J. G. Cleary and I. H. Witten: "A comparison of enumerative and adaptive codes," IEEE Trans. on Inform. Theory, vol.IT-30, pp.306-315.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....n then go to step 3) Otherwise terminate the algorithm. 3) Set F j (s) F j (s) 1 (j = k; k 1; 1 1 1 ; M ) and u(s) u(s) 1. Then, set s (x i ; s) and i i 1. Go to step 2) 2 Remark 1: The above algorithm is an adaptive arithmetic coding using Laplace s law of succession (see e.g. [13], 14] for the estimation of the generic distribution P of the unifilar source. It should be noted that the variable L is introduced in order to check if the next input symbol can be encoded into the same codeword. Next theorem shows the asymptotic performance of the proposed codes. Theorem 2: ....

J. G. Cleary and I. H. Witten: "A comparison of enumerative and adaptive codes," IEEE Trans. on Inform. Theory, vol.IT-30, pp.306-315.

....deterministic algorithm of the encoder is emulated by the decoder to deduce the state that was selected for coding without any side information about the actual model. In on line modelling, the coding penalty is incorporated into the inaccurate probability estimates from early in the sequence [13]. Thus, refinement coding penalties are incorporated into the records of past performance based on those estimates. Automated experiments with various parametrizations of our executable taxonomy, which evaluate the effect of loose constant lower bounds on the off line coding cost of model ....

Cleary, J. G. and Witten, I. H. (1984) A comparison of enumerative and adaptive codes. IEEE Trans. Inform. Theory, IT-30, 306--315.

....ideal code length (IL) defined in [RL81] for a model structure and individual finite data strings by using the context dependent relative frequencies obtained from a posteriori occurrence counts. A one pass measure, enumerative code length (EL) is similarly defined using a closed form formula [CW84]. Both measures are reviewed in [Lan90] and are defined for a single conditioning context z and 8 bits per symbol (256 values) as follows: IL(z) N(z) log 2 N(z) Gamma 255 X i=0 n(ijz) log 2 n(ijz) 4:1) EL(z) 1 log 2 N(z) Gamma 255 X i=0 (log 2 (n[ijz] Gamma 1) 4:2) In ....

J. Cleary and I. Witten. A comparison of enumerative and adaptive codes. IEEE Trans. on Info. Theory, IT-30(2):306 -- 315, March 1984.

....a misnomer to call this a semi adaptive model since the model adapts throughout the second pass, but we apply the term here since the symbol probability estimates are based primarily on the first pass. The decrementing count idea appears in the analysis of enumerative codes by Cleary and Witten [7]. The resulting code length for a semi adaptive decrementing code is L SD = Gamma lg k Y i=1 c i OE t : 1) It is straightforward to show that for all input files, the code length of a semiadaptive decrementing code is at most that of a semi adaptive static code, equality ....

....the model. If we assume that all symbol distributions are equally likely for a given file length t, the cost of transmitting the exact model statistics is LM = lg(number of possible distributions) lg t n Gamma 1 n Gamma 1 (2) n lg(et=n) A similar result appears in [2] and [7]. For a typical file LM is only about 2560 bits, or 320 bytes. The assumption of equally likely distributions is not very good for text files; in practice we can reduce the cost of encoding the model by 50 percent or more by encoding each of the counts using a suitable encoding of the integers, ....

[Article contains additional citation context not shown here]

J. G. Cleary & I. H. Witten, "A Comparison of Enumerative and Adaptive Codes," IEEE Trans. Inform. Theory IT--30 (Mar. 1984), 306--315.

....of text, all the best methods are adaptive (Bell et al. 1990) An important property of adaptive techniques is that they are very robust. In the limit of long messages, they can guarantee to achieve the entropy of the sequence plus a term that is logarithmic in message size (Bell et al. 1990; Cleary and Witten, 1984). In contrast, the use of predetermined statistics can cause the data to be expanded by an unbounded constant factor. For any particular problem, adaptive algorithms can be constructed in many different ways for computing the complexity, depending on what statistics are accumulated and how they ....

Cleary, J.G. and Witten, I.H. (1984) "A comparison of enumerative and adaptive codes." IEEE Trans Information Theory IT30 (2): 306--315; March.

....model is based upon predicting words rather than single characters. If the word is novel (has not occurred before) then the model escapes down to a character based model similar to PPM. Methods proposed for estimating the escape probability are shown in Table 1. Methods A and B are described in [3, 8]. Method A is based upon Laplace s Law whereas method B still classifies a character as being novel even if it has occurred once before. Method C proposed in [6] uses the number of times a novel character has occurred before as the basis of its probability. Method D [4] is a minor modification to ....

Cleary, J.G. and Witten, I.H. (1984) "A comparison of enumerative and adaptive codes," IEEE Transactions on Information Theory, 30(2), 306--315.

No context found.

Cleary, J.G., and I.H. Witten, "A comparison of enumerative and adaptive codes," IEEE Trans. on Inf. Theory, IT-30, 2, pp. 306-315, Mar. 1984.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC