公开(公告)号：WO03012737A2

公开(公告)日：2003-02-13

申请号：PCT/EP0208473

申请日：2002-07-11

Applicant: IBM ,  IBM FRANCE

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL

IPC: G06N3/08 ,  G06K9/62 ,  G06N3/00 ,  G06N3/063

CPC classification number: G06K9/6276 ,  G06N3/063

Abstract: There is disclosed an improved artificial neural network (ANN) (120') comprised of a conventional ANN (120), a database block (220) and a compare & update circuit (230). The conventional ANN is formed by a plurality of n neurons (130), each neuron having a prototype memory (140) dedicated to store a prototype and a distance evaluator (150) to evaluate the distance between the input pattern presented to the ANN and the prototype stored therein. The data base block is comprised of three data bases: a first data base (222) containing all the p prototypes arranged in s slices, each slice being capable to store up to n prototypes, a second data base (224) being capable to store the q input patterns to be presented to the ANN (queries) and a third data base (226) being capable to store the q distances resulting of said evaluation performed during the recognition/classification phase. The role of the compare & update circuit is to compare said distance with the distance previously found for the same input pattern (or pre-existing at initialization) and based upon the result of that comparison, to update or not said distance previously stored.

Abstract translation: 公开了一种由常规ANN（120），数据库块（220）和比较和更新电路（230）组成的改进的人造神经网络（ANN）（120'）。 常规ANN由多个n个神经元（130）形成，每个神经元具有专用于存储原型的原型存储器（140）和距离评估器（150），以评估呈现给ANN的输入模式与 原型存储在其中。 数据库块由三个数据库组成：第一数据库（222），包含以s个片段排列的所有p个原型，每个片段能够存储多达n个原型，第二数据库（224）能够存储 要呈现给ANN（查询）的q个输入模式和第三数据库（226）能够存储在识别/分类阶段期间进行的所述评估的结果。 比较和更新电路的作用是将所述距离与先前针对相同输入模式找到的距离进行比较（或者在初始化时预先存在），并且基于该比较的结果来更新或不更新先前存储的所述距离。

公开(公告)号：JP2001236207A

公开(公告)日：2001-08-31

申请号：JP2000387095

申请日：2000-12-20

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  LOUIS DIDIER ,  TANNHOF PASCAL

IPC: G06F7/02 ,  G06F7/22

Abstract: PROBLEM TO BE SOLVED: To retrieves the minimum/maximum values in the group of numbers. SOLUTION: At first, p numbers encoded on q bits are turned into K (q>=K×n) partial values encoded on n bits, and parameters k (k=1-K) for respectively assigning ranks to the partial value of each number are defined so that K bit slices can be formed, and each slice is constituted of the plural partial values having the same rank. Then, each partial value is encoded on m bits (m>n) by using a 'thermometric' encoding technique. Afterwards, the minimum partial value in the first slice (MSB) of the encoded partial values is decided by a parallel type retrieval, and all the numbers related with the larger partial values than the partial value are selectively released. An evaluation process is repeated in the same configuration until the last slice (LSB) is processed, and the number selected as it is in the final stage is allowed to have the minimum value.

公开(公告)号：CA2351040A1

公开(公告)日：2002-01-13

申请号：CA2351040

申请日：2001-06-19

Applicant: IBM

Inventor： TANNHOF PASCAL ,  IMBERT DE TREMIOLLES GHISLAIN

IPC: G06N3/06 ,  G06K9/62 ,  G06N3/063 ,  G06N3/02

Abstract: The method and circuits of the present invention aim to associate a norm to each component of an input pattern presented to an input space mapping algorithm based artificial neural network (ANN) during the distance evaluation process. The set of norm s, referred to as the "component" norms is memorized in specific memorization means in the ANN . In a first embodiment, the ANN is provided with a global memory, common for all the neurons of the ANN, that memorizes all the component norms. For each component of the input pattern, all the neurons perform the elementary (or partial) distance calculation with the corresponding prototype components stored therein during the distance evaluation process using the associated component norm. The distance elementary calculations are then combined using a "distance" norm to determine the final distance between the input pattern and the prototypes stored in the neurons. In another embodiment, the set of component norms is memorized in t he neurons themselves in the prototype memorization means, so that the global memory is no longer physically necessary. This implementation allows to significantly optimize t he consumed silicon area when the ANN is integrated in a silicon chip.

