#### DMCA

## Cooperative learning in multi-agent systems from intermittent measurements

Citations: | 1 - 0 self |

### Citations

1213 | Coordination of Groups of Mobile Autonomous Agents Using Nearest Neighbor Rules
- Jadbabaie, Lin, et al.
- 2003
(Show Context)
Citation Context ... a generalization of earlier results on leader-following, which achieved similar conclusions either without the assumptions of noise, or on fixed graphs, or with the assumption of a fixed leader (see =-=[23]-=-, [43], [38], [40]). Our protocol is very much in the spirit of this earlier literature. All the previous protocols (as well as ours) may be thought of consensus protocols driven by noisy inputs, and ... |

595 | Markov games as a framework for multi-agent reinforcement learning
- Littman
- 1994
(Show Context)
Citation Context ...le of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], [10], [21], [39], [1], [37], [24] as well as the classic works =-=[34]-=-, [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracted some recent attention; we refer the reader to [34... |

367 | The dynamics of reinforcement learning in cooperative multiaent systems
- Claus, Boutilier
- 1998
(Show Context)
Citation Context ...decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], [10], [21], [39], [1], [37], [24] as well as the classic works [34], =-=[19]-=- which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracted some recent attention; we refer the reader to [34], [56... |

364 | Multi-Agent Systems: A Survey from a Machine Learning Perspective
- Stone, Veloso
- 2000
(Show Context)
Citation Context ... some recent attention; we refer the reader to [34], [56], [50]. We make no attempt to survey these literatures here and refer the reader to the references in the above papers, as well as the surveys =-=[54]-=-, [47]. Moreover, we note that much of the recent literature in distributed robotics has focused on distributed algorithms robust to faults and communication link failures. We refer the reader to the ... |

305 | Multi-agent reinforcement learning: Independent vs. cooperative agents
- Tan
- 1993
(Show Context)
Citation Context ...19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracted some recent attention; we refer the reader to [34], =-=[56]-=-, [50]. We make no attempt to survey these literatures here and refer the reader to the references in the above papers, as well as the surveys [54], [47]. Moreover, we note that much of the recent lit... |

255 | Introduction to Optimization - Polyak - 1987 |

228 | Distributed subgradient methods for multi-agent optimization
- Nedic, Ozdaglar
- 2009
(Show Context)
Citation Context ...ysis as in [41]. On the other hand, we introduce a stepsize ∆(t) which we will later choose to decay to zero with t at an appropriate speed by analogy with the recent work on multi-agent optimization =-=[42]-=-, [55], [60]. The use of a stepsize ∆(t) is crucial for the system to be able to successfully learn the unknown vector µ. Intuitively, as t gets large, the nodes should avoid overreacting by changing ... |

192 | The electrical resistance of a graph captures its commute and cover times
- Chandra, Raghavan, et al.
- 1989
(Show Context)
Citation Context ...r. A recent preprint [59] shows that for many graphs, hitting times are proportional to the inverse degrees. In a 2D grid, hitting time is O(n log 2 n) while in the 3D grid hitting time is O(n log n) =-=[15]-=-. We illustrate the convergence times of Theorem 2 with a concrete example. Suppose we have a collection of nodesE[Z(t) | v(0)] ≤ 9lT 2 ( Mσ 2 + nT (σ ′ ) 2) (24T H) 1/ɛ t 2−2ɛ + Z(0)e −((t/T −1)ɛ −1... |

180 | Cooperative Multi-Agent Learning: The State of the Art
- Panait, Luke
- 2005
(Show Context)
Citation Context ...recent attention; we refer the reader to [34], [56], [50]. We make no attempt to survey these literatures here and refer the reader to the references in the above papers, as well as the surveys [54], =-=[47]-=-. Moreover, we note that much of the recent literature in distributed robotics has focused on distributed algorithms robust to faults and communication link failures. We refer the reader to the repres... |

169 | Distributed Kalman filtering for sensor networks
- Olfati-Saber
- 2007
(Show Context)
Citation Context ...to faults and communication link failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], =-=[46]-=-, [52], [3], [53], [13], [35], [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repea... |

