run-select

#!/bin/bash
# run: run active learning experiment
# usage: run batchsize1 batchsize2 exp [h]
# 20170829 erikt(at)xs4all.nl

COMMAND="$0"
ID=$1
if [ "$ID" == "" ]; then echo usage: $COMMAND id >&2; fi
LASTID=$(($ID-1))
REVERSE=""
MAKEPROBS=make-probs-learner
BATCHSIZE=110
EXP=a
H=0.5
METHOD="r"
TRAIN=TRAIN
TEST=TEST
REST=REST.unique.0$LASTID
VECTORS=vectors-train-dev-$EXP.vec
TMPFILE=run.$$.$RANDOM
TMPTRAIN=$TMPFILE.train
TMPREST=$TMPFILE.rest
TMPPROBS=$TMPFILE.probs
TMPSTART=$TMPFILE.start
DIM=300
MINCOUNT=5
NBROFCLASSES=12
BINDIR=$HOME/projects/online-behaviour/machine-learning
SELECT=$BINDIR/active-select.py

function make-probs-bagging {
   MAKEPROBSTRAIN=$1
   MAKEPROBSREST=$2
   MAKEPROBSTMP=run.makeprobs.$$.$RANDOM
   TRAINSIZE=`cat $MAKEPROBSTRAIN|wc -l`
   for i in {1..10}
   do
      $SELECT -d $MAKEPROBSTRAIN -z $TRAINSIZE -r -w -h $H > $MAKEPROBSTMP.train
      ../fasttext supervised -input $MAKEPROBSTMP.train -output $MAKEPROBSTMP -dim $DIM \
            -pretrainedVectors $VECTORS -minCount $MINCOUNT >/dev/null 2>/dev/null
      ../fasttext predict-prob $MAKEPROBSTMP.bin $MAKEPROBSREST 1 | sed 's/ .*/ 1.0/' > $MAKEPROBSTMP.probs.$i
   done
   paste -d'#' $MAKEPROBSTMP.probs.* | sed 's/#/ # /g'
   rm -f $MAKEPROBSTMP.bin $MAKEPROBSTMP.vec $MAKEPROBSTMP.probs.* $MAKEPROBSTMP.train
}

function make-probs-learner {
   MAKEPROBSTRAIN=$1
   MAKEPROBSREST=$2
   MAKEPROBSTMP=run.makeprobs.$$.$RANDOM
   TRAINSIZE=`cat $MAKEPROBSTRAIN|wc -l`
   ../fasttext supervised -input $MAKEPROBSTRAIN -output $MAKEPROBSTMP -dim $DIM \
            -pretrainedVectors $VECTORS -minCount $MINCOUNT >/dev/null 2>/dev/null
   ../fasttext predict-prob $MAKEPROBSTMP.bin $MAKEPROBSREST $NBROFCLASSES
   rm -f $MAKEPROBSTMP.bin $MAKEPROBSTMP.vec
}

cat TRAIN ACTIVE-?? > $TMPTRAIN
if [ "$METHOD" == "c" -o "$METHOD" == "e" -o "$METHOD" == "m" ]; then
   $MAKEPROBS $TMPTRAIN $REST > $TMPPROBS
else
   echo -e "\c" > $TMPPROBS
fi
$SELECT -d $REST -p $TMPPROBS -z $BATCHSIZE -a -$METHOD $REVERSE -h $H -x > run-selected-a.0$ID

rm -f $TMPFILE $TMPFILE.??? $TMPFILE.???? $TMPFILE.????? $TMPSTART.bin $TMPSTART.train

exit 0