Sequential Classification With Empirically Observed Statistics

Motivated by real-world machine learning applications, we consider a statistical classification task in sequential setting where test samples arrive sequentially. In addition, the generating distributions are unknown and only set of empirically sampled sequences available to decision maker. The maker is tasked classify sequence which known be generated according either one distributions. particular, for binary case, wishes perform with minimum number samples, so, at each step, she declares that hypothesis 1 true, 2 or requests an additional sample. We propose classifier analyze type-I type-II error probabilities. demonstrate significant advantage our scheme compared existing non-sequential proposed Gutman. Finally, extend setup results multi-class scenario again variable-length nature problem affords advantages as can achieve same exponents Gutman’s fixed-length but without having rejection option.