151 | Fastest Mixing Markov Chain on a Graph
- Boyd, Diaconis, et al.
- 2004
(Show Context)
Citation Context ...ulti-agent consensus. On the one hand, the weights (1/4)/ max(d(i), d(j)) we accord to neighboring nodes are based on Metropolis weights (first introduced within the context of multi-agent control in =-=[9]-=-) and are chosen because they lead to a tractable Lyapunov analysis as in [41]. On the other hand, we introduce a stepsize ∆(t) which we will later choose to decay to zero with t at an appropriate spe... |

136 |
Distributed Kalman Filter with Embedded Consensus Filters
- Olfati-Saber
- 2005
(Show Context)
Citation Context ...obust to faults and communication link failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering =-=[45]-=-, [46], [52], [3], [53], [13], [35], [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then... |

128 | Tsitsiklis, Distributed averaging algorithms and quantization effects. http://arxiv.org/abs/0803.1202
- Nedić, Olshevsky, et al.
- 2008
(Show Context)
Citation Context ...accord to neighboring nodes are based on Metropolis weights (first introduced within the context of multi-agent control in [9]) and are chosen because they lead to a tractable Lyapunov analysis as in =-=[41]-=-. On the other hand, we introduce a stepsize ∆(t) which we will later choose to decay to zero with t at an appropriate speed by analogy with the recent work on multi-agent optimization [42], [55], [60... |

116 | Multiagent reinforcement learning in the iterated prisoner’s dilemma, iosystems
- Sandholm, Crites
- 1995
(Show Context)
Citation Context ...ich study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracted some recent attention; we refer the reader to [34], [56], =-=[50]-=-. We make no attempt to survey these literatures here and refer the reader to the references in the above papers, as well as the surveys [54], [47]. Moreover, we note that much of the recent literatur... |

83 | Convergence and no-regret in multiagent learning
- Bowling
- 2005
(Show Context)
Citation Context ...which has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], =-=[10]-=-, [21], [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning ... |

82 | Distributed stochastic subgradient projection algorithms for convex optimization
- Ram, Nedic, et al.
- 2010
(Show Context)
Citation Context ...ks in. Every choice of ɛ then provides a tradeoff between the transient size and the asymptotic rate of decay. This is to be contrasted with the usual situation in distributed optimization (see e.g., =-=[49]-=-, [55]) where a specific choice of stepsize usually results in the best bounds. Moreover, the constant in front of asymptotically dominant 1/t 2−2ɛ decay in Theorem 2 scales as some function of the gr... |

63 |
Incremental adaptive strategies over distributed networks
- Lopes, Sayed
- 2007
(Show Context)
Citation Context ...ink failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], =-=[35]-=-, [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through a cons... |

55 | Distributed Kalman filtering based on consensus strategies
- Carli, Chiuso, et al.
- 2008
(Show Context)
Citation Context ...resentative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], [14], [28], [30], [51] and especially =-=[12]-=-. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through a consensus-like iteration. The above-referenced p... |

54 | Bayesian learning in social networks
- Acemoglu, Dahleh, et al.
- 2011
(Show Context)
Citation Context ...d considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], [10], [21], [39], =-=[1]-=-, [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracted som... |

53 | Distributing the kalman filter for large-scale systems
- Khan, Moura
- 2008
(Show Context)
Citation Context ...efer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], [14], [28], =-=[30]-=-, [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through a consensus-like iterati... |

44 | Distributed and collaborative estimation over wireless sensor networks
- Speranzon, Fischione, et al.
- 2006
(Show Context)
Citation Context ...munication link failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], =-=[53]-=-, [13], [35], [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined th... |

41 |
Distributed Kalman Filtering Using Weighted Averaging
- Alriksson, Rantzer
- 2006
(Show Context)
Citation Context ...d communication link failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], =-=[3]-=-, [53], [13], [35], [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combi... |

40 | On synchronous robotic networks - Part I: models, tasks, complexity
- Martinez, Bullo, et al.
- 2007
(Show Context)
Citation Context ...hat much of the recent literature in distributed robotics has focused on distributed algorithms robust to faults and communication link failures. We refer the reader to the representative papers [4], =-=[36]-=-. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], [14], [28], [30], [51] and especially [12]. These works consider t... |

39 | On a stochastic approximation method - Chung - 1954 |

36 | Bounds for eigenvalues of certain stochastic matrices - Landau, Odlyzko - 1981 |

