This paper addressed a statistical analysis for the recall of parallel intraconnected bidirectional associative memory-Modified Intraconnected Bidirectional Associative Memory (MIBAM) and proved the conclusions: two M...This paper addressed a statistical analysis for the recall of parallel intraconnected bidirectional associative memory-Modified Intraconnected Bidirectional Associative Memory (MIBAM) and proved the conclusions: two MIBAM with the equal total number of neurons have the equal recalling probability for m pairs of stored pattern pairs if m is not too large. So they have the same capacity and same error correcting ability, i. e., their performances are statistically equivalent. The results of simulation support the conclusions well.展开更多
A new bidirectional associative memory model named as HOMIBAM is introduced. The relationships of HOMIBAM with the models existed are pointed out. Both theoretical analysis and simulations show that the capacity and r...A new bidirectional associative memory model named as HOMIBAM is introduced. The relationships of HOMIBAM with the models existed are pointed out. Both theoretical analysis and simulations show that the capacity and recall performance of HOMIBAM are superior to that of modified intraconnected BAM (MIBAM), higher-order BAM (HOBAM ) greatly.展开更多
A unified bidirectional associative memory model (UBAM) isproposed- Its two special cases, UHOBAM and UEBAM, are the modifica-tions of intraconnected BAM (IBAM) and higher-order BAM (HOBAM),exponential BAM (EBAM) and ...A unified bidirectional associative memory model (UBAM) isproposed- Its two special cases, UHOBAM and UEBAM, are the modifica-tions of intraconnected BAM (IBAM) and higher-order BAM (HOBAM),exponential BAM (EBAM) and modified exponential BAM (MEBAM) , re-展开更多
In this paper we propose a new discrete bidirectional associative memory (DBAM) which is derived from our previous continuous linear bidirectional associative memory (LBAM). The DBAM performs bidirectionally the opti...In this paper we propose a new discrete bidirectional associative memory (DBAM) which is derived from our previous continuous linear bidirectional associative memory (LBAM). The DBAM performs bidirectionally the optimal associative mapping proposed by Kohonen. Like LBAM and NBAM proposed by one of the present authors,the present BAM ensures the guaranteed recall of all stored patterns,and possesses far higher capacity compared with other existing BAMs,and like NBAM, has the strong ability to suppress the noise occurring in the output patterns and therefore reduce largely the spurious patterns. The derivation of DBAM is given and the stability of DBAM is proved. We also derive a learning algorithm for DBAM,which has iterative form and make the network learn new patterns easily. Compared with NBAM the present BAM can be easily implemented by software.展开更多
Based on Tai’ s high - order bidirectional associative memory ( HOBAM) and Stmposon’ s intraconnected BAM (IBAM) , two Improved models are first presented and discussed in this paper. The improved models not only re...Based on Tai’ s high - order bidirectional associative memory ( HOBAM) and Stmposon’ s intraconnected BAM (IBAM) , two Improved models are first presented and discussed in this paper. The improved models not only retain the advantages of both HOBAM and ISAM but overcome the shortcomings of Kosko’ s BAM also. Secondly their recall stabilities in synchronous and asyn-chronous update modes have been proven by defining corresponding energy functions which decrease as the re-call process proceeds such that the systems can ensure all the training pattern pairs to become local minima of the energy surfaces. Finally with signal - to - noise ratio (SNR) approach, we show that their storage capacities and error correction capabilities are better than that of the HOBAM.展开更多
基金Supported by Climbing Program-National Key Project for Fundamental Research in China
文摘This paper addressed a statistical analysis for the recall of parallel intraconnected bidirectional associative memory-Modified Intraconnected Bidirectional Associative Memory (MIBAM) and proved the conclusions: two MIBAM with the equal total number of neurons have the equal recalling probability for m pairs of stored pattern pairs if m is not too large. So they have the same capacity and same error correcting ability, i. e., their performances are statistically equivalent. The results of simulation support the conclusions well.
基金Supported by Climbing Progamme-National Key Project for Fundamental Research in China
文摘A new bidirectional associative memory model named as HOMIBAM is introduced. The relationships of HOMIBAM with the models existed are pointed out. Both theoretical analysis and simulations show that the capacity and recall performance of HOMIBAM are superior to that of modified intraconnected BAM (MIBAM), higher-order BAM (HOBAM ) greatly.
文摘A unified bidirectional associative memory model (UBAM) isproposed- Its two special cases, UHOBAM and UEBAM, are the modifica-tions of intraconnected BAM (IBAM) and higher-order BAM (HOBAM),exponential BAM (EBAM) and modified exponential BAM (MEBAM) , re-
文摘In this paper we propose a new discrete bidirectional associative memory (DBAM) which is derived from our previous continuous linear bidirectional associative memory (LBAM). The DBAM performs bidirectionally the optimal associative mapping proposed by Kohonen. Like LBAM and NBAM proposed by one of the present authors,the present BAM ensures the guaranteed recall of all stored patterns,and possesses far higher capacity compared with other existing BAMs,and like NBAM, has the strong ability to suppress the noise occurring in the output patterns and therefore reduce largely the spurious patterns. The derivation of DBAM is given and the stability of DBAM is proved. We also derive a learning algorithm for DBAM,which has iterative form and make the network learn new patterns easily. Compared with NBAM the present BAM can be easily implemented by software.
文摘Based on Tai’ s high - order bidirectional associative memory ( HOBAM) and Stmposon’ s intraconnected BAM (IBAM) , two Improved models are first presented and discussed in this paper. The improved models not only retain the advantages of both HOBAM and ISAM but overcome the shortcomings of Kosko’ s BAM also. Secondly their recall stabilities in synchronous and asyn-chronous update modes have been proven by defining corresponding energy functions which decrease as the re-call process proceeds such that the systems can ensure all the training pattern pairs to become local minima of the energy surfaces. Finally with signal - to - noise ratio (SNR) approach, we show that their storage capacities and error correction capabilities are better than that of the HOBAM.
