Optimal Trees for Searching in Codebook.pdfVIP

,YDQ.RSHþHN )DFXOW\RI,QIRUPDWLFV0DVDU\N8QLYHUVLW\ %RWDQLFNiD%UQR]HFK5HSXEOLF HPDLO: kopecek@fi.muni.cz http://www.fi.muni.cz/~kopecek/ Abstract. Finding nearest neighbour of a given vector in a codebook leads to the following model of searching. In a metric space V, a vector x and a finite subset of vectors S (representing a codebook) are given. We have to find an element of S which is „nearest“ to the element x. In what follows, the problem is formulated more exactly and a characterization of optimal search trees for this model of searching is given. It turns out, that balanced quasi-ternary trees are optimal search trees for the discussed problem. The result enables to speed up finding a codebook representation vector of a given acoustic vector, which is important for applications in speech recognition and synthesis. 9HFWRUTXDQWL]DWLRQFUHDWLRQRIDFRGHERRNDQGWUHHVHDUFKLQWKHFRGHERRNEHORQJWR VWDQGDUGWHFKQLTXHVXVHGLQVSHHFKV\QWKHVLVDQGUHFRJQLWLRQVHHHJ[2], [3], [8]). In what follows, we will formulate the problem of tree searching in a codebook in a slightly generalized way and characterize optimal search trees. ∈ ⊆ /HW9 0GEHDPHWULFVSDFH6DILQLWHVXEVHWRI0DQG[ ; 0 UHVWULFWLRQWR;FDQDYRLGFDVHVZKHQVHDUFKWUHHGRHVQRWH[LVW2XUSUREOHPLVWR PLQLPL]HG[\ZLWKUHVSHFWWRWKHYDULDEOH\LQ6DFRGHERRN/HWXVUHPDUNWKDW WKHSUREOHPFDQDULVHLQVRPHRWKHUDSSOLFDWLRQVVHHHJ@ :HVXSSRVHWKDWZHKDYHFRQVWUXFWHGDURRWHGWUHH5ZKLFK QRGHV DUH HOHPHQWVRI0DQGOHDYH