Active Learning on Non-Stationary Functions
نویسنده
چکیده
We consider the problem of active learning on a function that varies with time. A common approach is to employ a learning algorithm that reduces the contribution of data points as they become older. With this method an active learner will decide to resample regions of the space where all the data has become fairly old. We propose an alternative that calls for the learner to retain all data. As the function varies in time, it may stray away from a certain mapping, but may also return to it at some point in the future. Our algorithm reviews its historical data to see if the current mapping is similar to one held previously and re-uses the old data points when this occurs. Our results show that this can provide better approximation of time-varying functions and can free an active learner from having to resample regions of the space where historical data has become relevant again.
منابع مشابه
Nonlinear and Non-stationary Vibration Analysis for Mechanical Fault Detection by Using EMD-FFT Method
The Hilbert-Huang transform (HHT) is a powerful method for nonlinear and non-stationary vibrations analysis. This approach consists of two basic parts of empirical mode decomposition (EMD) and Hilbert spectral analysis (HSA). To achieve the reliable results, Bedrosian and Nuttall theorems should be satisfied. Otherwise, the phase and amplitude functions are mixed together and consequently, the ...
متن کاملThe Time Adaptive Self Organizing Map for Distribution Estimation
The feature map represented by the set of weight vectors of the basic SOM (Self-Organizing Map) provides a good approximation to the input space from which the sample vectors come. But the timedecreasing learning rate and neighborhood function of the basic SOM algorithm reduce its capability to adapt weights for a varied environment. In dealing with non-stationary input distributions and changi...
متن کاملActive Imitation Learning via Reduction to I.I.D. Active Learning
In standard passive imitation learning, the goal is to learn a target policy by passively observing full execution trajectories of it. Unfortunately, generating such trajectories can require substantial expert effort and be impractical in some cases. In this paper, we consider active imitation learning with the goal of reducing this effort by querying the expert about the desired action at indi...
متن کاملInput Warping for Bayesian Optimization of Non-Stationary Functions
Bayesian optimization has proven to be a highly effective methodology for the global optimization of unknown, expensive and multimodal functions. The ability to accurately model distributions over functions is critical to the effectiveness of Bayesian optimization. Although Gaussian processes provide a flexible prior over functions which can be queried efficiently, there are various classes of ...
متن کاملActive lmitation learning: formal and practical reductions to I.I.D. learning
In standard passive imitation learning, the goal is to learn a policy that performs as well as a target policy by passively observing full execution trajectories of it. Unfortunately, generating such trajectories can require substantial expert effort and be impractical in some cases. In this paper, we consider active imitation learning with the goal of reducing this effort by querying the exper...
متن کامل