We introduce a nonlinear method for directly embedding large, sparse, stochastic graphs into low-dimensional spaces, without requiring vertex features to reside in, or be transformed into, a metric space. Graph data and models are prevalent in real-world applications. Direct graph embedding is fundamental to many graph analysis tasks, in addition to graph visualization. We name the novel approach SG-t-SNE, as it is inspired by and builds upon the core principle of t-SNE, a widely used method for nonlinear dimensionality reduction and data visualization. We also introduce t-SNE-$\Pi$, a high-performance software for 2D, 3D embedding of large sparse graphs on personal computers with superior efficiency. It empowers SG-t-SNE with modern computing techniques for exploiting in tandem both matrix structures and memory architectures. We present elucidating embedding results on one synthetic graph and four real-world networks.

In topological data analysis, persistent homology is used to study the "shape of data". Persistent homology computations are completely characterized by a set of intervals called a bar code. It is often said that the long intervals represent the "topological signal" and the short intervals represent "noise". We give evidence to dispute this thesis, showing that the short intervals encode geometric information. Specifically, we prove that persistent homology detects the curvature of disks from which points have been sampled. We describe a general computational framework for solving inverse problems using the average persistence landscape, a continuous mapping from metric spaces with a probability measure to a Hilbert space. In the present application, the average persistence landscapes of points sampled from disks of constant curvature results in a path in this Hilbert space which may be learned using standard tools from statistical and machine learning.

Zero-shot learning (ZSL) is a framework to classify images belonging to unseen classes based on solely semantic information about these unseen classes. In this paper, we propose a new ZSL algorithm using coupled dictionary learning. The core idea is that the visual features and the semantic attributes of an image can share the same sparse representation in an intermediate space. We use images from seen classes and semantic attributes from seen and unseen classes to learn two dictionaries that can represent sparsely the visual and semantic feature vectors of an image. In the ZSL testing stage and in the absence of labeled data, images from unseen classes can be mapped into the attribute space by finding the joint sparse representation using solely the visual data. The image is then classified in the attribute space given semantic descriptions of unseen classes. We also provide an attribute-aware formulation to tackle domain shift and hubness problems in ZSL. Extensive experiments are provided to demonstrate the superior performance of our approach against the state of the art ZSL algorithms on benchmark ZSL datasets.

We propose a novel feature coding method that exploits invariance. We consider the setting where the transformations that preserve the image contents compose a finite group of orthogonal matrices. This is the case in many image transformations such as image rotations and image flipping. We prove that the group-invariant feature vector contains sufficient discriminative information when we learn a linear classifier using convex loss minimization. From this result, we propose a novel feature modeling for principal component analysis, and k-means clustering, which are used for most feature coding methods, and global feature functions that explicitly consider the group action. Although the global feature functions are complex nonlinear functions in general, we can calculate the group action on this space easily by constructing the functions as the tensor product representations of basic representations, resulting in the explicit form of invariant feature functions. We demonstrate the effectiveness of our methods on several image datasets.

Recently, matrix factorization-based recommendation methods have been criticized for the problem raised by the triangle inequality violation. Although several metric learning-based approaches have been proposed to overcome this issue, existing approaches typically project each user to a single point in the metric space, and thus do not suffice for properly modeling the intensity and the heterogeneity of user-item relationships in implicit feedback. In this paper, we propose TransCF to discover such latent user-item relationships embodied in implicit user-item interactions. Inspired by the translation mechanism popularized by knowledge graph embedding, we construct user-item specific translation vectors by employing the neighborhood information of users and items, and translate each user toward items according to the user's relationships with the items. Our proposed method outperforms several state-of-the-art methods for top-N recommendation on seven real-world data by up to 17% in terms of hit ratio. We also conduct extensive qualitative evaluations on the translation vectors learned by our proposed method to ascertain the benefit of adopting the translation mechanism for implicit feedback-based recommendations.

