org.openimaj.lsh.functions.FloatHammingFactory Maven / Gradle / Ivy

Go to download
/*
	AUTOMATICALLY GENERATED BY jTemp FROM
	/Users/jsh2/Work/openimaj/target/checkout/machine-learning/nearest-neighbour/src/main/jtemp/org/openimaj/lsh/functions/#T#HammingFactory.jtemp
*/
/**
 * Copyright (c) 2011, The University of Southampton and the individual contributors.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification,
 * are permitted provided that the following conditions are met:
 *
 *   * 	Redistributions of source code must retain the above copyright notice,
 * 	this list of conditions and the following disclaimer.
 *
 *   *	Redistributions in binary form must reproduce the above copyright notice,
 * 	this list of conditions and the following disclaimer in the documentation
 * 	and/or other materials provided with the distribution.
 *
 *   *	Neither the name of the University of Southampton nor the names of its
 * 	contributors may be used to endorse or promote products derived from this
 * 	software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
package org.openimaj.lsh.functions;

import org.openimaj.citation.annotation.Reference;
import org.openimaj.citation.annotation.ReferenceType;
import org.openimaj.feature.FloatFVComparison;
import org.openimaj.util.array.SparseFloatArray;

import cern.jet.random.Uniform;
import cern.jet.random.engine.MersenneTwister;

/**
 * A hash function factory for producing hash functions that approximate the
 * Hamming distance.
 * 
 * @author Jonathon Hare ([email protected])
 * 
 */
 @Reference(
	type = ReferenceType.Inproceedings,
	author = { "Indyk, Piotr", "Motwani, Rajeev" },
	title = "Approximate nearest neighbors: towards removing the curse of dimensionality",
	year = "1998",
	booktitle = "Proceedings of the thirtieth annual ACM symposium on Theory of computing",
	pages = { "604", "", "613" },
	url = "http://doi.acm.org/10.1145/276698.276876",
	publisher = "ACM",
	series = "STOC '98"
)
public class FloatHammingFactory extends FloatHashFunctionFactory {
	private class Function extends FloatHashFunction {
		private int ham;

		Function(FloatHammingFactory options, int ndims, MersenneTwister rng) {
			super(rng);

			Uniform uniform = new Uniform(rng);
			
			if (options.bitsPerDim == 0)
				ham = (int) uniform.nextIntFromTo(0, ndims - 1);
			else
				ham = (int) uniform.nextIntFromTo(0, (ndims * options.bitsPerDim) - 1);
		}

		@Override
		public int computeHashCode(float[] point) {
			// which hash function
			if (bitsPerDim == 0) {
				return point[ham] == 0 ? 0 : 1;
			} else {
				// compact binary data
				final int m = ham % bitsPerDim;
				final int d = ham / bitsPerDim;
				return (int) (HammingHelper.convert(point[d]) >>> m & 1L);
			}
		}
		
		@Override
		public int computeHashCode(SparseFloatArray array) {
			// which hash function
			if (bitsPerDim == 0) {
				return array.get(ham) == 0 ? 0 : 1;
			} else {
				// compact binary data
				final int m = ham % bitsPerDim;
				final int d = ham / bitsPerDim;
				return (int) (HammingHelper.convert(array.get(d)) >>> m & 1L);
			}			
		}
	}

	int bitsPerDim;

	/**
	 * Construct a new factory using the given parameters.
	 * 
	 * @param ndims
	 *            The number of dimensions (i.e. length of the vector being
	 *            hashed)
	 * @param rng
	 *            A random number generator
	 * @param bitsPerDim
	 *            The number of bits per dimension. If the data is packed, then
	 *            this will be greater than zero, and internally a single bit
	 *            will be sampled for computing the hash. If zero, then it is
	 *            assumed that every element of the vector being hashed is
	 *            either a zero or one.
	 */
	public FloatHammingFactory(int ndims, MersenneTwister rng, int bitsPerDim) {
		super(ndims, rng);

		this.bitsPerDim = bitsPerDim;
	}

	@Override
	public Function create() {
		return new Function(this, ndims, rng);
	}

	@Override
	public FloatFVComparison fvDistanceFunction() {
		if (bitsPerDim == 0)
			return FloatFVComparison.HAMMING;
		else
			return FloatFVComparison.PACKED_HAMMING;
	}
}