| 摘要 |
The invention relates to a computer system for data analysis. The aim of the invention is to significantly reduce the training time by means of technical measures and to considerably reduce the storage space required by the implementation of said technical measures. To achieve this, the invention discloses an electronic data processing system for analysing data, comprising at least one analysis server and at least one in-situ client computer. The analysis server is designed and programmed to implement a self-adapting neural network that can be trained in a database comprising a plurality of data records with many attributes. The in-situ client computer is designed to subject supplied data to data processing and/or data compression before the data is transmitted by the in-situ client computer to the analysis server via an electronic network. The analysis server is designed and programmed to train the self-adapting neural network using the received pre-processed/compressed data, whereby said data is iteratively presented to the self-adapting neural network and an analysis is then carried out in order to create a self-adapting neural network model. The analysis server is further designed and programmed to initiate the sending of the self-adapting neural network model from the analysis server to the in-situ client computer. Finally, the in-situ client computer is designed and programmed to subject the data of the self-adapting neural network model to decompression. |
