Online Aggregation of Unbounded Losses Using Shifting Experts with Confidence
This work provides an incremental improvement in online learning algorithms for time-series forecasting, benefiting researchers and practitioners in fields like energy management.
The paper tackles the problem of sequential prediction with shifting experts and unbounded losses by extending AdaHedge with Fixed Share and a smooth version of specialized experts, achieving improved regret bounds in adversarial settings, as demonstrated in numerical experiments for electricity consumption forecasting.
We develop the setting of sequential prediction based on shifting experts and on a "smooth" version of the method of specialized experts. To aggregate experts predictions, we use the AdaHedge algorithm, which is a version of the Hedge algorithm with adaptive learning rate, and extend it by the meta-algorithm Fixed Share. Due to this, we combine the advantages of both algorithms: (1) we use the shifting regret which is a more optimal characteristic of the algorithm; (2) regret bounds are valid in the case of signed unbounded losses of the experts. Also, (3) we incorporate in this scheme a "smooth" version of the method of specialized experts which allows us to make more flexible and accurate predictions. All results are obtained in the adversarial setting -- no assumptions are made about the nature of data source. We present results of numerical experiments for short-term forecasting of electricity consumption based on a real data.