred million grid cells are nowadays possible using the modern supercomputing facilities, such as those available at the Leibniz Supercomputing Centre. These models provide quantitative estimates on the inaccessible parameters, such as buoyancy and temperature, as well as predictions of the associated gravity field and seismic wavefield that can be tested against Earth observations.</p><p>3-D visualizations of the computed physical parameters allow us to inspect the models such as if one were actually travelling down into the Earth. This way, convective processes that occur thousands of kilometres below our feet are virtually accessible by combining the simulations with high-end VR techniques.</p><p>The large data set used here poses severe challenges for real time visualization, because it cannot fit into graphics memory, while requiring rendering with strict deadlines. This raises the necessity to balance the amount of displayed data versus the time needed for rendering it.</p><p>As a solution, we introduce a rendering framework and describe our workflow that allows us to visualize this geoscientific dataset. Our example exceeds 16 TByte in size, which is beyond the capabilities of most visualization tools. To display this dataset in real-time, we reduce and declutter the dataset through isosurfacing and mesh optimization techniques.</p><p>Our rendering framework relies on multithreading and data decoupling mechanisms that allow to upload data to graphics memory while maintaining high frame rates. The final visualization application can be executed in a CAVE installation as well as on head mounted displays such as the HTC Vive or Oculus Rift. The latter devices will allow for viewing our example on-site at the EGU conference.</p>

scholarly journals Fully Convolutional Neural Networks for Automotive Radar Interference Mitigation

2020 ◽  
Author(s):  
Nicolae-Catalin Ristea ◽  
Andrei Anghel ◽  
Radu Tudor Ionescu
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Large Scale ◽  
Interference Mitigation ◽  
Continuous Wave ◽  
Automotive Radar ◽  
Driver Assistance Systems ◽  
Data Set ◽  
Radar Sensors ◽  
Fully Convolutional Neural Networks

The interest of the automotive industry has progressively focused on subjects related to driver assistance systems as well as autonomous cars. In order to achieve remarkable results, cars combine a variety of sensors to perceive their surroundings robustly. Among them, radar sensors are indispensable because of their independence of light conditions and the possibility to directly measure velocity. However, radar interference is an issue that becomes prevalent with the increasing amount of radar systems in automotive scenarios. In this paper, we address this issue for frequency modulated continuous wave (FMCW) radars with fully convolutional neural networks (FCNs), a state-of-the-art deep learning technique. We propose two FCNs that take spectrograms of the beat signals as input, and provide the corresponding clean range profiles as output. We propose two architectures for interference mitigation which outperform the classical zeroing technique. Moreover, considering the lack of databases for this task, we release as open source a large scale data set that closely replicates real world automotive scenarios for single-interference cases, allowing others to compare objectively their future work in this domain. The data set is available for download at: http://github.com/ristea/arim.

scholarly journals A review of the use of convolutional neural networks in agriculture

2018 ◽  
Vol 156 (3) ◽  
pp. 312-322 ◽  
Author(s):  
A. Kamilaris ◽  
F. X. Prenafeta-Boldú
Keyword(s):  
Neural Networks ◽  
Image Processing ◽  
Convolutional Neural Networks ◽  
High Performance ◽  
Performance Metrics ◽  
Specific Class ◽  
Data Set ◽  
Advantages And Disadvantages ◽  
Future Potential ◽  
Processing Techniques

AbstractDeep learning (DL) constitutes a modern technique for image processing, with large potential. Having been successfully applied in various areas, it has recently also entered the domain of agriculture. In the current paper, a survey was conducted of research efforts that employ convolutional neural networks (CNN), which constitute a specific class of DL, applied to various agricultural and food production challenges. The paper examines agricultural problems under study, models employed, sources of data used and the overall precision achieved according to the performance metrics used by the authors. Convolutional neural networks are compared with other existing techniques, and the advantages and disadvantages of using CNN in agriculture are listed. Moreover, the future potential of this technique is discussed, together with the authors’ personal experiences after employing CNN to approximate a problem of identifying missing vegetation from a sugar cane plantation in Costa Rica. The overall findings indicate that CNN constitutes a promising technique with high performance in terms of precision and classification accuracy, outperforming existing commonly used image-processing techniques. However, the success of each CNN model is highly dependent on the quality of the data set used.

scholarly journals An effective drug-disease associations prediction model based on graphic representation learning over multi-biomolecular network

2022 ◽  
Vol 23 (1) ◽  
Author(s):  
Hanjing Jiang ◽  
Yabing Huang
Keyword(s):  
High Performance ◽  
Large Scale ◽  
Representation Learning ◽  
Biological Data ◽  
Graph Representation ◽  
Data Set ◽  
Validation Experiment ◽  
Biomolecular Network ◽  
Disease Associations ◽  
Drug Reposition

Abstract Background Drug-disease associations (DDAs) can provide important information for exploring the potential efficacy of drugs. However, up to now, there are still few DDAs verified by experiments. Previous evidence indicates that the combination of information would be conducive to the discovery of new DDAs. How to integrate different biological data sources and identify the most effective drugs for a certain disease based on drug-disease coupled mechanisms is still a challenging problem. Results In this paper, we proposed a novel computation model for DDA predictions based on graph representation learning over multi-biomolecular network (GRLMN). More specifically, we firstly constructed a large-scale molecular association network (MAN) by integrating the associations among drugs, diseases, proteins, miRNAs, and lncRNAs. Then, a graph embedding model was used to learn vector representations for all drugs and diseases in MAN. Finally, the combined features were fed to a random forest (RF) model to predict new DDAs. The proposed model was evaluated on the SCMFDD-S data set using five-fold cross-validation. Experiment results showed that GRLMN model was very accurate with the area under the ROC curve (AUC) of 87.9%, which outperformed all previous works in terms of both accuracy and AUC in benchmark dataset. To further verify the high performance of GRLMN, we carried out two case studies for two common diseases. As a result, in the ranking of drugs that were predicted to be related to certain diseases (such as kidney disease and fever), 15 of the top 20 drugs have been experimentally confirmed. Conclusions The experimental results show that our model has good performance in the prediction of DDA. GRLMN is an effective prioritization tool for screening the reliable DDAs for follow-up studies concerning their participation in drug reposition.

Sign in / Sign up

Export Citation Format

Share Document