83.txt

This complete matlab for neural network

发信人: fervvac (高远), 信区: DataMining
标  题: Re: Data compression一问
发信站: 南京大学小百合站 (Wed Apr  9 13:46:17 2003)

See 

Approximate Computation of Multidimensional Aggregates of Sparse Data Using W
avelets (J. S. Vitter and M. Wang), Proceedings of the 1999 ACM SIGMOD Intern
ational Conference on Management of Data (SIGMOD 1999), Philadelphia, June 19
99, pp.193-204.

http://www.cs.duke.edu/%7Eminw/Publication/sigmod99.pdf

Basically, they choose a set of coefficiency from all coefficiencies under
a space budget. They try to optimaize the process by minimizing some
error fucntions. But I don't think they have theoretical assurance of the
fidelity of the compression.


btw, r u working on data compression?

【 在 GeminiGirl (双子MM) 的大作中提到: 】
: Apply the DWT to the array
: {0,1,0,4,3,4,4,0,0,1,5,5,5,6,4,0}
: L1: 0.5,2,3.5,2,0.5,5,5.5,2 H1: -0.5,-2,-0.5,2,-0.5,0,-0.5,2
: L2: 1.25,2.75,2.75,3.75 H2: -0.75,0.75,-2.25,1.75
: L3: 2,3.25 H3: -0.75,-0.5
: L4: 2.625 H4: -0.625
: Am I right?
: What should be the final result?
: 2.625,-0.625,-0.75,-0.5,-0.75,0.75,-2.25,1.75,-0.5,-2,-0.5,2,-0.5,0,-0.5,2
: or only
: 2.625, -0.625
: 
: If it should be the first, how can we "compress" the data?
: If the second, why should we count the high pass values of Hi?


--
※ 修改:．fervvac 于 Apr  9 13:49:57 修改本文．[FROM: 143.89.43.15]
※ 来源:．南京大学小百合站 bbs.nju.edu.cn．[FROM: 143.89.43.15]