34 | Diffusion strategies for distributed Kalman filtering and smoothing
- Cattivelli, Sayed
- 2010
(Show Context)
Citation Context ...tion link failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], =-=[13]-=-, [35], [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through ... |

31 | Cooperative control and potential games
- Marden, Arslan, et al.
(Show Context)
Citation Context ...tracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], [10], [21], =-=[39]-=-, [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracte... |

29 |
Distributed detection via Gaussian running consensus: Large deviations asymptotic analysis
- Bajovic, Jakovetic, et al.
- 2011
(Show Context)
Citation Context ...obtain bounds on their performance, usually stated in terms of solutions to certain Lyapunov equations. Our work is also related to a number of recent papers on distributed detection [28], [25], [5], =-=[6]-=-, [30], [29], [26], [27] which seek to evaluate protocols for networked cooperative hypothesis testing and related problems. Like the previously mentioned work on distributed filtering, these papers u... |

29 | Maximum hitting time for random walks on graphs - Brightwell, Winkler - 1990 |

29 | Approximate distributed Kalman filtering in sensor networks with quantifiable performance
- Spanos, Olfati-Saber, et al.
(Show Context)
Citation Context ...lts and communication link failures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], =-=[52]-=-, [3], [53], [13], [35], [14], [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly ... |

28 |
Distributed detection over adaptive networks using diffusion adaptation
- Cattivelli, Sayed
- 2011
(Show Context)
Citation Context ...ilures. We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], =-=[14]-=-, [28], [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through a consensus-... |

28 | Distributed asynchronous constrained stochastic optimization
- Srivastava, Nedic
- 2011
(Show Context)
Citation Context ...s in [41]. On the other hand, we introduce a stepsize ∆(t) which we will later choose to decay to zero with t at an appropriate speed by analogy with the recent work on multi-agent optimization [42], =-=[55]-=-, [60]. The use of a stepsize ∆(t) is crucial for the system to be able to successfully learn the unknown vector µ. Intuitively, as t gets large, the nodes should avoid overreacting by changing their ... |

26 | Convergence rate analysis of distributed gossip (linear parameter) estimation: Fundamental limits and tradeoffs
- Kar, Moura
- 2011
(Show Context)
Citation Context ...ds on their performance, usually stated in terms of solutions to certain Lyapunov equations. Our work is also related to a number of recent papers on distributed detection [28], [25], [5], [6], [30], =-=[29]-=-, [26], [27] which seek to evaluate protocols for networked cooperative hypothesis testing and related problems. Like the previously mentioned work on distributed filtering, these papers use the idea ... |

21 |
Leader-following consensus of multi-agent systems under fixed and switching topologies
- Ni, Cheng
- 2010
(Show Context)
Citation Context ...eralization of earlier results on leader-following, which achieved similar conclusions either without the assumptions of noise, or on fixed graphs, or with the assumption of a fixed leader (see [23], =-=[43]-=-, [38], [40]). Our protocol is very much in the spirit of this earlier literature. All the previous protocols (as well as ours) may be thought of consensus protocols driven by noisy inputs, and we not... |

17 | Compositional modeling and analysis of multi-hop control networks
- Alur, D’Innocenzo, et al.
- 2011
(Show Context)
Citation Context ...ote that much of the recent literature in distributed robotics has focused on distributed algorithms robust to faults and communication link failures. We refer the reader to the representative papers =-=[4]-=-, [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], [14], [28], [30], [51] and especially [12]. These works cons... |

17 | An architectural view of game theoretic control
- Gopalakrishnan, Marden, et al.
(Show Context)
Citation Context ...has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], [10], =-=[21]-=-, [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has at... |

16 |
Social interactions, information use, and the evolution of collective migration
- Guttal, Couzin
(Show Context)
Citation Context ...nts of the migration route suggest that the majority of individuals rely on the more accessible observations of the relative motion of their near neighbors when they update their own velocities vi(t) =-=[22]-=-. III. OUR RESULTS We now describe the protocol which we analyze for the remainder of this paper. If at time t node i does not have a measurement of µ, it moves its velocity in the direction of its ne... |

