Improving floating point compression through binary masks | IEEE Conference Publication | IEEE Xplore