Search

We study the problem of experiment design to learn causal structures from interventional data. We consider an active learning setting in which the experimenter decides to intervene on one of the variables in the system in each step and uses the results of the intervention to recover further causal relationships among the variables. The goal is to fully identify the causal structures with minimum number of interventions. We present the first deep reinforcement learning based solution for the problem of experiment design. In the proposed method, we embed input graphs to vectors using a graph neural network and feed them to another neural network which outputs a variable for performing intervention in each step. Both networks are trained jointly via a Q-iteration algorithm. Experimental results show that the proposed method achieves competitive performance in recovering causal structures with respect to previous works, while significantly reducing execution time in dense graphs. • Active learning of causal structures by training an agent with a deep reinforcement learning algorithm. • The agent is trained by jointly optimizing an embedding network and a score network. • The proposed solution achieves competitive performance compared with previous work while reducing execution time. [ABSTRACT FROM AUTHOR]

We propose two nonlinear regression methods, namely, Adversarial Orthogonal Regression (AdOR) for additive noise models and Adversarial Orthogonal Structural Equation Model (AdOSE) for the general case of structural equation models. Both methods try to make the residual of regression independent from regressors, while putting no assumption on noise distribution. In both methods, two adversarial networks are trained simultaneously where a regression network outputs predictions and a loss network that estimates mutual information (in AdOR) and KL-divergence (in AdOSE). These methods can be formulated as a minimax two-player game; at equilibrium, AdOR finds a deterministic map between inputs and output and estimates mutual information between residual and inputs, while AdOSE estimates a conditional probability distribution of output given inputs. The proposed methods can be used as subroutines to address several learning problems in causality, such as causal direction determination (or more generally, causal structure learning) and causal model estimation. Experimental results on both synthetic and real-world data demonstrate that the proposed methods have remarkable performance with respect to previous solutions. • Two adversarial regression methods making the residual statistically independent of regressors in models with unknown distributions. • Both methods can be utilized in several causal inference tasks by measuring the dependence between residual and regressors. • Proposed regression methods obtain the best fitted curves compared to previous methods in terms of Integral Squared Error (ISE). [ABSTRACT FROM AUTHOR]

Abstract: In this paper, we study the impact of sensing error and channel fading on the decision process of a multiple secondary user network in a primary network whose channel occupancy states are modelled as a Bernoulli process. We present a randomized access strategy to maximize total secondary network throughput. The proposed method guarantees that the probability of collision between primary and secondary users in each channel is less than the predefined value of P c = ξ. To find the optimal access strategy, we formulate secondary network throughput as an optimization problem. Then, using the KKT method to find the solution, we break the original problem into multiple sub-problems. Then, we employ an iterative algorithm and a sifting technique to reduce the number of sub-problems and find the final solution. The solution contains both the SUs access strategies and their detector operating points. In other words, we present a solution for the cross layer design in cognitive radio networks. At the end, we compare our results with two other heuristic strategies (1) equi-probable strategy and (2) weighted strategy and show that the proposed method outperforms them in terms of total throughput. [Copyright &y& Elsevier]

• Proposing two distributed algorithms for multi-choice majority voting in the beep model. • Showing the correctness of the algorithms and analyzing their time, message, and space complexities. • Comparing with the baseline methods analytically. • Evaluating the proposed methods experimentally on different network topologies and initial distributions of votes. We consider the problem of distributed multi-choice voting in a setting that each node can communicate with its neighbors merely by sending beep signals. Given its simplicity, the beep communication model is of practical importance in different applications such as system biology and wireless sensor networks. Yet, the distributed majority voting has not been resolved in this setting. In this paper, we propose two algorithms, named Distributed Voting with Beeps, to resolve this problem. In the first proposed algorithm, the adjacent nodes having the same value form a set called spot. Afterwards, the spots with majority value try to corrode the spots with non-majority values. The second proposed algorithm is based on pairwise interactions between nodes. The proposed algorithms have a termination detection procedure to check whether voting is achieved. We establish theoretical guarantees for the convergence of these algorithms. In particular, we show that the success probability of the first algorithm tends to one as the percentage of the initial votes in majority increases. For the second algorithm, we show that it returns the correct output with high probability. Our experiments show that the algorithms are fairly invariant to the network topology, initial distribution of values, and network size. [ABSTRACT FROM AUTHOR]