Minimizing Dynamic Remorse and Adaptive Remorse Concurrently
Authors: Lijun Zhang, Shiyin Lu, Tianbao Yang
Summary: Remorse minimization is handled because the golden rule within the conventional examine of on-line studying. Nevertheless, remorse minimization algorithms are inclined to converge to the static optimum, thus being suboptimal for altering environments. To deal with this limitation, new efficiency measures, together with dynamic remorse and adaptive remorse have been proposed to information the design of on-line algorithms. The previous one goals to reduce the worldwide remorse with respect to a sequence of adjusting comparators, and the latter one makes an attempt to reduce each native remorse with respect to a hard and fast comparator. Current algorithms for dynamic remorse and adaptive remorse are developed independently, and solely goal one efficiency measure. On this paper, we bridge this hole by proposing novel on-line algorithms which might be in a position to decrease the dynamic remorse and adaptive remorse concurrently. Actually, our theoretical assure is even stronger within the sense that one algorithm is ready to decrease the dynamic remorse over any interval
