Across many scientific domains, there is a common need to automatically extract simplified view or coarse-graining of how complex system's components interact. This general task called community detection in networks and analogous searching for clusters independent vector data. It evaluate the performance algorithms by their ability find so-called ground truth communities. works well synthetic with planted communities because these networks' links are formed explicitly based on those known...

This paper details the winning method in IEEE GOLD category of PHM psila08 Data Challenge. The task was to estimate remaining useable life left an unspecified complex system using a purely data driven approach. involves construction Multi-Layer Perceptron and Radial Basis Function networks for regression. A suitable selection these has been successfully combined ensemble Kalman filter. filter provides mechanism fusing multiple neural network model predictions over time. essential initial...

Assortative mixing in networks is the tendency for nodes with same attributes, or metadata, to link each other. It a property often found social manifesting as higher of links occurring between people age, race, political belief. Quantifying level assortativity disassortativity (the preference linking different attributes) can shed light on factors involved formation and contagion processes complex networks. common practice measure according coefficient, modularity case discrete-valued...

Interactions among people or objects are often dynamic in nature and can be represented as a sequence of networks, each providing snapshot the interactions over brief period time. An important task analyzing such evolving networks is change-point detection, which we both identify times at large-scale pattern changes fundamentally quantify how large what kind change occurred. Here, formalize for first time network detection problem within an online probabilistic learning framework introduce...

Time-of-Flight cameras provide high-frame-rate depth measurements within a limited range of distances. These readings can be extremely noisy and display unique errors, for instance, where scenes contain discontinuities or materials with low infrared reflectivity. Previous works have treated the amplitude each sample as measure confidence. In this paper, we demonstrate shortcomings common lone heuristic, propose an improved per-pixel confidence using Random Forest regressor trained real-world...

We study the fundamental limits on learning latent community structure in dynamic networks. Specifically, we stochastic block models where nodes change their membership over time, but edges are generated independently at each time step. In this setting (which is a special case of several existing models), able to derive detectability threshold exactly, as function rate and strength communities. Below threshold, claim that no algorithm can identify communities better than chance. then give...

Modular and hierarchical community structures are pervasive in real-world complex systems. A great deal of effort has gone into trying to detect study these structures. Important theoretical advances the detection modular have included identifying fundamental limits detectability by formally defining structure using probabilistic generative models. Detecting introduces additional challenges alongside those inherited from detection. Here we present a on networks, which thus far not received...

Recent advances in high‐throughput technologies are bringing the study of empirical genotype‐phenotype (GP) maps to fore. Here, we use data from protein‐binding microarrays an GP map transcription factor (TF) ‐binding preferences. In this map, each genotype is a DNA sequence. The phenotype sequence its ability bind one or more TFs. We using networks, which nodes represent genotypes with same phenotype, and edges connect if their differ by single small mutation. describe structure arrangement...

We develop a Bayesian hierarchical model to identify communities in networks for which we do not observe the edges directly, but instead series of interdependent signals each nodes. Fitting provides an end-to-end community detection algorithm that does extract information as sequence point estimates propagates uncertainties from raw data labels. Our approach naturally supports multiscale well selection optimal scale using comparison. study properties synthetic and apply it daily returns...

We consider the network constraints on bounds of assortativity coefficient, which aims to quantify tendency nodes with same attribute values be connected. The coefficient can considered as Pearson's correlation node metadata across edges and lies in interval $[\ensuremath{-}1,1]$. However, properties network, such degree distribution values, place upon attainable coefficient. This is important a particular value may say much about topology how are distributed over network---a fact often...

This paper considers the problem of algorithm selection for community detection. The aim detection is to identify sets nodes in a network which are more interconnected relative their connectivity rest network. A large number algorithms have been developed tackle this problem, but as with any machine learning task there no "one-size-fits-all" and each excels specific part space. examines performance weighted networks against those using unweighted different parts space (parameterised by...

Nodes in real world networks often have class labels, or underlying attributes, that are related to the way which they connect other nodes. Sometimes this relationship is simple, for instance nodes of same may be more likely connected. In cases, however, not true, and link a network exhibits different, complex their attributes. Here, we consider know how connected, but do labels relate links. We wish identify best subset label learn between node attributes can then use discovered accurately...

We study the problem of recovering a planted hierarchy partitions in network. The detectability single partition has previously been analyzed detail and phase transition identified below which cannot be detected. Here we show that, hierarchical setting, there exist additional phases presence multiple consistent can either help or hinder detection. Accordingly, limit for nonhierarchical typically provides insufficient information about complete structure, as highlight with several...

We study the effects of individual perceptions payoffs in two-player games. In particular we consider setting which individuals' game are influenced by their previous experiences and outcomes. Accordingly, introduce a framework based on evolutionary games where individuals have capacity to perceive interactions different ways. Starting from narrative social behaviors pub as an illustration, first combination prisoner's dilemma harmony two alternative same situation. Considering selection...

Professional team sports provide an excellent domain for studying the dynamics of social competitions. These games are constructed with simple, well-defined rules and payoffs that admit a high-dimensional set possible actions nontrivial scoring dynamics. The resulting gameplay efforts to predict its evolution object great interest both professionals enthusiasts. In this paper, we consider two online prediction problems sports: given partially observed game Who will score next? ultimately...

Collective classification models attempt to improve performance by taking into account the class labels of related instances. However, they tend not learn patterns interactions between classes and/or make assumption that instances same link each other (assortativity assumption). Blockmodels provide a solution these issues, being capable modelling assortative and disassortative interactions, learning pattern in form summary network. The Supervised Blockmodel provides good using structure...