Filters








309 Hits in 5.8 sec

Program-to-Circuit: Exploiting GNNs for Program Representation and Circuit Translation [article]

Nan Wu, Huake He, Yuan Xie, Pan Li, Cong Hao
2021 arXiv   pre-print
Third, we discuss three sets of real-world benchmarks for GNN generalization evaluation, and analyze the performance gap between standard programs and the real-case ones.  ...  GNNs targeting this high-demand circuit design area.  ...  Traditional EDA tools for circuit design usually take hours to days to accurately predict circuit quality and require extensive manual efforts; however, even though high-level synthesis (HLS) tools accelerate  ... 
arXiv:2109.06265v1 fatcat:fvx5axayprd7xa6vihz35eihgq

Model-Architecture Co-Design for High Performance Temporal GNN Inference on FPGA [article]

Hongkuan Zhou, Bingyi Zhang, Rajgopal Kannan, Viktor Prasanna, Carl Busart
2022 arXiv   pre-print
Taking advantage of the model optimizations, we propose a principled hardware architecture using batching, pipelining, and prefetching techniques to further improve the performance.  ...  Real-world applications require high performance inference on real-time streaming dynamic graphs.  ...  The accelerators are developed using Xilinx High-level Synthesis (HLS). HLS is a pragma-directive programming language that allows user to develop the accelerator design using C/C++.  ... 
arXiv:2203.05095v1 fatcat:5xtb6tklvjedrphekn4qztwqnu

G-CoS: GNN-Accelerator Co-Search Towards Both Better Accuracy and Efficiency [article]

Yongan Zhang, Haoran You, Yonggan Fu, Tong Geng, Ang Li, Yingyan Lin
2021 arXiv   pre-print
Extensive experiments and ablation studies show that the GNNs and accelerators generated by G-CoS consistently outperform SOTA GNNs and GNN accelerators in terms of both task accuracy and hardware efficiency  ...  design spaces of GNNs and their accelerators.  ...  GNN variants and their implementations Many advanced GNN variants have recently been proposed to consider different aggregation functions and introduce additional attention modules or sampling functions  ... 
arXiv:2109.08983v1 fatcat:hvxnl3ytwza4vky6iat7rnre2m

AdaGCN:Adaptive Boosting Algorithm for Graph Convolutional Networks on Imbalanced Node Classification [article]

S. Shi, Kai Qiao, Shuai Yang, L. Wang, J. Chen, Bin Yan
2021 arXiv   pre-print
Experiments show that the AdaGCN model we proposed achieves better performance than GCN, GraphSAGE, GAT, N-GCN and the most of advanced reweighting and resampling methods on synthetic imbalanced datasets  ...  GNN.  ...  It is difficult for a single model to accurately predict rare and few points on an imbalanced dataset, and overall performance is limited.  ... 
arXiv:2105.11625v1 fatcat:b3mnj5hat5gjtn4yk2bpxvue44

Accelerating SpMM Kernel with Cache-First Edge Sampling for Graph Neural Networks [article]

Chien-Yu Lin, Liang Luo, Luis Ceze
2021 arXiv   pre-print
To evaluate ES-SpMM's performance, we integrated it with a popular GNN framework, DGL, and tested it using representative GNN models and datasets.  ...  Our analysis shows that 95% of the inference time could be spent on SpMM when running popular GNN models on NVIDIA's advanced V100 GPU.  ...  INTRODUCTION GNNs are a class of powerful deep learning (DL) models that can extract high-level embeddings from graph-structured data.  ... 
arXiv:2104.10716v2 fatcat:mljvwog2nngz3j3liz72mem42u

Quantitative Evaluation of Explainable Graph Neural Networks for Molecular Property Prediction [article]

Jiahua Rao, Shuangjia Zheng, Yuedong Yang
2021 arXiv   pre-print
In this work, we first build three levels of benchmark datasets to quantitatively assess the interpretability of the state-of-the-art GNN models.  ...  Advances in machine learning have led to graph neural network-based methods for drug discovery, yielding promising results in molecular design, chemical synthesis planning, and molecular property prediction  ...  Based on the benchmarks, we implemented state-of-theart XAI and commonly used GNN models, and provided a uniform and rigorous framework to evaluate their performances. for example, McCloskey et al. (2019  ... 
arXiv:2107.04119v2 fatcat:w5ajrkh3ajg2tmm4toj5yp6mqi

GNN4REL: Graph Neural Networks for Predicting Circuit Reliability Degradation [article]

Lilas Alrahis, Johann Knechtel, Florian Klemme, Hussam Amrouch, Ozgur Sinanoglu
2022 arXiv   pre-print
This problem is exacerbated for advanced technology nodes, where transistor dimensions reach atomic levels and established margins are severely constrained.  ...  incorporated into the GNN model via training by the foundry.  ...  Besides, this work is also supported by Advantest as part of the Graduate School "Intelligent Methods for Test and Reliability" (GS-IMTR) at the University of Stuttgart.  ... 
arXiv:2208.02868v1 fatcat:am5a5zhl4bhippfc67yjmjnhva

Hybrid Graph Models for Logic Optimization via Spatio-Temporal Information [article]

Nan Wu, Jiwon Lee, Yuan Xie, Cong Hao
2022 arXiv   pre-print
The key idea is to simultaneously leverage spatio-temporal information from hardware designs and logic synthesis flows to forecast performance (i.e., delay/area) of various synthesis flows on different  ...  on hardware designs by combining a virtually added supernode or a sequence processing model with conventional GNN models.  ...  Aiming at a practical use of ML-based performance model, the generalization across different designs and flow lengths is a necessity.  ... 
arXiv:2201.08455v1 fatcat:63h7zpdmffgaxgms3i3gdryewu