公开(公告)号：DE60206194T2

公开(公告)日：2006-06-14

申请号：DE60206194

申请日：2002-07-11

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL

IPC: G06N3/00 ,  G06N3/08 ,  G06K9/62 ,  G06N3/063

Abstract: An improved Artificial Neural Network (ANN) is disclosed that comprises a conventional ANN, a database block, and a compare and update circuit. The conventional ANN is formed by a plurality of neurons, each neuron having a prototype memory dedicated to store a prototype and a distance evaluator to evaluate the distance between the input pattern presented to the ANN and the prototype stored therein. The database block has: all the prototypes arranged in slices, each slice being capable to store up to a maximum number of prototypes; the input patterns or queries to be presented to the ANN; and the distances resulting of the evaluation performed during the recognition/classification phase. The compare and update circuit compares the distance with the distance previously found for the same input pattern updates or not the distance previously stored.

公开(公告)号：DE60035171T2

公开(公告)日：2008-02-14

申请号：DE60035171

申请日：2000-11-14

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL ,  LOUIS DIDIER

IPC: G06F7/02 ,  G06F7/22 ,  G06F7/544

Abstract: In the search of the minimum value among a set of p Numbers coded on q bits, each Number is split into K sub-values coded on n bits (q>=Kxn). Parameter K thus assigns a rank to each sub-value so that K slices of bits are formed wherein each slice is composed of sub-values of the same rank. Each sub-value is then encoded on m bits (m>n) using a "thermometric" coding technique. A parallel search is then performed on the first slice of encoded sub-values (MSBs) to determine the minimum sub-value of that slice. All the Numbers associated to sub-values that are greater than the minimum sub-value that has been evaluated are deselected. The evaluation process is continued the same way until the last slice (LSBs) has been processed. At the end of the evaluation process, the Number which remains selected has the minimum value. The response time (i.e. the number of processing steps) now only depends upon the number K of sub-values in which the Numbers have been split up. The method applies to search the maximum as well.

公开(公告)号：DE60035171D1

公开(公告)日：2007-07-26

申请号：DE60035171

申请日：2000-11-14

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL ,  LOUIS DIDIER

IPC: G06F7/02 ,  G06F7/22 ,  G06F7/544

Abstract: In the search of the minimum value among a set of p Numbers coded on q bits, each Number is split into K sub-values coded on n bits (q>=Kxn). Parameter K thus assigns a rank to each sub-value so that K slices of bits are formed wherein each slice is composed of sub-values of the same rank. Each sub-value is then encoded on m bits (m>n) using a "thermometric" coding technique. A parallel search is then performed on the first slice of encoded sub-values (MSBs) to determine the minimum sub-value of that slice. All the Numbers associated to sub-values that are greater than the minimum sub-value that has been evaluated are deselected. The evaluation process is continued the same way until the last slice (LSBs) has been processed. At the end of the evaluation process, the Number which remains selected has the minimum value. The response time (i.e. the number of processing steps) now only depends upon the number K of sub-values in which the Numbers have been split up. The method applies to search the maximum as well.

