Joe, Sounds good. Regarding tuning the sfrsd2 erasure probabilities for the HF use case, I played around with that some and I reached the conclusion that we should tune the algorithm using statistics derived from vectors that required soft-symbol decoding (as opposed to using the bulk of our data, which can be decoded using BM).
Unfortunately, I have only about 500 such vectors from my data set, which is not really enough to do a good job of filling in the 8x8 probability matrices. Thus, I propose the following --- once you are satisfied with your algorithm for selecting candidates, let’s each use that version to write an s3_hf.bin file using our HF data sets. We can then share those files and concatenate them to combine your data and mine. That should give us enough soft-symbol-decodable vectors upon which to base the iterative self-tuning scheme. Steve > On Oct 5, 2015, at 4:35 PM, Joe Taylor <j...@princeton.edu> wrote: > > Hi STeve, > > I think I understand what's going on with the less-than-perfect > selection of candidate frequencies at which to attempt JT65 decoding. I > hope to spend some time on it in the next couple of days. If I don't > get it sorted out then, it may be delayed for about a week. I'll be > away between Oct 8 and Oct 14. > > -- Joe, K1JT > > ------------------------------------------------------------------------------ > _______________________________________________ > wsjt-devel mailing list > wsjt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wsjt-devel ------------------------------------------------------------------------------ _______________________________________________ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
