Information retrieval system

• US 5,400,436 A
• Filed: 12/23/1991
• Issued: 03/21/1995
• Est. Priority Date: 12/26/1990
• Status: Expired due to Term

• Current Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Inventors: Iwamoto, Takashi, Banzhaf, Wolfgang, Nara, Shigetoshi, Ishii, Toshinao
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US07/813,159
• Time in Patent Office: 1,184 Days
• Field of Search: 395/13, 395/21, 395/22, 395/600, 395/934, 395/11, 382/10, 382/14, 382/41
• US Class Current: 706/13
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: G06V 10/75 Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
• CPC Class Codes: Y10S 706/934 Information retrieval or Information management

Abstract

An information retrieval system retrieves stored information on the basis of incomplete or noisy retrieval key information within a realistic processing time. The information to be recalled is stored in neural network associative memory. Genetic algorithms are adopted to avoid a thorough retrieval of data in a huge breadth and depth of high-order space which must otherwise be searched due to ambiguity of retrieval key information. The information containing the given retrieval key is effectively sought. A processor employs a neural network. A memory is correctly recalled from partial inputs or inputs in which noise is extremely pronounced by repeating a sum-of-product operation of a firing pattern within a synapse coupling matrix and also threshold value processing.

Claims

1. An information retrieval system comprising:

a memory means for storing a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with data inputted as retrieval key information;

fetching means for fetching one or more individual pieces of data from said memory means;

operation means, connected to said fetching means, for effecting a genetic operation with respect to said fetched individual piece of data;

processing means, composed of a neural network including a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory, for extracting recall information by receiving said firing pattern as a result of being operated by said operating means, executing a state updating operation of performing a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix and recursively executing said state updating operation till said firing pattern converges in a steady-state; and

evaluation means for giving an evaluation by collating said recall information extracted by said processing means with the retrieval key information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information;

wherein the genetic operation includes a crossover operation, mutational operation, a complementing operation and a reverse operation which are executed in parallel.

2. The system of claim 1 wherein the correlation is an auto-correlation.

3. The system of claim 2 wherein the correlation also includes a cross-correlation.

4. The system of claim 1 wherein the correlation is a cross-correlation.

5. An information retrieval system comprising:

a memory means for storing a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with data inputted as retrieval key information;

fetching means for fetching one or more individual pieces of data from said memory means;

operation means, connected to said fetching means, for effecting a genetic operation with respect to said fetched individual piece of data;

processing means, composed of a neural network including a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory, for extracting recall information by receiving said firing pattern as a result of being operated by said operating means, executing a state updating operation of performing a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix and recursively executing said state updating operation till said firing pattern converges in a steady-state; and

evaluation means for giving an evaluation by collating said recall information extracted by said processing means with the retrieval key information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information;

wherein said genetic operation includes a crossover operation, a mutational operation, a complementing operation and a reverse operation which are sequentially executed.

6. An information retrieval system comprising:

a memory means for storing a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with data inputted as retrieval key information;

fetching means for fetching one or more individual pieces of data from said memory means;

operation means, connected to said fetching means, for effecting a genetic operation with respect to said fetched individual piece of data;

processing means, composed of a neural network including a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory, for extracting recall information by receiving said firing pattern as a result of being operated by said operating means, executing a state updating operation of performing a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix and recursively executing said state updating operation till said firing pattern converges in a steady-state; and

evaluation means for giving an evaluation by collating said recall information extracted by said processing means with the retrieval key information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information;

wherein said genetic operation includes a shift operation of sequentially shifting suffixes of said firing patterns.

7. An information retrieval system comprising:

a memory means for storing a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with data inputted as retrieval key information;

fetching means for fetching one or more individual pieces of data from said memory means;

operation means, connected to said fetching means, for effecting a genetic operation with respect to said fetched individual piece of data;

processing means, composed of a neural network including a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory, for extracting recall information by receiving said firing pattern as a result of being operated by said operating means, executing a state updating operation of performing a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix and recursively executing said state updating operation till said firing pattern converges in a steady-state; and

evaluation means for giving an evaluation by collating said recall information extracted by said processing means with the retrieval key information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information;

wherein the genetic operation includes an operation of incorporating rows in which values are all a single value.