Graph kernels are an instance of the class of $\mathcal{R}$-Convolution kernels, which measure the similarity of objects by comparing their substructures. Despite their empirical success, most graph kernels use a naive aggregation of the final set of substructures, usually a sum or average, thereby potentially discarding valuable information about the distribution of individual components. Furthermore, only a limited instance of these approaches can be extended to continuously attributed graphs. We propose a novel method that relies on the Wasserstein distance between the node feature vector distributions of two graphs, which allows to find subtler differences in data sets by considering graphs as high-dimensional objects, rather than simple means. We further propose a Weisfeiler-Lehman inspired embedding scheme for graphs with continuous node attributes and weighted edges, enhance it with the computed Wasserstein distance, and thus improve the state-of-the-art prediction performance on several graph classification tasks.

Structured prediction can be considered as a generalization of many standard supervised learning tasks, and is usually thought as a simultaneous prediction of multiple labels. One standard approach is to maximize a score function on the space of labels, which decomposes as a sum of unary and pairwise potentials, each depending on one or two specific labels, respectively. For this approach, several learning and inference algorithms have been proposed over the years, ranging from exact to approximate methods while balancing the computational complexity. However, in contrast to binary and multiclass classification, results on the necessary number of samples for achieving learning is still limited, even for a specific family of predictors such as factor graphs. In this work, we provide minimax bounds for a class of factor-graph inference models for structured prediction. That is, we characterize the necessary sample complexity for any conceivable algorithm to achieve learning of factor-graph predictors.

Structured prediction can be thought of as a simultaneous prediction of multiple labels. This is often done by maximizing a score function on the space of labels, which decomposes as a sum of pairwise and unary potentials. The above is naturally modeled with a graph, where edges and vertices are related to pairwise and unary potentials, respectively. We consider the generative process proposed by Globerson et al. and apply it to general connected graphs. We analyze the structural conditions of the graph that allow for the exact recovery of the labels. Our results show that exact recovery is possible and achievable in polynomial time for a large class of graphs. In particular, we show that graphs that are bad expanders can be exactly recovered by adding small edge perturbations coming from the Erd\H{o}s-R\'enyi model. Finally, as a byproduct of our analysis, we provide an extension of Cheeger's inequality.

Electronic Health Records (EHRs) have been heavily used to predict various downstream clinical tasks such as readmission or mortality. One of the modalities in EHRs, clinical notes, has not been fully explored for these tasks due to its unstructured and inexplicable nature. Although recent advances in deep learning (DL) enables models to extract interpretable features from unstructured data, they often require a large amount of training data. However, many tasks in medical domains inherently consist of small sample data with lengthy documents; for a kidney transplant as an example, data from only a few thousand of patients are available and each patient's document consists of a couple of millions of words in major hospitals. Thus, complex DL methods cannot be applied to these kinds of domains. In this paper, we present a comprehensive ensemble model using vector space modeling and topic modeling. Our proposed model is evaluated on the readmission task of kidney transplant patients and improves 0.0211 in terms of c-statistics from the previous state-of-the-art approach using structured data, while typical DL methods fail to beat this approach. The proposed architecture provides the interpretable score for each feature from both modalities, structured and unstructured data, which is shown to be meaningful through a physician's evaluation.

Traditional structured prediction models try to learn the conditional likelihood, i.e., p(y x), to capture the relationship between the structured output y and the input features x. For many models, computing the likelihood is intractable. These models are therefore hard to train, requiring the use of surrogate objectives or variational inference to approximate likelihood. In this paper, we propose conditional Glow (c-Glow), a conditional generative flow for structured output learning. C-Glow benefits from the ability of flow-based models to compute p(y x) exactly and efficiently. Learning with c-Glow does not require a surrogate objective or performing inference during training. Once trained, we can directly and efficiently generate conditional samples to do structured prediction. We evaluate this approach on different structured prediction tasks and find c-Glow's structured outputs comparable in quality with state-of-the-art deep structured prediction approaches.