Search NASA⌕ Search

SEARCH · Search NASA

Results for “Convolution”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 703 records · Page 39

Evolving Energy Efficient Convolutional Neural Networks

As deep neural networks have been deployed in more and more applications over the past half decade and are finding their way into an ever increasing number of operational systems, their energy consumption becomes a concern whether running in the datacenter or on edge devices. Hyperparameter optimization and automated network design for deep learning is a quickly growing field, but much of the focus has remained only on optimizing for the performance of the machine learning task. In this work, we demonstrate that the best performing networks created through this automated network design process have radically different computational characteristics (e.g. energy usage, model size, inference time), presenting the opportunity to utilize this optimization process to make deep learning networks more energy efficient and deployable to smaller devices. Optimizing for these computational characteristics is critical as the number of applications of deep learning continues to expand.

Young, Steven↗

H-GCN: A Graph Convolutional Network Accelerator on Versal ACAP Architecture

Recently Graph Neural Networks (GNNs) have drawn tremendous attentions due to their unique capability to extend the Machine Learning (ML) approaches to broadly defined applications with unstructured data, especially graphs. Comparing with other ML modalities, the acceleration of GNNs is as critical but even more challenging due to the irregularity and heterogeneity from graph typologies that together limit the performance. Existing efforts mainly focus on handling graphs’ irregularity, however, have not studied the heterogeneity. To this end, in this work, we propose H-GCN, a PL-AIE-based hybrid accelerator that leverages the emerging heterogeneity of Xilinx Versal ACAPs to achieve high-performance GNN inference. In particular, H-GCN partitions each graph into three subgraphs based on its inherent heterogeneity and processes them using PL and the newly emerged AIE respectively. To further improve the performance, we explore the sparsity support of AIE and develop an efficient density-aware method to map tiles of SpMM onto the systolic tensor array automatically. Compared with the current state-of-the-art GCN accelerator, HGCN achieves on average 1.5× speedups.

Zhang, Chengming↗

Hashed Binary Search Sampling for Convolutional Network Training with Large Overhead Image Patches

Very large overhead imagery associated with ground truth maps has the potential to generate billions of training image patches for machine learning algorithms. However, random sampling selection criteria often leads to redundant and noisy-image patches for model training. With minimal research efforts behind this challenge, the current status spells missed opportunities to develop supervised learning algorithms that generalize over wide geographical scenes. In addition, much of the computational cycles for large scale machine learning are poorly spent crunching through noisy and redundant image patches. We demonstrate a potential framework to address these challenges specifically, while evaluating a human settlement detection task. A novel binary search tree sampling scheme is fused with a kernel based hashing procedure that maps image patches into hash-buckets using binary codes generated from image content. The framework exploits inherent redundancy within billions of image patches to promote mostly high variance preserving samples for accelerating algorithmic training and increasing model generalization.

Lunga, Dalton↗

Convolutional Neural Network Surrogate Models for the Mechanical Properties of Periodic Structures

The work in this paper describes neural network surrogate models for calculating the effective mechanical properties of a periodic composites. The models achieve good accuracy even when only provided with training data sampling a small portion of the design space. As an example, the surrogate models are applied to solving the inverse design problem of finding structures with optimal mechanical properties. The surrogate models are sufficiently accurate to recover optimal solutions in general agreement with established topology optimization methods. However, improvements will be required to develop robust, efficient neural network-based surrogate models and several directions for future research are highlighted here.

