15 results match your criteria gnn structures

  • Page 1 of 1

Deep Constraint-based Propagation in Graph Neural Networks.

IEEE Trans Pattern Anal Mach Intell 2021 Apr 15;PP. Epub 2021 Apr 15.

The popularity of deep learning techniques renewed the interest in neural architectures able to process complex structures that can be represented using graphs, inspired by Graph Neural Networks (GNNs). We focus our attention on the originally proposed GNN model of Scarselli et al. 2009, which encodes the state of the nodes of the graph by means of an iterative diffusion procedure that, during the learning stage, must be computed at every epoch, until the fixed point of a learnable state transition function is reached, propagating the information among the neighbouring nodes. Read More

View Article and Full-Text PDF

Multigraph Transformer for Free-Hand Sketch Recognition.

IEEE Trans Neural Netw Learn Syst 2021 Apr 7;PP. Epub 2021 Apr 7.

Learning meaningful representations of free-hand sketches remains a challenging task given the signal sparsity and the high-level abstraction of sketches. Existing techniques have focused on exploiting either the static nature of sketches with convolutional neural networks (CNNs) or the temporal sequential property with recurrent neural networks (RNNs). In this work, we propose a new representation of sketches as multiple sparsely connected graphs. Read More

View Article and Full-Text PDF

Anomaly Detection on Attributed Networks via Contrastive Self-Supervised Learning.

IEEE Trans Neural Netw Learn Syst 2021 Apr 5;PP. Epub 2021 Apr 5.

Anomaly detection on attributed networks attracts considerable research interests due to wide applications of attributed networks in modeling a wide range of complex systems. Recently, the deep learning-based anomaly detection methods have shown promising results over shallow approaches, especially on networks with high-dimensional attributes and complex structures. However, existing approaches, which employ graph autoencoder as their backbone, do not fully exploit the rich information of the network, resulting in suboptimal performance. Read More

View Article and Full-Text PDF

Learning Aligned Image-Text Representations Using Graph Attentive Relational Network.

IEEE Trans Image Process 2021 18;30:1840-1852. Epub 2021 Jan 18.

Image-text matching aims to measure the similarities between images and textual descriptions, which has made great progress recently. The key to this cross-modal matching task is to build the latent semantic alignment between visual objects and words. Due to the widespread variations of sentence structures, it is very difficult to learn the latent semantic alignment using only global cross-modal features. Read More

View Article and Full-Text PDF
January 2021

Prediction of Liquid Chromatographic Retention Time with Graph Neural Networks to Assist in Small Molecule Identification.

Anal Chem 2021 02 7;93(4):2200-2206. Epub 2021 Jan 7.

College of Chemistry and Chemical Engineering, Central South University, Changsha 410083, China.

The predicted liquid chromatographic retention times (RTs) of small molecules are not accurate enough for wide adoption in structural identification. In this study, we used the graph neural network to predict the retention time (GNN-RT) from structures of small molecules directly without the requirement of molecular descriptors. The predicted accuracy of GNN-RT was compared with random forests (RFs), Bayesian ridge regression, convolutional neural network (CNN), and a deep-learning regression model (DLM) on a METLIN small molecule retention time (SMRT) dataset. Read More

View Article and Full-Text PDF
February 2021

Learning on Attribute-Missing Graphs.

IEEE Trans Pattern Anal Mach Intell 2020 Oct 19;PP. Epub 2020 Oct 19.

Graphs with complete node attributes have been widely explored recently. While in practice, there is a graph where attributes of only partial nodes could be available and those of the others might be entirely missing. This attribute-missing graph is related to numerous real-world applications and there are limited studies investigating the corresponding learning problems. Read More

View Article and Full-Text PDF
October 2020

Improving Attention Mechanism in Graph Neural Networks via Cardinality Preservation.

Shuo Zhang Lei Xie

IJCAI (U S) 2020 Jul;2020:1395-1402

Ph.D. Program in Computer Science, The Graduate Center, The City University of New York.

Graph Neural Networks (GNNs) are powerful for the representation learning of graph-structured data. Most of the GNNs use a message-passing scheme, where the embedding of a node is iteratively updated by aggregating the information from its neighbors. To achieve a better expressive capability of node influences, attention mechanism has grown to be popular to assign trainable weights to the nodes in aggregation. Read More

View Article and Full-Text PDF

GNNExplainer: Generating Explanations for Graph Neural Networks.

Adv Neural Inf Process Syst 2019 Dec;32:9240-9251

Department of Computer Science, Stanford University.

Graph Neural Networks (GNNs) are a powerful tool for machine learning on graphs. GNNs combine node feature information with the graph structure by recursively passing neural messages along edges of the input graph. However, incorporating both graph structure and feature information leads to complex models and explaining predictions made by GNNs remains unsolved. Read More

View Article and Full-Text PDF
December 2019

Aggregation kinetics of short peptides: All-atom and coarse-grained molecular dynamics study.

Biophys Chem 2019 10 5;253:106219. Epub 2019 Jul 5.