8. An information retrieval method comprising:

(A) the steps of generating a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with a data inputted as a retrieval key information;

(B) fetching one or more pieces of individual data from said individual data set;

(C) effecting a genetic operation with respect to said fetched individual data;

(D) generating a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory;

(E) receiving said firing pattern as a result of undergoing said genetic operation and executing a state updating operation of effecting a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix;

(F) extracting recall information by recursively executing said step (E) till said firing pattern converges in a steady-state;

(G) providing an evaluation by collating said extracted recall information with key retrieval information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information; and

(H) repeating said steps (B) through (G) till said evaluated values of all said individual data fall within an allowable range;

wherein said genetic operation includes a crossover operation, a mutational operation, a complementing operation and a reverse operation are executed in parallel.

9. The system of claims 8 wherein the correlation is an auto-correlation.

10. The system of claim 9 wherein the correlation also includes a cross-correlation.

11. The system of claim 8 wherein the correlation is a cross-correlation.

12. An information retrieval method comprising:

(A) the steps of generating a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with a data inputted as a retrieval key information;

(B) fetching one or more pieces of individual data from said individual data set;

(C) effecting a genetic operation with respect to said fetched individual data;

(D) generating a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory;

(E) receiving said firing pattern as a result of undergoing said genetic operation and executing a state updating operation of effecting a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix;

(F) extracting recall information by recursively executing said step (E) till said firing pattern converges in a steady-state;

(G) providing an evaluation by collating said extracted recall information with key retrieval information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information; and

(H) repeating said steps (B) through (G) till said evaluated values of all said individual data fall within an allowable range;

wherein said genetic operation includes a crossover operation, a mutational operation, a complementing operation and a reverse operation which are sequentially executed.

13. An information retrieval method comprising:

(A) the steps of generating a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with a data inputted as a retrieval key information;

(B) fetching one or more pieces of individual data from said individual data set;

(C) effecting a genetic operation with respect to said fetched individual data;

(D) generating a synapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory;

(E) receiving said firing pattern as a result of undergoing said genetic operation and executing a state updating operation of effecting a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix;

(F) extracting recall information by recursively executing said step (E) till said firing pattern converges in a steady-state;

(G) providing an evaluation by collating said extracted recall information with key retrieval information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information; and

(H) repeating said steps (B) through (G) till said evaluated values of all said individual data fall within an allowable range;

wherein said genetic operation includes a shift operation of sequentially shifting suffixes of said firing patterns.

14. An information retrieval method comprising:

(A) the steps of generating a set of individual data in which a firing pattern indicating a characteristic of information to be retrieved is combined with a value indicating a degree of coincidence of said firing pattern with a data inputted as a retrieval key information;

(B) fetching one or more pieces of individual data from said individual data set;

(C) effecting a genetic operation with respect to said fetched individual data;

(D) generating a syanapse coupling matrix generated by taking a correlation from an aggregation of said firing patterns to be embedded in the form of memory;

(E) receiving said firing pattern as a result of undergoing said genetic operation and executing a state updating operation of effecting a sum-of-product operation and a threshold value process of said firing pattern with said synapse coupling matrix;

(F) extracting recall information by recursively executing said step (E) till said firing pattern converges in a steady-state;

(G) providing an evaluation by collating said extracted recall information with key retrieval information, replacing, if a value of said evaluation is higher than an evaluation value of said fetched individual data, said individual data within said memory means with said recall information and rejecting, if lower than said evaluation value, said recall information; and

(H) repeating said steps (B) through (G) till said evaluated values of all said individual data fall within an allowable range;

wherein said genetic operation includes an operation of incorporating rows in which values are all a single value.