0000000000061690
AUTHOR
Mahmoud Ramezani-mayiami
Graph recursive least squares filter for topology inference in causal data processes
In this paper, we introduce the concept of recursive least squares graph filters for online topology inference in data networks that are modelled as Causal Graph Processes (CGP). A Causal Graph Process (CGP) is an auto regressive process in the time series associated to different variables, and whose coefficients are the so-called graph filters, which are matrix polynomials with different orders of the graph adjacency matrix. Given the time series of data at different variables, the goal is to estimate these graph filters, hence the associated underlying adjacency matrix. Previously proposed algorithms have focused on a batch approach, assuming implicitly stationarity of the CGP. We propose…
Graph Topology Learning and Signal Recovery Via Bayesian Inference
The estimation of a meaningful affinity graph has become a crucial task for representation of data, since the underlying structure is not readily available in many applications. In this paper, a topology inference framework, called Bayesian Topology Learning, is proposed to estimate the underlying graph topology from a given set of noisy measurements of signals. It is assumed that the graph signals are generated from Gaussian Markov Random Field processes. First, using a factor analysis model, the noisy measured data is represented in a latent space and its posterior probability density function is found. Thereafter, by utilizing the minimum mean square error estimator and the Expectation M…
Robust Graph Topology Learning and Application in Stock Market Inference
In many applications, there are multiple interacting entities, generating time series of data over the space. To describe the relation within the set of data, the underlying topology may be used. In many real applications, not only the signal/data of interest is measured in noise, but it is also contaminated with outliers. The proposed method, called RGTL, infers the graph topology from noisy measurements and removes these outliers simultaneously. Here, it is assumed that we have no information about the space graph topology, while we know that graph signal are sampled consecutively in time and thus the graph in time domain is given. The simulation results show that the proposed algorithm h…
Joint Graph Learning and Signal Recovery via Kalman Filter for Multivariate Auto-Regressive Processes
In this paper, an adaptive Kalman filter algorithm is proposed for simultaneous graph topology learning and graph signal recovery from noisy time series. Each time series corresponds to one node of the graph and underlying graph edges express the causality among nodes. We assume that graph signals are generated via a multivariate auto-regressive processes (MAR), generated by an innovation noise and graph weight matrices. Then we relate the state transition matrix of Kalman filter to the graph weight matrices since both of them can play the role of signal propagation and transition. Our proposed Kalman filter for MAR processes, called KF-MAR, runs three main steps; prediction, update, and le…
Topology Inference and Signal Representation Using Dictionary Learning
This paper presents a Joint Graph Learning and Signal Representation algorithm, called JGLSR, for simultaneous topology learning and graph signal representation via a learned over-complete dictionary. The proposed algorithm alternates between three main steps: sparse coding, dictionary learning, and graph topology inference. We introduce the “transformed graph” which can be considered as a projected graph in the transform domain spanned by the dictionary atoms. Simulation results via synthetic and real data show that the proposed approach has a higher performance when compared to the well-known algorithms for joint undirected graph topology inference and signal representation, when there is…
JOINT TOPOLOGY LEARNING AND GRAPH SIGNAL RECOVERY VIA KALMAN FILTER IN CAUSAL DATA PROCESSES
In this paper, a joint graph-signal recovery approach is investigated when we have a set of noisy graph signals generated based on a causal graph process. By leveraging the Kalman filter framework, a three steps iterative algorithm is utilized to predict and update signal estimation as well as graph topology learning, called Topological Kalman Filter or TKF. Similar to the regular Kalman filter, we first predict the a posterior signal state based on the prior available data and then this prediction is updated and corrected based on the recently arrived measurement. But contrary to the conventional Kalman filter algorithm, we have no information of the transition matrix and hence we relate t…