It would be possible to traverse the training database once (in parallel) for many input samples. Currently, samples are classified one-by-one, and the training database is traversed each time.
Could probably accomplish this by modifying query_per_sequence.cc (the smart/tedious way: by allowing multiple .jf inputs, put in a heap, keep track as in count_in_file.cc) or loop over samples in the function form_y in Classify.py (dumb/easy way).
It would be possible to traverse the training database once (in parallel) for many input samples. Currently, samples are classified one-by-one, and the training database is traversed each time.
Could probably accomplish this by modifying
query_per_sequence.cc(the smart/tedious way: by allowing multiple.jfinputs, put in a heap, keep track as incount_in_file.cc) or loop over samples in the functionform_yinClassify.py(dumb/easy way).