公开(公告)号：DE60234900D1

公开(公告)日：2010-02-11

申请号：DE60234900

申请日：2002-11-13

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL

IPC: G06T3/40

Abstract: An artificial neural network (ANN) based system that is adapted to process an input pattern to generate an output pattern related thereto having a different number of components than the input pattern. The system (26) is comprised of an ANN (27) and a memory (28), such as a DRAM memory, that are serially connected. The input pattern (23) is applied to a processor (22), where it can be processed or not (the most general case), before it is applied to the ANN and stored therein as a prototype (if learned). A category is associated with each stored prototype. The processor computes the coefficients that allow the determination of the estimated values of the output pattern, these coefficients are the components of a so-called intermediate pattern (24). Assuming the ANN has already learned a number of input patterns, when a new input pattern is presented to the ANN in the recognition phase, the category of the closest prototype is output therefrom and is used as a pointer to the memory. In turn, the memory outputs the corresponding intermediate pattern. The input pattern and the intermediate pattern are applied to the processor to construct the output pattern (25) using the coefficients. Typically, the input pattern is a block of pixels in the field of scaling images.

公开(公告)号：AT453903T

公开(公告)日：2010-01-15

申请号：AT02368124

申请日：2002-11-13

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL

IPC: G06T3/40

Abstract: An artificial neural network (ANN) based system that is adapted to process an input pattern to generate an output pattern related thereto having a different number of components than the input pattern. The system (26) is comprised of an ANN (27) and a memory (28), such as a DRAM memory, that are serially connected. The input pattern (23) is applied to a processor (22), where it can be processed or not (the most general case), before it is applied to the ANN and stored therein as a prototype (if learned). A category is associated with each stored prototype. The processor computes the coefficients that allow the determination of the estimated values of the output pattern, these coefficients are the components of a so-called intermediate pattern (24). Assuming the ANN has already learned a number of input patterns, when a new input pattern is presented to the ANN in the recognition phase, the category of the closest prototype is output therefrom and is used as a pointer to the memory. In turn, the memory outputs the corresponding intermediate pattern. The input pattern and the intermediate pattern are applied to the processor to construct the output pattern (25) using the coefficients. Typically, the input pattern is a block of pixels in the field of scaling images.

公开(公告)号：AT364866T

公开(公告)日：2007-07-15

申请号：AT00480106

申请日：2000-11-14

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL ,  LOUIS DIDIER

IPC: G06F7/02 ,  G06F7/22 ,  G06F7/544

Abstract: In the search of the minimum value among a set of p Numbers coded on q bits, each Number is split into K sub-values coded on n bits (q>=Kxn). Parameter K thus assigns a rank to each sub-value so that K slices of bits are formed wherein each slice is composed of sub-values of the same rank. Each sub-value is then encoded on m bits (m>n) using a "thermometric" coding technique. A parallel search is then performed on the first slice of encoded sub-values (MSBs) to determine the minimum sub-value of that slice. All the Numbers associated to sub-values that are greater than the minimum sub-value that has been evaluated are deselected. The evaluation process is continued the same way until the last slice (LSBs) has been processed. At the end of the evaluation process, the Number which remains selected has the minimum value. The response time (i.e. the number of processing steps) now only depends upon the number K of sub-values in which the Numbers have been split up. The method applies to search the maximum as well.

公开(公告)号：DE60206194D1

公开(公告)日：2005-10-20

申请号：DE60206194

申请日：2002-07-11

Applicant: IBM

Inventor： IMBERT DE TREMIOLLES GHISLAIN ,  TANNHOF PASCAL

IPC: G06N3/08 ,  G06K9/62 ,  G06N3/00 ,  G06N3/063

Abstract: An improved Artificial Neural Network (ANN) is disclosed that comprises a conventional ANN, a database block, and a compare and update circuit. The conventional ANN is formed by a plurality of neurons, each neuron having a prototype memory dedicated to store a prototype and a distance evaluator to evaluate the distance between the input pattern presented to the ANN and the prototype stored therein. The database block has: all the prototypes arranged in slices, each slice being capable to store up to a maximum number of prototypes; the input patterns or queries to be presented to the ANN; and the distances resulting of the evaluation performed during the recognition/classification phase. The compare and update circuit compares the distance with the distance previously found for the same input pattern updates or not the distance previously stored.