Generating a hash value from a vector representing a data object
AssigneeHewlett-Packard Development Company, L.P.
InventorsKave Eshghi, Shyam Sundar Rajaram
Life of this patent
Estimating the term…
Abstract
To generate at least one hash value for a feature vector that represents a data object, a discrete orthogonal transform is applied on a second vector produced from the feature vector. Applying the discrete orthogonal transform on the second vector produces a third vector. At least one value is selected from the third vector to produce the hash value. The at least one hash value is used to perform an action.
Claims19
The A method executed by a computer, comprising:
- generating at least one hash value for a feature vector that represents a data object, wherein generating the at least one hash value comprises:
- applying a permutation to a collection that includes elements in the feature vector to obtain a second vector;
- applying a discrete orthogonal transform on the second vector, wherein applying the discrete orthogonal transform on the second vector produces a third vector;
- and
- selecting at least one value from the third vector to produce the hash value;
- and
- using the at least one hash value to perform an action.
A method executed by a computer, comprising:
- generating at least one hash value for a feature vector that represents a data object, wherein generating the at least one hash value comprises:
- repeating the feature vector multiple times, wherein the elements of the repeated feature vectors make up a collection of elements;
- applying a permutation to the collection of elements to obtain a second vector, wherein repeating the feature vector multiple times before applying the permutation enables a length of the hash value to be increased;
- applying a discrete orthogonal transform on the second vector, wherein applying the discrete orthogonal transform on the second vector produces a third vector;
- and
- selecting at least one value from the third vector to produce the hash value;
- and
- using the at least one hash value to perform an action.
A method executed by a computer, comprising:
- generating hash values for corresponding feature vectors that represent corresponding data objects,
- wherein generating each of the hash values for a respective one of the feature vectors comprises:
- applying a permutation to a collection that includes elements in the feature vector to obtain a second vector;
- applying a discrete orthogonal transform on the second vector to produce a third vector, and
- selecting a value from the third vector, wherein the hash value is generated from the selected value.
An article comprising at least one computer-readable storage medium containing instructions that when executed cause a computer to:
- generate at least one hash value for a feature vector that represents a data object, wherein generating the at least one hash value comprises:
- applying a permutation to a collection that includes elements in the feature vector to obtain a second vector;
- applying a discrete orthogonal transform on the second vector, wherein applying the discrete orthogonal transform on the second vector produces a third vector;
- and
- selecting at least one value from the third vector to produce the hash value;
- and
- using the at least one hash value to perform an action.
An article comprising at least one computer-readable storage medium containing instructions that when executed cause a computer to:
- generate at least one hash value for a feature vector that represents a data object, wherein generating the at least one hash value comprises:
- repeating the feature vector multiple times, wherein the elements of the repeated feature vectors make up a collection of elements;
- applying a permutation to the collection of elements to obtain a second vector, wherein repeating the feature vector multiple times before applying the permutation enables a length of the hash value to be increased;
- applying a discrete orthogonal transform on the second vector, wherein applying the discrete orthogonal transform on the second vector produces a third vector;
- and
- selecting at least one value from the third vector to produce the hash value;
- and
- using the at least one hash value to perform an action.
A system comprising:
- at least one processor to:
- generate at least one hash value for a feature vector that represents a data object, wherein generating the at least one hash value comprises:
- applying a permutation to a collection that includes elements in the feature vector to obtain a second vector;
- applying a discrete orthogonal transform on the second vector, wherein applying the discrete orthogonal transform on the second vector produces a third vector;
- and
- selecting at least one value from the third vector to produce the hash value;
- and
- use the at least one hash value to perform an action.
Dependencies are inferred from the claim text.