| public Build ( MetricDB db, int symbolsPerHash ) : void | ||
| db | MetricDB | |
| symbolsPerHash | int | |
| return | void | |
public static Parameters EstimateParameters(MetricDB db, int max_instances, int k, double expected_recall, int num_estimation_queries)
{
var seq = new Sequential ();
seq.Build (db);
var I = new NeighborhoodHash ();
int symbolsPerHash = 3;
I.Build (db, symbolsPerHash);
var Q = RandomSets.GetRandomSubSet (num_estimation_queries, db.Count);
// k > 1 since Q is a subset of the database
if (k == 1) {
++k;
}
++k;
var res_array = new HashSet<int> [Q.Length];
for (int i = 0; i < Q.Length; ++i) {
var s = KnrFP.GetFP (db [Q [i]], seq, k);
res_array [i] = new HashSet<int> (s);
}
int num_instances = 0;
--I.NeighborhoodExpansion;
double cost = 0.0;
double time = 0.0;
do {
++I.NeighborhoodExpansion;
var c = db.NumberDistances;
var t = DateTime.Now.Ticks;
num_instances = _EstimateParameters(k, expected_recall, I, Q, res_array);
cost = (db.NumberDistances - c) / Q.Length * num_instances;
time = TimeSpan.FromTicks((DateTime.Now.Ticks - t) / Q.Length).TotalSeconds * num_instances;
Console.WriteLine("==== expansion: {0}, num_instances: {1}, search-cost: {2}, search-time: {3}", I.NeighborhoodExpansion, num_instances, cost, time);
} while (num_instances > max_instances);
return new Parameters() {
Index = I,
NumberOfInstances = num_instances
};
}
/// <summary>
/// Creates an index for db using the specified number of instances.
/// </summary>
public void Build(MetricDB db, Parameters uparams)
{
var seed = RandomSets.GetRandomInt ();
this.A = new NeighborhoodHash[uparams.NumberOfInstances];
this.DB = db;
this.A [0] = uparams.Index;
for(int i = 1; i < uparams.NumberOfInstances; ++i) {
Console.WriteLine ("==== creating {0}/{1} instances", i + 1, uparams.NumberOfInstances);
var I = new NeighborhoodHash ();
I.Build(db, uparams.Index.SymbolsPerHash, uparams.Index.NeighborhoodExpansion, new Random(seed + i));
this.A [i] = I;
}
}