Towards Real-Time Information Processing of Sensor Network Data Using Computationally Efficient Multi-output Gaussian Processes

By Michael A. Osborne, Stephen J. Roberts, Alex Rogers, Sarvapali D. Ramchurn, Nicholas R. Jennings. 2008

189 citations recorded by the source corpus.

In this paper, we describe a novel, computationally efficient algorithm that facilitates the autonomous acquisition of readings from sensor networks (deciding when and which sensor to acquire readings from at any time), and which can, with minimal domain knowledge, perform a range of information processing tasks including modelling the accuracy of the sensor readings, predicting the value of missing sensor readings, and predicting how the monitored environmental variables will evolve into the future. Our motivating scenario is the need to provide situational awareness support to first responders at the scene of a large scale incident, and to this end, we describe a novel iterative formulation of a multi-output Gaussian process that can build and exploit a probabilistic model of the environmental variables being measured (including the correlations and delays that exist between them). We validate our approach using data collected from a network of weather sensors located on the south coast of England.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Kernels for Vector-Valued Functions: a Review (2011, methodology): “The intrinsic coregionalization model has also been used by [72].”
Gaussian Process Regression Networks (2011, background): “…an explosion of interest in extending the Gaussian process regression framework to account for ﬁxed correlations between output variables (Alvarez and Lawrence, 2011; Yu et al., 2009; Alvarez and Lawrence, 2008; Bonilla et al., 2008; Osborne et al., 2008; Teh et al., 2005; Boyle and Frean, 2004).”
Machine learning of linear differential equations using Gaussian processes (2017, methodology): “Moreover, systems of linear integro-diﬀerential equations can be addressed using multi-output Gaussian process regressions [40, 23, 22].”
Computationally Efficient Convolved Multiple Output Gaussian Processes (2011, background): “In sensor networks, for example, missing signals from failing sensors may be predicted due to correlations with signals acquired from other sensors (Osborne et al., 2008).”
Sparse Convolved Gaussian Processes for Multi-output Regression (2008, methodology): “We followed [12] in simulating sensor failure by introducing some missing ranges for these signals.”
Sequential Bayesian Prediction in the Presence of Changepoints and Faults (2010, methodology): “Following [21], we take a grid of hyperparameter samples θs ×e φu,e, where φu,e is a column vector of unique samples for the eth hyperparameter and× is the Cartesian product.”
Gaussian processes for time-series modelling (2013, unknown): “[2] for more detailed discussions.”
Bounded approximate decentralised coordination via the max-sum algorithm (2009, methodology): “A GP is a principled Bayesian method of performing inference over functions, and have been shown to be very suitable for modelling spatial phenomena [40, 37, 36, 41].”
Remarks on multi-output Gaussian process regression (2018, methodology): “[46] employed the ICM model and parameterized the correlation matrix as A = diag(e)SSdiag(e), where e corresponds to the length scale of each output and S is an upper triangular matrix describing the particular spherical coordinates of points.”
Robust Gaussian Process-Based Global Optimization Using a Fully Bayesian Expected Improvement Criterion (2011, methodology): “An alternative approach using Bayesian quadrature rules [29] has been proposed in [17–19].”
Machine Learning in Wireless Sensor Networks: Algorithms, Strategies, and Applications (2014, background): “3) Assessing accuracy and reliability of sensor nodes us-ing multi-output Gaussian processes: Osborne et al. [127] presented a real-time algorithm to determine a set of nodes that are capable of handling information processing tasks such as assessing the accuracy of collected sensor readings and…”
Inferring solutions of differential equations using noisy multi-fidelity data (2016, background): “An equally important direction involves solving systems of linear partial differential equations, which can be addressed using multi-output GP regression [26,27].”
Multitask Gaussian Processes for Multivariate Physiological Time-Series Analysis (2015, methodology): “In [13], a parametrisation of Kc based on a spherical decomposition was used.”
Gaussian Processes for Global Optimization (2008, methodology): “We use also the sequential formulation of a GP given by [6], a natural fit for our sequential decision problem.”