Impulse noised outliers are data points that differ significantly from other observations. They are generally removed from the data set through local regression or the Kalman filter algorithm. However, these methods, or their generalizations, are not well suited when the number of outliers is of the same order as the number of low-noise data (often called nominal measurement ). In this article, we propose a new model for impulsed noise outliers. It is based on a hierarchical model and a simple linear Gaussian process as with the Kalman Filter. We present a fast forward-backward algorithm to filter and smooth sequential data and which also detects these outliers. We compare the robustness and efficiency of this algorithm with classical methods. Finally, we apply this method on a real data set from a Walk Over Weighing system admitting around 60 % of outliers. For this application, we further develop an (explicit) EM algorithm to calibrate some algorithm parameters.
Path exits for transhumant livestock are a major problem in many Sub-Saharan African countries. These problems contribute to many community conflicts. Several solutions are currently being studied, including dialogues between stakeholders. In this paper, we propose a numerical approach to address the problem. Based on a commonly accepted model of agent movement, we propose a path simulator to estimate and quantify the risk of exiting the path. This enables quantitative estimation of the exit rates of transhumant animals as a function of the geometric properties of the routes. This model is tested on real transhumance routes in Chad to evaluate the risks of exits along these routes. These new data allow us to better understand the geometric properties on real routes and to evaluate them in terms of exit risk, giving new information to this complex problem. Although our approach does not deal with the whole complexity of this problem, it opens the door to field experimentation with geolocation sensors.
We propose a fine analysis of the various possible long time behaviours of the solutions of the replicator-mutator equation with so-called Kingman's house of cards mutations. In particular, we give what is to our knowledge the first concentration result for this model.
Impulse noised outliers are data points that differs significantly from other observations. They are generally removed from the data set through local regression or Kalman filter algorithm. However, these methods, or their generalizations, are not well suited when the number of outliers is of the same order as the number of low-noise data. In this article, we propose a new model for impulse noised outliers based on simple latent linear Gaussian processes as in the Kalman Filter. We present a fast forward-backward algorithm to filter and smooth sequential data and which also detect these outliers. We compare the robustness and efficiency of this algorithm with classical methods. Finally, we apply this method on a real data set from a Walk Over Weighing system admitting around $60\%$ of outliers. For this application, we further develop an (explicit) EM algorithm to calibrate some algorithm parameters.
In this work, we consider a general time-periodic linear transport equation with integral source term. We prove the existence of a Floquet principal eigenvalue, namely a real number such that the equation rescaled by this number admits nonnegative periodic solutions. We also prove the exponential attractiveness of these solutions. The method relies on general spectral results about positive operators.