Learning Execution through Neural Code Fusion [article]

Zhan Shi, Kevin Swersky, Daniel Tarlow, Parthasarathy Ranganathan, Milad Hashemi
2020 arXiv   pre-print
As an illustration of this, we apply the new model to challenging dynamic tasks (branch prediction and prefetching) from the SPEC CPU benchmark suite, outperforming the state-of-the-art by 26% and 45%  ...  We show that this leads to improved performance over similar methods that do not use execution and it opens the door to applying GNN models to new tasks that would not be feasible from static code alone  ...  First, instead of using high level source code, we construct a new graph representation of low-level assembly code and model it with a graph neural network.  ... 
arXiv:1906.07181v2 fatcat:czxoxlbifbgn3kvssezsyzjclu

MuxLink: Circumventing Learning-Resilient MUX-Locking Using Graph Neural Network-based Link Prediction [article]

Lilas Alrahis, Satwik Patnaik, Muhammad Shafique, Ozgur Sinanoglu
2021 arXiv   pre-print
The proposed MuxLink achieves key prediction accuracy and precision up to 100% on D-MUX and symmetric MUX-locked ISCAS-85 and ITC-99 benchmarks, fully unlocking the designs.  ...  Our trained GNN model learns the structure of the given circuit and the composition of gates around the non-obfuscated wires, thereby generating meaningful link embeddings that help decipher the secret  ...  bit at a time and perform re-synthesis.  ... 
arXiv:2112.07178v1 fatcat:4zqymdsq6zgdjk5hbq4fgnc3pi

Regio-selectivity prediction with a machine-learned reaction representation and on-the-fly quantum mechanical descriptors

yanfei guan, Connor W Coley, Haoyang Wu, Ranasinghe Duminda, Esther Heid, Thomas James Struble, Lagnajit Pattanaik, William H. Green, Klavs F Jensen
2021 Chemical Science  
Accurate and rapid evaluation of whether substrates can undergo the desired the transformation iscrucial and challenging for both human knowledge and computer predictions.  ...  Acknowledgements The authors thank the Machine Learning for Pharmaceutical Discovery and Synthesis (MLPDS) Consortium for support.  ...  E.H. acknowledges support from the Austrian Science Fund (FWF), project J-4415.We thank Pritha Verma and Jessica Xu for helpful comments and discussions on this manuscript.  ... 
doi:10.1039/d0sc04823b pmid:34163985 pmcid:PMC8179287 fatcat:xgp53x4x5fd6vgwuvddkrypada

A Comprehensive Survey on Electronic Design Automation and Graph Neural Networks: Theory and Applications

Daniela Sánchez Lopera, Lorenzo Servadei, Gamze Naz Kiprit, Robert Wille, Wolfgang Ecker
2022 ACM Transactions on Design Automation of Electronic Systems  
The trending Graph Neural Networks (GNNs) are an opportunity to solve EDA problems directly using graph structures for circuits, intermediate RTLs, and netlists.  ...  We map those works to a design pipeline by defining graphs, tasks, and model types. Furthermore, we analyze their practical implications and outcomes.  ...  ACKNOWLEDGMENTS The work described herein is partly funded by the German Federal Ministry for Economic Afairs and Energy (BMWi) as part of the research project ProgressivKI (19A21006C).  ... 
doi:10.1145/3543853 fatcat:halhg7zwgvdpjkzf7g7ctxzuby

Machine intelligence for chemical reaction space

Philippe Schwaller, Alain C. Vaucher, Ruben Laplaza, Charlotte Bunne, Andreas Krause, Clemence Corminboeuf, Teodoro Laino
2022 Wiley Interdisciplinary Reviews. Computational Molecular Science  
Discovering new reactions, optimizing their performance, and extending the synthetically accessible chemical space are critical drivers for major technological advances and more sustainable processes.  ...  Machine intelligence has emerged as a potential game-changer for chemical reaction space exploration and the synthesis of novel molecules and materials.  ...  High top-N accuracy on those benchmarks does not necessarily translate to a good model performance when used for multi-step synthesis planning, as other criteria, like the diversity of the suggested reactions  ... 
doi:10.1002/wcms.1604 fatcat:3p2wz5i5pvgojaeiyaikvw4rua

PDBench: Evaluating Computational Methods for Protein Sequence Design [article]

Leonardo V. Castorina, Rokas Petrenas, Kartic Subr, Christopher W. Wood
2021 arXiv   pre-print
We compare five existing models with two novel models for sequence prediction.  ...  Our robust benchmark provides biological insight into the behaviour of design methods, which is essential for evaluating their performance and utility.  ...  Acknowledgements CWW is supported by an Engineering and Physical Sciences Research Council Fellowship (EP/S003002/1).  ... 
arXiv:2109.07925v3 fatcat:mzqo7hn3nzhqjm2am5zyos3uli

A Survey of Trustworthy Graph Learning: Reliability, Explainability, and Privacy Protection [article]

Bingzhe Wu, Jintang Li, Junchi Yu, Yatao Bian, Hengtong Zhang, CHaochao Chen, Chengbin Hou, Guoji Fu, Liang Chen, Tingyang Xu, Yu Rong, Xiaolin Zheng (+8 others)
2022 arXiv   pre-print
In contrast to conventional graph learning research which mainly cares about model performance, TwGL considers various reliability and safety aspects of the graph learning framework including but not limited  ...  Deep graph learning has achieved remarkable progresses in both business and scientific areas ranging from finance and e-commerce, to drug and advanced material discovery.  ...  uncertainty can be used for rejecting unreliable decisions with high model predictive uncertainties.  ... 
arXiv:2205.10014v2 fatcat:aobv34rwg5ehpka4fsuar2gm7i
« Previous Showing results 1 — 15 out of 309 results