Near-optimal sensor placements: maximizing information while minimizing communication cost

By Andreas Krause, Carlos Guestrin, Anupam Gupta, Jon M. Kleinberg. 2006

482 citations recorded by the source corpus.

Subfields: Computational Theory and Mathematics, Computer Networks and Communications

How researchers see this paper

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

In Computational Theory and Mathematics

In Computer Networks and Communications

Citations in the corpus

Optimally-Weighted Herding is Bayesian Quadrature (2012, background): “A similar independence is observed in other optimal experimental design problems involving Gaussian processes (Krause et al., 2006).”
Bayesian Active Learning for Classification and Preference Learning (2011, background): “[Guestrin et al., 2005, Krause et al., 2006] also use mutual information.”
Sensor network data fault types (2009, methodology): “Gaussian processes have been used to model the environment for sensor placement [Krause et al. 2006].”
Near-Optimal Sensor Placements in Gaussian Processes: Theory, Efficient Algorithms and Empirical Studies (2008, unknown): “However, we can show that even if we do not specifically optimize for robustness, our sensor placements will be inherently robust: Proposition 11 (Krause et al., 2006) Consider a submodular function F(·) on a ground set S , a set B ⊆ S , and a probability distribution over subsets A of B with the pr”
Near-optimal sensor placements in Gaussian processes (2005, unknown): “However, we can show that even if we do not specifically optimize for robustness, our sensor placements will be inherently robust: Proposition 11 (Krause et al., 2006) Consider a submodular function F (·) on a ground set S, a set B ⊆ S, and a probability distribution over subsets A of B with the pro”
Efficient Informative Sensing using Multiple Robots (2014, background): “Krause et al. (2006) extended this work to include communication cost between sensors while optimizing the sensor placement.”
Robots for Environmental Monitoring: Significant Advancements and Applications (2012, methodology): “[103] present a unified approach based on Gaussian processes for not only determining the placement of senor network nodes to optimize the modeling of processes being observed but also to predict communication costs that are considered in the placement algorithm.”
Submodular Function Maximization (2014, background): “For submodular functions that satisfy an additional locality property, which arises naturally in spatial monitoring problems, improved algorithms can be obtained, both for tree (Krause et al., 2006) and path (Singh et al., 2009) constraints.”
Fast algorithms for maximizing submodular functions (2014, background): “Another one is the problem of intelligent gathering of information, where sensors should be placed in order to maximize the information that can be collected from them [14, 21, 28, 23, 20, 22].”
Near-optimal Observation Selection using Submodular Functions (2007, methodology): “In our recent work (Krause & Guestrin 2005c; Guestrin, Krause, & Singh 2005; Krause et al. 2006; 2007; Singh et al. 2007), we have presented several efficient algorithms for approximately solving these optimization problems .”
Adaptive Active Learning for Image Classification (2013, methodology): “In [10, 17], instances are selected to maximize the increase of mutual information between the selected set of instances and the remaining ones based on Gaussian Process models.”
Computational Sustainability (2011, methodology): “For example, when using wireless networks to monitor spatial phenomena, the selection of the best sensor placement in order to maximize the information gain while minimizing communication costs is a complex optimization problem requiring new solution techniques (Krause et al. 2006, 2008).”
Submodular Optimization with Submodular Cover and Submodular Knapsack Constraints (2013, methodology): “modular function subject to a cardinality equality constraint, and is hence NP hard [36]. Hence this algorithm in the worst case, must be exponential. Furthermore, a similar problem was considered in [25] with one specic instance of a function f, which is not submodular. They also use, a considerabl”
Robust Submodular Observation Selection (2008, methodology): “Krause et al. (2006) describe such an approach, where the cost g(A) measures the expected number of retransmissions required for sending messages across an optimal routing tree connecting the sensors A .”
Towards Real-Time Information Processing of Sensor Network Data Using Computationally Efficient Multi-output Gaussian Processes (2008, methodology): “Examples here include the use of GPs to represent spatial correlations between sensors in order that they may be positioned to maximise mutual information [ 10 ], and the use of multi-variate Gaussians to represent correlations between different sensors and sensor types for energy efficient querying”
Optimal approximation for submodular and supermodular optimization with bounded curvature (2013, background): “The problem of maximizing a submodular function subject to various constraints is a meta-problem that appears in various settings, from combinatorial auctions [30, 12, 35] and viral marketing in social networks [23] to optimal sensor placement in machine learning [26, 27, 28, 25].”