32,613 Hits in 3.1 sec

Edge-featured Graph Neural Architecture Search [article]

Shaofei Cai, Liang Li, Xinzhe Han, Zheng-jun Zha, Qingming Huang
2021 arXiv   pre-print
To solve this problem, we incorporate edge features into graph search space and propose Edge-featured Graph Neural Architecture Search to find the optimal GNN architecture.  ...  Recently, researchers study neural architecture search (NAS) to reduce the dependence of human expertise and explore better GNN architectures, but they over-emphasize entity features and ignore latent  ...  We randomly sample a graph neural architecture from edge-featured graph search space, termed as "Random".  ... 
arXiv:2109.01356v1 fatcat:wp4tyza3xfh27edowubhnmedlq

Network Graph Based Neural Architecture Search [article]

Zhenhan Huang, Chunheng Jiang, Pin-Yu Chen, Jianxi Gao
2021 arXiv   pre-print
Here we propose a new way of searching neural network where we search neural architecture by rewiring the corresponding graph and predict the architecture performance by graph properties.  ...  Our work proposes a new way of searching neural architecture and provides insights on neural architecture design.  ...  By targeting the optimal graph structure, we will able to locate the optimal neural network architecture. Figure 1 shows the graph-based neural architecture search.  ... 
arXiv:2112.07805v1 fatcat:rag2gwxbovenfntrzvvytxjxoe

Deep Neural Architecture Search with Deep Graph Bayesian Optimization [article]

Lizheng Ma, Jiaxu Cui, Bo Yang
2019 arXiv   pre-print
In this work, we propose a Bayesian graph neural network as a new surrogate, which can automatically extract features from deep neural architectures, and use such learned features to fit and characterize  ...  Based on the new surrogate, we then develop a graph Bayesian optimization framework to address the challenging task of deep neural architecture search.  ...  space is modeled as an attributed graph q = {V, E, F V , F G }, where V is a set of nodes denoting the layers of neural architecture, E is a set of edges, F V is the feature set of nodes and F G is the  ... 
arXiv:1905.06159v1 fatcat:fb7capt2mrbaflf3ze2tqampwi

Rethinking Graph Neural Architecture Search from Message-passing [article]

Shaofei Cai, Liang Li, Jincan Deng, Beichen Zhang, Zheng-Jun Zha, Li Su, Qingming Huang
2021 arXiv   pre-print
Inspired by the strong searching capability of neural architecture search (NAS) in CNN, this paper proposes Graph Neural Architecture Search (GNAS) with novel-designed search space.  ...  Specifically, we design Graph Neural Architecture Paradigm (GAP) with tree-topology computation procedure and two types of fine-grained atomic operations (feature filtering and neighbor aggregation) from  ...  Graph Neural Architecture Search Following graph neural architecture paradigm (GAP), we design a three-level search space.  ... 
arXiv:2103.14282v4 fatcat:rshdr2tqvbbm7ijborgvtjju7a

Automatic Relation-aware Graph Network Proliferation [article]

Shaofei Cai, Liang Li, Xinzhe Han, Jiebo Luo, Zheng-Jun Zha, Qingming Huang
2022 arXiv   pre-print
Graph neural architecture search has sparked much attention as Graph Neural Networks (GNNs) have shown powerful reasoning capability in many relational tasks.  ...  However, the currently used graph search space overemphasizes learning node features and neglects mining hierarchical relational information.  ...  6 i Page 1 Preliminaries A graph neural architecture uses graph-structured data as input and outputs high-dimensional node and edge features (relation features).  ... 
arXiv:2205.15678v1 fatcat:4hrh3t2akncdhbqpt5mickgmwy

Dynamic Graph: Learning Instance-aware Connectivity for Neural Networks [article]

Kun Yuan, Quanquan Li, Dapeng Chen, Aojun Zhou, Junjie Yan
2020 arXiv   pre-print
One practice of employing deep neural networks is to apply the same architecture to all the input instances.  ...  We generate edge weights by a learnable module router and select the edges whose weights are larger than a threshold, to adjust the connectivity of the neural network structure.  ...  Recently, Neural Architecture Search (NAS) has been widely used for automatic network architecture design.  ... 
arXiv:2010.01097v1 fatcat:gndgp47sgbbb7fhun6pkkyijxe

Neural Graph Embedding for Neural Architecture Search

Wei Li, Shaogang Gong, Xiatian Zhu
Existing neural architecture search (NAS) methods often operate in discrete or continuous spaces directly, which ignores the graphical topology knowledge of neural networks.  ...  This leads to suboptimal search performance and efficiency, given the factor that neural networks are essentially directed acyclic graphs (DAG).  ...  Figure 1 : 1 The concept of neural graph architecture search.  ... 
doi:10.1609/aaai.v34i04.5903 fatcat:iphvazyxbbhirkhh5cop2e7tya

Graph Neural Network Architecture Search for Molecular Property Prediction [article]

Shengli Jiang, Prasanna Balaprakash
2020 arXiv   pre-print
Neural architecture search (NAS) is a promising approach to discover high-performing neural network architectures automatically.  ...  By treating molecule structure as graphs, where atoms and bonds are modeled as nodes and edges, graph neural networks (GNNs) have been widely used to predict molecular properties.  ...  Neural architecture search (NAS) has been designed to automatically search for the best network architecture for a given dataset.  ... 
arXiv:2008.12187v1 fatcat:urzxk3qefjb6bohxm4qacfpqgu

