public static void main(String[] args) throws IOException, Exception {
BasicConfigurator.configure();
if (args.length < 6) {
System.err.println(SYNOPSIS);
System.exit(1);
}
String inf = null;
String ouf = null;
String lif = null;
String inDir = null;
int ufv = 0;
boolean savePartialEstimates = false;
// number of iterations
int n = 10;
// number of cores
int c = Runtime.getRuntime().availableProcessors();
Density.Flags flags = Flags.fAllParams;
for (int i = 0; i < args.length; ++i) {
if (args[i].equals("-i")) inf = args[++i];
else if (args[i].equals("-o")) ouf = args[++i];
else if (args[i].equals("-p")) {
int tc = Integer.parseInt(args[++i]);
// if (tc > c)
// throw new RuntimeException("too many cores requested!");
if (tc > 0) c = tc;
} else if (args[i].equals("-s")) c = -1;
else if (args[i].equals("-n")) n = Integer.parseInt(args[++i]);
else if (args[i].equals("-l")) lif = args[++i];
else if (args[i].equals("-d")) inDir = args[++i];
else if (args[i].equals("--update")) {
String arg = args[++i].toLowerCase();
flags = new Density.Flags(arg.contains("w"), arg.contains("m"), arg.contains("v"));
} else if (args[i].equals("--save-partial-estimates")) {
savePartialEstimates = true;
} else if (args[i].equals("--ufv")) {
ufv = Integer.parseInt(args[++i]);
} else {
System.err.println("Unknown argument: " + args[i]);
System.exit(-1);
}
}
if (inf == null) {
System.err.println("no input file specified");
System.exit(1);
}
if (ouf == null) {
System.err.println("no output file specified");
System.exit(1);
}
if (lif == null) {
System.err.println("no list file specified");
System.exit(1);
}
logger.info("Reading from " + inf + "...");
Mixture initial, estimate;
initial = Mixture.readFromFile(new File(inf));
if (savePartialEstimates) initial.writeToFile(new File(ouf + ".0"));
if (c == -1) {
logger.info("Caching feature data...");
LinkedList<Sample> data = new LinkedList<Sample>();
ChunkedDataSet set = new ChunkedDataSet(new File(lif), inDir, ufv);
ChunkedDataSet.Chunk chunk;
while ((chunk = set.nextChunk()) != null) {
FrameInputStream r = chunk.getFrameReader();
double[] buf = new double[r.getFrameSize()];
while (r.read(buf)) data.add(new Sample((short) 0, buf));
}
logger.info(data.size() + " samples cached");
logger.info("Starting " + n + " EM iterations: single-core, cached data");
estimate = initial;
for (int i = 0; i < n; ++i) {
estimate = Trainer.em(estimate, data);
if (savePartialEstimates) estimate.writeToFile(new File(ouf + "." + (i + 1)));
}
} else {
logger.info("Starting " + n + " EM iterations on " + c + " cores");
ParallelEM pem =
new ParallelEM(initial, new ChunkedDataSet(new File(lif), inDir, ufv), c, flags);
for (int i = 0; i < n; ++i) {
pem.iterate();
if (savePartialEstimates) pem.current.writeToFile(new File(ouf + "." + (i + 1)));
}
estimate = pem.current;
}
logger.info("Saving new estimate...");
estimate.writeToFile(new File(ouf));
}