16 | Multi-agent learning for engineers
- Mannor, Shamma
- 2007
(Show Context)
Citation Context ...siderable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], [16], [10], [21], [39], [1], =-=[37]-=-, [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement learning has attracted some rece... |

15 | On stochastic gradient and subgradient methods with adaptive steplength sequences
- Yousefian, Nedic, et al.
(Show Context)
Citation Context ...41]. On the other hand, we introduce a stepsize ∆(t) which we will later choose to decay to zero with t at an appropriate speed by analogy with the recent work on multi-agent optimization [42], [55], =-=[60]-=-. The use of a stepsize ∆(t) is crucial for the system to be able to successfully learn the unknown vector µ. Intuitively, as t gets large, the nodes should avoid overreacting by changing their estima... |

14 | Leaderless and leader-following consensus with communication and input delays under a directed network topology
- Meng, Ren, et al.
(Show Context)
Citation Context ...of earlier results on leader-following, which achieved similar conclusions either without the assumptions of noise, or on fixed graphs, or with the assumption of a fixed leader (see [23], [43], [38], =-=[40]-=-). Our protocol is very much in the spirit of this earlier literature. All the previous protocols (as well as ours) may be thought of consensus protocols driven by noisy inputs, and we note there are ... |

14 |
Hitting and commute times in large graphs are often misleading
- Luxburg, Radl, et al.
(Show Context)
Citation Context ...We first mention the general result that H = O(n 2 ) for an arbitrary connected graph from [44]. On a variety of reasonably connected graphs, hitting times are considerably smaller. A recent preprint =-=[59]-=- shows that for many graphs, hitting times are proportional to the inverse degrees. In a 2D grid, hitting time is O(n log 2 n) while in the 3D grid hitting time is O(n log n) [15]. We illustrate the c... |

12 | Efficient no-regret multiagent learning
- Banerjee, Peng
- 2005
(Show Context)
Citation Context ... classic subject which has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], =-=[8]-=-, [17], [16], [10], [21], [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinf... |

12 | Distributed detection over noisy networks: Large deviations analysis
- Jakovetic, Moura, et al.
- 2012
(Show Context)
Citation Context ...effect and obtain bounds on their performance, usually stated in terms of solutions to certain Lyapunov equations. Our work is also related to a number of recent papers on distributed detection [28], =-=[25]-=-, [5], [6], [30], [29], [26], [27] which seek to evaluate protocols for networked cooperative hypothesis testing and related problems. Like the previously mentioned work on distributed filtering, thes... |

12 |
Context-dependent interaction leads to emergent search behavior in social aggregates
- Torney, Neufeld, et al.
- 2009
(Show Context)
Citation Context ...0], [33]. Our time varying stepsize rule is similar to models of context-dependent interaction in which individuals reduce their reliance on social cues when they are progressing towards their target =-=[57]-=-. We now proceed to set the background for our main result, which bounds the rate at which the estimates vi(t) converge to µ. We first state a proposition which assures us that the estimates vi(t) do ... |

11 | Large deviations performance of consensus+innovations distributed detection with non-gaussian observations,” available on arXiv:1111.4555v2
- Bajovic, Jakovetic, et al.
- 2012
(Show Context)
Citation Context ... and obtain bounds on their performance, usually stated in terms of solutions to certain Lyapunov equations. Our work is also related to a number of recent papers on distributed detection [28], [25], =-=[5]-=-, [6], [30], [29], [26], [27] which seek to evaluate protocols for networked cooperative hypothesis testing and related problems. Like the previously mentioned work on distributed filtering, these pap... |

9 | Gossip and Distributed Kalman Filtering: Weak Consensus under Weak Detectability
- Kar, Moura
- 2011
(Show Context)
Citation Context .... We refer the reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], [14], =-=[28]-=-, [30], [51] and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through a consensus-like i... |

8 |
Oblivious equilibrium: an approximation to large population dynamic games with concave utility
- Adlakha, Johari, et al.
- 2009
(Show Context)
Citation Context ...games is a classic subject which has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers =-=[2]-=-, [7], [8], [17], [16], [10], [21], [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distrib... |

7 | Aspiration learning in coordination games
- Chasparis, Arapostathis, et al.
(Show Context)
Citation Context ...bject which has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], [17], =-=[16]-=-, [10], [21], [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforcement lea... |

