Deep learning has revolutionized many machine learning tasks in recent years, ranging from image classification and video processing to speech recognition and natural language understanding. The data in these tasks are typically represented in the Euclidean space. However, there is an increasing number of applications, where data are generated from non-Euclidean domains and are represented as graphs with complex relationships and interdependency between objects. The complexity of graph data has imposed significant challenges on the existing machine learning algorithms. Recently, many studies on extending deep learning approaches for graph data have emerged. In this article, we provide a comprehensive overview of graph neural networks (GNNs) in data mining and machine learning fields. We propose a new taxonomy to divide the state-of-the-art GNNs into four categories, namely, recurrent GNNs, convolutional GNNs, graph autoencoders, and spatial–temporal GNNs. We further discuss the applications of GNNs across various domains and summarize the open-source codes, benchmark data sets, and model evaluation of GNNs. Finally, we propose potential research directions in this rapidly growing field.

/pdf/a-comprehensive-survey-on-graph-neural-networks-4lw9i4v183.pdf

A Comprehensive Survey on Graph Neural Networks

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

This chapter explores the ecology of human development, those forces in the person's environment that affect and influence development. Urie Bronfenbrenner's model of the human ecosystem guides the discussion, making connections between children in families and in communities and the larger society that surrounds them. The human ecosystem model is much like the study of the natural ecology, focusing on the interactions between subjects at various levels of the environment as they affect each other. The interaction between individual and environment forms the basis of an ecological approach to human development. This view sees the process of development as the expansion of the child's conception of the world and the child's ability to act on that world. Risks to development can come from both direct threats and the absence of opportunities for development. Sociocultural risk refers to the impoverishment in the child's world of essential experiences and relationships.

The Ecology of Human Development

A morphable model for the synthesis of 3D faces

The Cambridge Grammar of the English Language

Convolutional neural networks (CNNs) have recently led to incredible breakthroughs on a variety of pattern recognition problems. Banks of finite-impulse response filters are learned on a hierarchy of layers, each contributing more abstract information than the previous layer. The simplicity and elegance of the convolutional filtering process makes them perfect for structured problems, such as image, video, or voice, where vertices are homogeneous in the sense of number, location, and strength of neighbors. The vast majority of classification problems, for example in the pharmaceutical, homeland security, and financial domains are unstructured. As these problems are formulated into unstructured graphs, the heterogeneity of these problems, such as number of vertices, number of connections per vertex, and edge strength, cannot be tackled with standard convolutional techniques. We propose a novel neural learning framework that is capable of handling both homogeneous and heterogeneous data while retaining the benefits of traditional CNN successes. Recently, researchers have proposed variations of CNNs that can handle graph data. In an effort to create learnable filter banks of graphs, these methods either induce constraints on the data or require preprocessing. As opposed to spectral methods, our framework, which we term Graph-CNNs, defines filters as polynomials of functions of the graph adjacency matrix. Graph-CNNs can handle both heterogeneous and homogeneous graph data, including graphs having entirely different vertex or edge sets. We perform experiments to validate the applicability of Graph-CNNs to a variety of structured and unstructured classification problems and demonstrate state-of-the-art results on document and molecule classification problems.

Robust Spatial Filtering With Graph Convolutional Neural Networks

Intelligent character recognition using fully convolutional neural networks

A computing system for automatically identifying individual regions in a digital image that include a human face, forming combined padded regions and automatically evaluating each of the regions including calculating a fitness score for each of the combined padded regions and ignoring at least one of the combined padded regions whose fitness score is below a known threshold

Image recomposition from face detection and facial features

The disclosed method uses empirically derived image quality data in order to determine an image resolution for a particular zoom and/or crop operation performed on an image. An acceptability value is specified for the image, e.g., a specific % acceptable, where the acceptability value relates to empirically derived image quality as perceived by a human viewer. Given selected input and output devices, an image resolution is then generated from a combination of the particular zoom and crop and the specified acceptability value. This resolution is specified as the image resolution, e.g., a minimally acceptable resolution, to produce the desired zoom and crop.

Method for determining necessary resolution for zoom and crop images

A method of operating a digital image capture device including storing a plurality of digital image template files, each of the digital image template files include features of a digital image template such as a number, location, and shape of openings in the digital image template Image attribute requirements are also stored in association with each of the openings for screening images to be disposed in an opening The templates are displayed to a user for selecting a template Images are captured and stored and each of the images include image attribute data Stored images are automatically searched to identify images that may be automatically digitally placed into openings in the digital template The image attributes of found images are those which satisfy image attribute requirements of the openings A file containing a digital image template together with the images digitally disposed therein is saved in the capture device for future use such as printing the file or transmitting the file over a network

Raymond Ptucha

Papers

Robust Spatial Filtering With Graph Convolutional Neural Networks

Intelligent character recognition using fully convolutional neural networks

Image recomposition from face detection and facial features

Method for determining necessary resolution for zoom and crop images

Image capture method with artistic template design