Graph neural architecture search: A survey

Babatounde Moctard Oloulade, Jianliang Gao, Jiamin Chen, Tengfei Lyu, Raeed Al-Sabri
2022 Tsinghua Science and Technology  
In academia and industries, graph neural networks (GNNs) have emerged as a powerful approach to graph data processing ranging from node classification and link prediction tasks to graph clustering tasks  ...  Hence, many studies have focused on building automated machine learning frameworks to search for the best GNN models for targeted tasks.  ...  The edge features of a graph are represented by a matrix X e 2 R m c , where m and c represent the number of edges and an edge feature size respectively.  ... 
doi:10.26599/tst.2021.9010057 fatcat:f4dck2qxjfbgbjc6h5u6ptw4ny

Smooth Variational Graph Embeddings for Efficient Neural Architecture Search [article]

Jovita Lukasik and David Friede and Arber Zela and Frank Hutter and Margret Keuper
2021 arXiv   pre-print
In this paper, we propose a two-sided variational graph autoencoder, which allows to smoothly encode and accurately reconstruct neural architectures from various search spaces.  ...  Neural architecture search (NAS) has recently been addressed from various directions, including discrete, sampling-based methods and efficient differentiable approaches.  ...  Neural Architecture Search via Bayesian Optimization.  ... 
arXiv:2010.04683v3 fatcat:geopbtghf5ftfcyxwwr4ckd7ym

Interpretable Neural Architecture Search via Bayesian Optimisation with Weisfeiler-Lehman Kernels [article]

Binxin Ru, Xingchen Wan, Xiaowen Dong, Michael Osborne
2021 arXiv   pre-print
Current neural architecture search (NAS) strategies focus only on finding a single, good, architecture.  ...  and graph-like search spaces amenable to BO.  ...  Introduction Neural architecture search (NAS) is a popular research direction recently that aims to automate the design process of good neural network architectures for a given task/dataset.  ... 
arXiv:2006.07556v2 fatcat:j7uu5zl3n5hjfepubenhxeel2y

Deep Graph Learning for Search & Recommender Systems

Bee-Chung Chen, Andrew Zhai
2022 Zenodo  
In this talk, we describe the architecture of deep-learning-based search and recommender systems, show how to apply GNNs learned from billions of nodes and edges to these systems to improve their performance  ...  Graph neural networks (GNNs) are a useful tool for learning such a representation not only based on the entity itself, but also its relationships with other entities.  ...  Inductive Learning for Adaptation ••• Graph Size: 3B nodes, 18B edges • Represent node via context features not unique id for inductive inference ○ Extract pin embedding seconds after creation Sample Method  ... 
doi:10.5281/zenodo.6507203 fatcat:zerrpyba5bb7nl2eaoh6aink6e

Automating Neural Architecture Design without Search [article]

Zixuan Liang, Yanan Sun
2022 arXiv   pre-print
Neural structure search (NAS), as the mainstream approach to automate deep neural architecture design, has achieved much success in recent years.  ...  We implemented the proposed approach by using a graph neural network for link prediction and acquired the knowledge from NAS-Bench-101.  ...  [45] uses Graph Convolutional Networks to generate feature embedding of the neural network architectures and pair these embeddings with their accuracies to train a neural predictor.  ... 
arXiv:2204.11838v1 fatcat:ge3ewpsqkjd4pjwz3wxwoq55xy

Graph Property Prediction on Open Graph Benchmark: A Winning Solution by Graph Neural Architecture Search [article]

Xu Wang and Huan Zhao and Lanning Wei and Quanming Yao
2022 arXiv   pre-print
PAS(Pooling Architecture Search).  ...  Aiming at two molecular graph datasets and one protein association subgraph dataset in OGB graph classification task, we design a graph neural network framework for graph classification task by introducing  ...  Graph Neural Architecture Search Neural architecture search (NAS) methods were proposed to automatically find SOTA CNN architectures in a pre-defined search space and representative methods are [15, 17  ... 
arXiv:2207.06027v1 fatcat:x7zxa3gyvvekfnkturqakfhnkm

A Generic Graph-based Neural Architecture Encoding Scheme for Predictor-based NAS [article]

Xuefei Ning, Yin Zheng, Tianchen Zhao, Yu Wang, Huazhong Yang
2020 arXiv   pre-print
This work proposes a novel Graph-based neural ArchiTecture Encoding Scheme, a.k.a. GATES, to improve the predictor-based neural architecture search.  ...  GATES is a more reasonable modeling of the neural architectures, and can encode architectures from both the "operation on node" and "operation on edge" cell search spaces consistently.  ...  .: Graph hypernetworks for neural architecture search. arXiv preprint arXiv:1810.05749 (2018) 30. Zoph, B., Le, Q.V.: Neural architecture search with reinforcement learning.  ... 
arXiv:2004.01899v3 fatcat:aofyv24iozhfvgej5esppusylq
« Previous Showing results 1 — 15 out of 32,613 results