@techreport{TR-IC-PFG-18-19,
   number = {IC-PFG-18-19},
   author = {Victor de Araujo Velloso and Andre Rodrigues Oliveira and
                   Zanoni Dias},
   title  =  {{Using  Reinforcement  Learning  on Genome Rearrangement
                   Problems}},
   month = {December},
   year = {2018},
   institution = {Institute of Computing, University of Campinas},
   note = {In English, 13 pages.
    \par\selectlanguage{english}\textbf{Abstract}
       Genome   Rearrangement   Problems   compare  different  genomes 
       considering  events  of  mutations affecting a large segment of
       their  DNA structure. Some examples of mutations are deletions,
       insertions,   reversals,   and  transpositions.  To  make  this 
       comparison, genomes are considered as a group of blocks, formed
       by  one  or  more genes, that are conserved between the genomes
       compared.   By   representing  these  blocks  as  numbers,  and 
       considering  that  there  are  no  repeated  blocks,  we obtain
       permutations,  and  the  distance  between  two  genomes is the
       minimum  number  of rearrangements events required to transform
       one  permutation into the other. Depending on the rearrangement
       events  considered this problem becomes $\mathcal{NP}$-hard, so
       finding  approximation  algorithms  with  a low factor to solve
       this  problem  is  of  interest.  In this work, we estimate the
       distance  in  permutations  of length 10 and 15 using different
       Reinforcement  Learning  algorithms.  The  rearrangement events
       considered  here  were reversals and transpositions. We compare
       the  performance and the results of each method with either the
       exact distance or the distance estimated by other approximation
       algorithms in the literature.
  }
}