This paper presents a coded block adaptive neural network system using a radical-partitioned structure for large-volume Chinese characters recognition VLSI. Using the coded block adaptive neural network system with a radical-partitioned structure, 1000 frequently-used Chinese characters have been successfully trained in 139.2 hours using an 18 MIPS computer. According to the simulation results,...

This paper presents several methods to construct trapdoor block ciphers. A trapdoor cipher contains some hidden structure; knowledge of this structure allows an attacker to obtain information on the key or to decrypt certain ciphertexts. Without this trapdoor information the block cipher seems to be secure. It is demonstrated that for certain block ciphers, trapdoors can be built-in that make t...

In this paper, we propose a systematic search method for finding the impossible differential characteristic for block cipher structures, better than the U-method introduced by Kim et al [6]. This method is referred as unified impossible differential (UID) cryptanalysis. We give practical UID cryptanalysis on some popular block ciphers and give the detailed impossible differential characteristic...