42 ENGINEERING↗

An energy fluence-convolution model for amorphous silicon EPID dose prediction

In this work, an amorphous silicon electronic portal imaging device (a-Si EPID) dose prediction model based on the energy fluence model of the Pinnacle treatment planning system Version 7 (Philips Medical Systems, Madison, WI) is developed. An energy fluence matrix at very high resolution (<1 mm) is used to incorporate multileaf collimator (MLC) leaf effects in the predicted EPID images. The primary dose deposited in the EPID is calculated from the energy fluence using experimentally derived radially dependent EPID interaction coefficients. Separate coefficients are used for the open beam energy fluence component and the component of the energy fluence transmitted through closed MLC leaves to each EPID pixel. A spatially invariant EPID dose deposition kernel that describes both radiative dose deposition, central axis EPID backscatter, and optical glare is convolved with the primary dose. The kernel is further optimized to give accurate EPID penumbra prediction and EPID scatter factor with changing MLC field size. An EPID calibration method was developed to reduce the effect of nonuniform backscatter from the support arm (E-arm) in a calibrated EPID image. This method removes the backscatter component from the pixel sensitivity (flood field) correction matrix retaining only field-specific backscatter in the images. The model was compared to EPID images for jaw and MLC defined open fields and eight head and neck intensity modulated radiotherapy (IMRT) fields. For the head and neck IMRT fields with 2%, 2 mm criteria 97.6{+-}0.6% (mean {+-}1 standard deviation) of points passed with a gamma index less than 1, and for 3%, 3 mm 99.4{+-}0.4% of points were within the criteria. For these fields, the 2%, 2 mm pass score reduced to 96.0{+-}1.5% when backscatter was present in the pixel sensitivity correction matrix. The model incorporates the effect of MLC leaf transmission, EPID response to open and MLC leakage dose components, and accurately predicts EPID images of IMRT fields. Removing the backscatter component of the pixel sensitivity matrix correction reduces the effect of nonuniform E-arm backscatter.

61 RADIATION PROTECTION AND DOSIMETRY↗

A convolution-superposition dose calculation engine for GPUs

Purpose: Graphic processing units (GPUs) are increasingly used for scientific applications, where their parallel architecture and unprecedented computing power density can be exploited to accelerate calculations. In this paper, a new GPU implementation of a convolution/superposition (CS) algorithm is presented. Methods: This new GPU implementation has been designed from the ground-up to use the graphics card's strengths and to avoid its weaknesses. The CS GPU algorithm takes into account beam hardening, off-axis softening, kernel tilting, and relies heavily on raytracing through patient imaging data. Implementation details are reported as well as a multi-GPU solution. Results: An overall single-GPU acceleration factor of 908x was achieved when compared to a nonoptimized version of the CS algorithm implemented in PlanUNC in single threaded central processing unit (CPU) mode, resulting in approximatively 2.8 s per beam for a 3D dose computation on a 0.4 cm grid. A comparison to an established commercial system leads to an acceleration factor of approximately 29x or 0.58 versus 16.6 s per beam in single threaded mode. An acceleration factor of 46x has been obtained for the total energy released per mass (TERMA) calculation and a 943x acceleration factor for the CS calculation compared to PlanUNC. Dose distributions also have been obtained for a simple water-lung phantom to verify that the implementation gives accurate results. Conclusions: These results suggest that GPUs are an attractive solution for radiation therapy applications and that careful design, taking the GPU architecture into account, is critical in obtaining significant acceleration factors. These results potentially can have a significant impact on complex dose delivery techniques requiring intensive dose calculations such as intensity-modulated radiation therapy (IMRT) and arc therapy. They also are relevant for adaptive radiation therapy where dose results must be obtained rapidly.

61 RADIATION PROTECTION AND DOSIMETRY↗

I-GCN: A Graph Convolutional Network Accelerator with Runtime Locality Enhancement through Islandization

In this paper, we propose a novel hardware accelerator for GCN inference called I-GCN that significantly improves data locality and reduces unnecessary computation through a new online graph restructuring algorithm we refer to as islandization. The proposed algorithm finds clusters of nodes with strong internal but weak external connections. The islandization process yields two major benefits. First, by processing islands rather than individual nodes, there is better on-chip data reuse and fewer off-chip memory accesses. Second, there is less redundant computation as aggregation for common/shared neighbors in an island can be reused. The parallel search, identification, and leverage of graph islands are all handled purely in hardware at runtime working in an incremental pipelined manner. This is done without any preprocessing of the graph data or adjustment of the GCN model structure.

Geng, Tong↗