Skip to search formSkip to main contentSkip to account menu

GraphSAINT: Graph Sampling Based Inductive Learning Method

GraphSAINT is proposed, a graph sampling based inductive learning method that improves training efficiency in a fundamentally different way and can decouple the sampling process from the forward and backward propagation of training, and extend GraphSAINT with other graph samplers and GCN variants.
View PDF on arXiv (opens in a new tab)

A Framework for Generating High Throughput CNN Implementations on FPGAs

A novel Concatenate-and-Pad (CaP) technique is proposed, which improves OaA significantly by reducing the "wasted" computation on the padded pixels, and includes a tool to automatically generate fully synthesizable \textttVerilog.
View on ACM (opens in a new tab)

GraphACT: Accelerating GCN Training on CPU-FPGA Heterogeneous Platforms

A novel accelerator for training GCNs on CPU-FPGA heterogeneous systems, by incorporating multiple algorithm-architecture co-optimizations and proposing a light-weight pre-processing step based on a graph theoretic approach to optimize the feature propagation within subgraphs.
View on ACM (opens in a new tab)

Decoupling the Depth and Scope of Graph Neural Networks

This work proposes a design principle to decouple the depth and scope of GNNs -- to generate representation of a target entity, first extract a localized subgraph as the bounded-size scope, and then apply a GNN of arbitrary depth on top of the subgraph.
View PDF on arXiv (opens in a new tab)

An FPGA framework for edge-centric graph processing

A Field-Programmable Gate Array (FPGA) framework to accelerate graph algorithms based on the edge-centric paradigm, which is flexible for accelerating general graph algorithms with various vertex attributes and update propagation functions, such as Sparse Matrix Vector Multiplication and PageRank.
View on ACM (opens in a new tab)

Accurate, Efficient and Scalable Graph Embedding

This paper proposes novel parallelization techniques for graph sampling-based GCNs that achieve superior scalable performance on very large graphs without compromising accuracy, and demonstrates that the parallel graph embedding outperforms state-of theart methods in scalability, efficiency and accuracy on several large datasets.
View on IEEE (opens in a new tab)

Deep Graph Neural Networks with Shallow Subgraph Samplers

A simple "deep GNN, shallow sampler" design principle is proposed to improve both the GNN accuracy and efficiency -- to generate representation of a target node, a deep GNN is used to pass messages only within a shallow, localized subgraph.
View PDF on arXiv (opens in a new tab)

Design and implementation of parallel PageRank on multicore platforms

This paper presents an efficient parallel PageRank design based on an edge-centric scatter-gather model and develops a fast and efficient partitioning technique to overcome the poor locality of PageRank and optimize the memory performance.
View on IEEE (opens in a new tab)

Fast generation of high throughput customized deep learning accelerators on FPGAs

An automatic code generation tool that synthesizes high throughput accelerators for CNN inferencing targeting broad types of CNNs and FPGAs, and adopts an algorithm-architecture co-design methodology based on frequency domain convolution.
View on IEEE (opens in a new tab)

Quickly finding a truss in a haystack

This work presents a novel algorithm for finding both the k-truss of the graph (for a given k), as well as the maximal k- truss using a dynamic graph formulation, which is architecture independent and able to concurrently detect deleted triangles in contrast to past sequential approaches.
View on IEEE (opens in a new tab)
By clicking accept or continuing to use the site, you agree to the terms outlined in our Privacy Policy (opens in a new tab), Terms of Service (opens in a new tab), and Dataset License (opens in a new tab)