Understanding the training dynamics of deep neural networks (DNNs) is important as it can lead to improved training efficiency and task performance. Recent works have demonstrated that representing the wirings of neurons in feedforward DNNs as graphs is an effective strategy for understanding how architectural choices can affect performance. However, these approaches fail to model training dynamics since a single, static graph cannot capture how DNNs change over the course of training. Thus, in this work, we propose a compact, expressive temporal graph framework that effectively captures the dynamics of many workhorse architectures in computer vision. Specifically, our framework extracts an informative summary of graph properties (e.g., degree, eigenvector centrality) over a sequence of DNN graphs obtained during training. We demonstrate that the proposed framework captures useful dynamics by accurately predicting trained, task performance when using a summary over early training epochs (<5) across four different architectures and two image datasets. Moreover, by using a novel, highly-scalable DNN graph representation, we further demonstrate that the proposed framework captures generalizable dynamics as summaries extracted from smaller-width networks are effective when evaluated on larger widths.

Network alignment, or the task of finding meaningful node correspondences between nodes in di↵erent graphs, is an important graph mining task with many scientific and industrial applications.An important principle for network alignment is matched neighborhood consistency (MNC): nodes that are close in one graph should be matched to nodes that are close in the other graph.We theoretically demonstrate a close relationship between MNC and alignment accuracy.As many existing network alignment methods struggle to preserve topological consistency in di cult scenarios, we show how to refine their solutions by improving their MNC.Our refinement method, RefiNA, is straightforward to implement, admits scalable sparse approximation, and can be paired post hoc with any network alignment method.Extensive experiments show that RefiNA increases the accuracy of diverse unsupervised network alignment methods by up to 90%, making them robust enough to align graphs that are 5⇥ more topologically di↵erent than were considered in prior work.

Depression and anxiety are critical public health issues affecting millions of people around the world. To identify individuals who are vulnerable to depression and anxiety, predictive models have been built that typically utilize data from one source. Unlike these traditional models, in this study, we leverage a rich heterogeneous dataset from the University of Notre Dame’s NetHealth study that collected individuals’ (student participants’) social interaction data via smartphones, health-related behavioral data via wearables (Fitbit), and trait data from surveys. To integrate the different types of information, we model the NetHealth data as a heterogeneous information network (HIN). Then, we redefine the problem of predicting individuals’ mental health conditions (depression or anxiety) in a novel manner, as applying to our HIN a popular paradigm of a recommender system (RS), which is typically used to predict the preference that a person would give to an item (e.g., a movie or book). In our case, the items are the individuals’ different mental health states. We evaluate four state-of-the-art RS approaches. Also, we model the prediction of individuals’ mental health as another problem type—that of node classification (NC) in our HIN, evaluating in the process four node features under logistic regression as a proof-of-concept classifier. We find that our RS and NC network methods produce more accurate predictions than a logistic regression model using the same NetHealth data in the traditional non-network fashion as well as a random-approach. Also, we find that the best of the considered RS approaches outperforms all considered NC approaches. This is the first study to integrate smartphone, wearable sensor, and survey data in a HIN manner and use RS or NC on the HIN to predict individuals’ mental health conditions.

Network alignment, the process of finding correspondences between nodes in different graphs, has many scientific and industrial applications. Existing unsupervised network alignment methods find suboptimal alignments that break up node neighborhoods, i.e. do not preserve matched neighborhood consistency. To improve this, we propose CONE-Align, which models intra-network proximity with node embeddings and uses them to match nodes across networks after aligning the embedding subspaces. Experiments on diverse, challenging datasets show that CONE-Align is robust and obtains 19.25% greater accuracy on average than the best-performing state-of-the-art graph alignment algorithm in highly noisy settings.

In the information overloaded web, personalized recommender systems are essential tools to help users find most relevant information. The most heavily-used recommendation frameworks assume user interactions that are characterized by a single relation. However, for many tasks, such as recommendation in social networks, user-item interactions must be modeled as a complex network of multiple relations, not only a single relation. Recently research on multi-relational factorization and hybrid recommender models has shown that using extended meta-paths to capture additional information about both users and items in the network can enhance the accuracy of recommendations in such networks. Most of this work is focused on unweighted heterogeneous networks, and to apply these techniques, weighted relations must be simplified into binary ones. However, information associated with weighted edges, such as user ratings, which may be crucial for recommendation, are lost in such binarization. In this paper, we explore a random walk sampling method in which the frequency of edge sampling is a function of edge weight, and apply this generate extended meta-paths in weighted heterogeneous networks. With this sampling technique, we demonstrate improved performance on multiple data sets both in terms of recommendation accuracy and model generation efficiency.

Recommender systems are key components in information-seeking contexts where personalization is sought. However, the dominant framework for recommendation is essentially two dimensional, with the interaction between users and items characterized by a single relation. In many cases, such as social networks, users and items are joined in a complex web of relations, not readily reduced to a single value. Recent multirelational approaches to recommendation focus on the direct, proximal relations in which users and items may participate. Our approach uses the framework of complex heterogeneous networks to represent such recommendation problems. We propose the weighted hybrid of low-dimensional recommenders (WHyLDR) recommendation model, which uses extended relations, represented as constrained network paths, to effectively augment direct relations. This model incorporates influences from both distant and proximal connections in the network. The WHyLDR approach raises the problem of the unconstrained proliferation of components, built from ever-extended network paths. We show that although component utility is not strictly monotonic with path length, a measure based on information gain can effectively prune and optimize such hybrids.

Social media sites accumulate a wide variety of information about users: likes and ratings, friend and follower links, annotations, posts, media uploads, just to name a few. Key challenges for recommender systems research are (a) to synthesize of all of this data into an integrated recommendation model and (b) to support a wide variety of recommendation types simultaneously (items, friends, tags, etc.) One approach that has been explored in recent research is to view this multi-faceted data as a heterogeneous network and use network-based methods of generating recommendations. However, most such approaches involve computationally-intensive model generation resulting in a single-purpose recommender system. Our approach is to create a component-based hybrid model whose components can be reused for multiple recommendation tasks. In this paper, we show how this model can be applied to heterogeneous networks.