7 | On existence of a quadratic comparison function for random weighted averaging dynamics and its implications - Touri, Nedić |

5 |
Note: The hitting and cover times of metropolis walks
- Nonaka, Ono, et al.
- 2010
(Show Context)
Citation Context ...times for various graphs which may be plugged into Theorem 2 to obtain precise topology-dependent estimates. We first mention the general result that H = O(n 2 ) for an arbitrary connected graph from =-=[44]-=-. On a variety of reasonably connected graphs, hitting times are considerably smaller. A recent preprint [59] shows that for many graphs, hitting times are proportional to the inverse degrees. In a 2D... |

4 | Consensus+innovations detection: Phase transition under communication noise
- Jakovetic, Moura, et al.
(Show Context)
Citation Context ...their performance, usually stated in terms of solutions to certain Lyapunov equations. Our work is also related to a number of recent papers on distributed detection [28], [25], [5], [6], [30], [29], =-=[26]-=-, [27] which seek to evaluate protocols for networked cooperative hypothesis testing and related problems. Like the previously mentioned work on distributed filtering, these papers use the idea of loc... |

4 | Consensus control for leader-following multi-agent systems with measurement noises
- Ma, Li, et al.
- 2010
(Show Context)
Citation Context ...ation of earlier results on leader-following, which achieved similar conclusions either without the assumptions of noise, or on fixed graphs, or with the assumption of a fixed leader (see [23], [43], =-=[38]-=-, [40]). Our protocol is very much in the spirit of this earlier literature. All the previous protocols (as well as ours) may be thought of consensus protocols driven by noisy inputs, and we note ther... |

2 | Game couplings: learning dynamics and applications
- Balcan, Constantin, et al.
- 2011
(Show Context)
Citation Context ... is a classic subject which has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], =-=[7]-=-, [8], [17], [16], [10], [21], [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed ... |

2 | Perturbed learning automata for potential games
- Chasparis, Shamma, et al.
- 2011
(Show Context)
Citation Context ...sic subject which has attracted considerable attention within the last couple of decades due in part to its applications to multiagent systems. We refer the reader to the recent papers [2], [7], [8], =-=[17]-=-, [16], [10], [21], [39], [1], [37], [24] as well as the classic works [34], [19] which study multi-agent learning in a game-theoretic context. Moreover, the related problem of distributed reinforceme... |

2 |
Effective leadership and decisionmakingin animal groups on the move,” Nature
- Couzin, Krause, et al.
- 2005
(Show Context)
Citation Context ...t the influence of each successive new sample decays with the stepsize ∆(t). Our protocol is also motivated by models used to analyze collective decision making and collective motion in animal groups =-=[20]-=-, [33]. Our time varying stepsize rule is similar to models of context-dependent interaction in which individuals reduce their reliance on social cues when they are progressing towards their target [5... |

2 |
Fast cooperative distributed learning,”Proceedings
- Jakovetic, Moura, et al.
- 2012
(Show Context)
Citation Context ...performance, usually stated in terms of solutions to certain Lyapunov equations. Our work is also related to a number of recent papers on distributed detection [28], [25], [5], [6], [30], [29], [26], =-=[27]-=- which seek to evaluate protocols for networked cooperative hypothesis testing and related problems. Like the previously mentioned work on distributed filtering, these papers use the idea of local ite... |

2 |
Decision versus compromise for animal groupsin motion
- Leonard, Shen, et al.
- 2012
(Show Context)
Citation Context ...influence of each successive new sample decays with the stepsize ∆(t). Our protocol is also motivated by models used to analyze collective decision making and collective motion in animal groups [20], =-=[33]-=-. Our time varying stepsize rule is similar to models of context-dependent interaction in which individuals reduce their reliance on social cues when they are progressing towards their target [57]. We... |

2 |
Diffusion adaptation over networks,” available at http://arxiv.org/abs/1205.4220
- Sayed
(Show Context)
Citation Context ...he reader to the representative papers [4], [36]. Our work here is very much in the spirit of the recent literature on distributed filtering [45], [46], [52], [3], [53], [13], [35], [14], [28], [30], =-=[51]-=- and especially [12]. These works consider the problem of tracking a time-varying signal from local measurements by each node, which are then repeatedly combined through a consensus-like iteration. Th... |