We propose a learning algorithm for residual symbolic finite-state automata
(RSFA) under the minimally adequate teacher model (Angluin, 1987).  MAT learning
algorithms for residual finite-state automata (Denis et al., 2002) and
deterministic symbolic finite state automata (Veanes et al., 2010) have been
proposed in (Bollig et al., 2009) and (Argyros and D'Antoni, 2018),
respectively.  This work extends these works.