Adam Mickiewicz University in Poznań, Faculty of Chemistry, Umultowska 89b, 61-614 Poznań, Poland. Electronic address:

Peptides can aggregate into ordered structures with different morphologies. The aggregation mechanism and evolving structures are the subject of intense research. In this paper we have used molecular dynamics to examine the sequence-dependence of aggregation kinetics for three short peptides: octaalanine (Ala8), octaasparagine (Asn8), and the heptapeptide GNNQQNY (abbreviated as GNN). Read More

View Article and Full-Text PDF
October 2019

Graphene nanonet for biological sensing applications.

Nanotechnology 2013 Sep 21;24(37):375302. Epub 2013 Aug 21.

Department of Physics and Astronomy, Seoul National University, Seoul, Korea.

We report a simple but efficient method to fabricate versatile graphene nanonet (GNN)-devices. In this method, networks of V2O5 nanowires (NWs) were prepared in specific regions of single-layer graphene, and the graphene layer was selectively etched via a reactive ion etching method using the V2O5 NWs as a shadow mask. The process allowed us to prepare large scale patterns of GNN structures which were comprised of continuous networks of graphene nanoribbons (GNRs) with chemical functional groups on their edges. Read More

View Article and Full-Text PDF
September 2013

Intrinsic dynamics of an extended hydrophobic core in the S. cerevisiae RNase III dsRBD contributes to recognition of specific RNA binding sites.

J Mol Biol 2013 Feb 28;425(3):546-62. Epub 2012 Nov 28.

Department of Chemistry and Biochemistry, and Molecular Biology Institute, University of California, Los Angeles, Los Angeles, CA 90095-1569, USA.

The Saccharomyces cerevisiae RNase III enzyme Rnt1p preferentially binds to double-stranded RNA hairpin substrates with a conserved (A/u)GNN tetraloop fold, via shape-specific interactions by its double-stranded RNA-binding domain (dsRBD) helix α1 to the tetraloop minor groove. To investigate whether conformational flexibility in the dsRBD regulates the binding specificity, we determined the backbone dynamics of the Rnt1p dsRBD in the free and AGAA hairpin-bound states using NMR spin-relaxation experiments. The intrinsic microsecond-to-millisecond timescale dynamics of the dsRBD suggests that helix α1 undergoes conformational sampling in the free state, with large dynamics at some residues in the α1-β1 loop (α1-β1 hinge). Read More

View Article and Full-Text PDF
February 2013

A novel family of RNA tetraloop structure forms the recognition site for Saccharomyces cerevisiae RNase III.

EMBO J 2001 Dec;20(24):7240-9

Department of Chemistry and Biochemistry, 405 Hilgard Avenue, PO Box 951569, University of California, Los Angeles, CA 90095-1569, USA.

RNases III are a family of double-stranded RNA (dsRNA) endoribonucleases involved in the processing and decay of a large number of cellular RNAs as well as in RNA interference. The dsRNA substrates of Saccharomyces cerevisiae RNase III (Rnt1p) are capped by tetraloops with the consensus sequence AGNN, which act as the primary docking site for the RNase. We have solved the solution structures of two RNA hairpins capped by AGNN tetraloops, AGAA and AGUU, using NMR spectroscopy. Read More

View Article and Full-Text PDF
December 2001

Obcells as proto-organisms: membrane heredity, lithophosphorylation, and the origins of the genetic code, the first cells, and photosynthesis.

T Cavalier-Smith

J Mol Evol 2001 Oct-Nov;53(4-5):555-95

Department of Zoology, University of Oxford, South Parks Road, Oxford, OX1 3PS, United Kingdom.

I attempt to sketch a unified picture of the origin of living organisms in their genetic, bioenergetic, and structural aspects. Only selection at a higher level than for individual selfish genes could power the cooperative macromolecular coevolution required for evolving the genetic code. The protein synthesis machinery is too complex to have evolved before membranes. Read More

View Article and Full-Text PDF
February 2002

A comparative study of ligand-receptor complex binding affinity prediction methods based on glycogen phosphorylase inhibitors.

S S So M Karplus

J Comput Aided Mol Des 1999 May;13(3):243-58

Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138, USA.

Finding an accurate method for estimating the affinity of protein ligands activity is the most challenging task in computer-aided molecular design. In this study we investigate and compare seven different prediction methods for a set of 30 glycogen phosphorylase (GP) inhibitors with known crystal structures. Five of the methods involve quantitative structure-activity relationships (QSAR) based on the 2D or 3D structures of the GP ligands alone. Read More

View Article and Full-Text PDF

Remarkable morphological variability of a common RNA folding motif: the GNRA tetraloop-receptor interaction.

J Mol Biol 1997 Feb;266(3):493-506

Department of Biochemistry and Molecular Biophysics, Columbia University, New York, NY 10032, USA.

One of the most common RNA tertiary interactions involves the docking of GNRA hairpin loops into stem-loop structures on other regions of RNA. Domain 5 of the group II intron interacts with Domain 1 through such an interaction, which has been characterized thermodynamically and kinetically for the ai5g intron. Using this system, it was possible to test the morphological tolerances of the GNRA tetraloop involved in tertiary interactions. Read More

View Article and Full-Text PDF
February 1997
  • Page 1 of 1