PINE: Universal Deep Embedding for Graph Nodes via Partial Permutation Invariant Set Functions.

Graph node embedding aims at learning a vector representation for all nodes given a graph. It is a central problem in many machine learning tasks (e.g., node classification, recommendation, community detection). The key problem in graph node embedding lies in how to define the dependence to neighbors. Existing approaches specify (either explicitly or implicitly) certain dependencies on neighbors, which may lead to loss of subtle but important structural information within the graph and other dependencies among neighbors. This intrigues us to ask the question: can we design a model to give the adaptive flexibility of dependencies to each node's neighborhood. In this paper, we propose a novel graph node embedding method (named PINE) via a novel notion of partial permutation invariant set function, to capture any possible dependence. Our method 1) can learn an arbitrary form of the representation function from the neighborhood, without losing any potential dependence structures, and 2) is applicable to both homogeneous and heterogeneous graph embedding, the latter of which is challenged by the diversity of node types. Furthermore, we provide theoretical guarantee for the representation capability of our method for general homogeneous and heterogeneous graphs. Empirical evaluation results on benchmark data sets show that our proposed PINE method outperforms the state-of-the-art approaches on producing node vectors for various learning tasks of both homogeneous and heterogeneous graphs.

Switching-State Dynamical Modeling of Daily Behavioral Data.

Emerging wearable and environmental sensor technologies provide health professionals with unprecedented capacity to continuously collect human behavioral data for health monitoring and management. This enables new solutions to mitigate globally emerging health problems such as obesity. With such outburst of dynamic sensor data, it is critical that appropriate mathematical models and computational methods are developed to translate the collected data into accurate characterization of the underlying health dynamics, enabling more reliable personalized monitoring, prediction, and intervention of health status changes. In addition to addressing common analytic challenges in analyzing sensor behavioral data, such as missing values and outliers, we focus on modeling heterogeneous dynamics to better capture health status changes under different conditions, which may lead to more effective state-dependent intervention strategies. We implement switching-state dynamic system models with different complexity levels on real-world daily behavioral data. Evaluation experiments of these models are conducted to demonstrate the importance of modeling the dynamic heterogeneity, as well as simultaneously conducting missing value imputation and outlier detection in achieving interpretable health dynamic models with better prediction of health status changes.

A scalable algorithm for structure identification of complex gene regulatory network from temporal expression data.

BACKGROUND: Gene regulatory interactions are of fundamental importance to various biological functions and processes. However, only a few previous computational studies have claimed success in revealing genome-wide regulatory landscapes from temporal gene expression data, especially for complex eukaryotes like human. Moreover, recent work suggests that these methods still suffer from the curse of dimensionality if a network size increases to 100 or higher. RESULTS: Here we present a novel scalable algorithm for identifying genome-wide gene regulatory network (GRN) structures, and we have verified the algorithm performances by extensive simulation studies based on the DREAM challenge benchmark data. The highlight of our method is that its superior performance does not degenerate even for a network size on the order of 104, and is thus readily applicable to large-scale complex networks. Such a breakthrough is achieved by considering both prior biological knowledge and multiple topological properties (i.e., sparsity and hub gene structure) of complex networks in the regularized formulation. We also validate and illustrate the application of our algorithm in practice using the time-course gene expression data from a study on human respiratory epithelial cells in response to influenza A virus (IAV) infection, as well as the CHIP-seq data from ENCODE on transcription factor (TF) and target gene interactions. An interesting finding, owing to the proposed algorithm, is that the biggest hub structures (e.g., top ten) in the GRN all center at some transcription factors in the context of epithelial cell infection by IAV. CONCLUSIONS: The proposed algorithm is the first scalable method for large complex network structure identification. The GRN structure identified by our algorithm could reveal possible biological links and help researchers to choose which gene functions to investigate in a biological event. The algorithm described in this article is implemented in MATLAB Ⓡ , and the source code is freely available from https://github.com/Hongyu-Miao/DMI.git .