The following information was submitted:
Transactions: INTERNATIONAL JOURNAL of CIRCUITS, SYSTEMS and SIGNAL PROCESSING
Transactions ID Number: 19-547
Full Name: Shunsuke Kobayakawa
Position: Doctor (Researcher)
Age: ON
Sex: Male
Address: 102 Kimiegakkendairejidensu 3-13-1 Shioya, Wakamatsu-ku, Kitakyushu-shi, Fukuoka 808-0131
Country: JAPAN
Tel: +81-80-5204-2641
Tel prefix:
Fax:
E-mail address: s-kobayakawa@live.jp
Other E-mails: valiable@live.jp
Title of the Paper: Evaluation for Independent Quantization Learning Predictive Coding using Electrocardiogram
Authors as they appear in the Paper: Shunsuke Kobayakawa and Hirokazu Yokoi
Email addresses of all the authors: s-kobayakawa@live.jp, yokoi@life.kyutech.ac.jp
Number of paper pages: 11
Abstract: This paper is presented a method for improving predictive coding. The method is the independent quantization predictive coding as its two predictors are learning. Its coding process is characterized in independently processing quantizations of an original series signal and a prediction series signal to eliminate quantization errors. It is performed to reduce prediction error as the predictors using error-convergence neuron network are learning. The method is the lossless data compression with the highest compression ratio, if quantization step size for an original series signal is the same as one when the signal was obtained. Then, computer simulations to evaluate its compression ratio were executed for a normal sinus rhythm electrocardiogram with using input-delay second-order Volterra neuron networks for neuron networks in an error-convergence neuron network predictor. As a result, the compression ratio was 1.71. In addition, an obtained quantization error series!
signal is more compressed with cabinet. Its compression ratio was 2.02. This method can be expected to perform excellent predictive coding for every signal with functional relationships between inputs and a prediction.
Keywords: Accuracy, Electrocardiogram, Lossless data compression, Neuron network, Predictive coding
EXTENSION of the file: .pdf
Special (Invited) Session: Proposal of Independent Quantization Predictive Coding as Learning
Organizer of the Session: 201-459
How Did you learn about congress:
IP ADDRESS: 203.114.211.232
