Neuroprocessing service

An apparatus for providing neuroprocessing service, comprising:

• a user terminal to obtain input from a user;
• and
• a neuroprocessing center including
• a neurocomputer, operatively connected to said user terminal and including a neural network, to execute neuroprocessing to generate a modified graphic pattern in response to a user request when a subset A.sub.0 of a set A consisting of original patterns, is modified according to modification rules by the user into a subset B.sub.0 of a set B, the subset B.sub.0 corresponding element-for-element to the subset A.sub.0, said neurocomputer estimating the modification rules by training the neural network with the subset B.sub.0 as teacher patterns and supplying to the neural network the subset A.sub.0 as input patterns and the subset B.sub.0 as training patterns, and said neurocomputer then applying the modification rules to all of the set A to generate the set B;
• and
• means for supplying from said neuroprocessing center to said user terminal the set B as the modified graphic pattern.

An apparatus for providing neuroprocessing service, comprising:

• a user terminal to obtain input from a user and including a neurocomputer without a learning unit;
• a communication line connected to said user terminal;
• and
• a neuroprocessing center including a neurocomputer having a neural network, and a learning unit for the neural network, said neuroprocessing center connected to said communication line to establish a communication network with said user terminal, to execute neuroprocessing to generate a modified graphic pattern in response to a request by the user supplied by said communication line,
• said neurocomputer in said neuroprocessing center accessible at any time by the user through said communication line to send a result of the neuroprocessing back to said user terminal through said communication line and to share the neuroprocessing between said user terminal and said neuroprocessing center,
• said learning unit of said neurocomputer in said neuroprocessing center learning a learning law as requested by the user, performing a learning operation based on the learning law to produce a result and transmitting the result to said neurocomputer in said user terminal for neuroprocessing using the result of the learning operation,
• said neurocomputer in said neuroprocessing center learning to produce the modified graphic pattern from a set A, including a subset A.sub.0 consisting of original patterns to be modified according to modification rules determined by said neurocomputer using the subset A.sub.0 as input patterns, where a subset B.sub.0 of a set B, which corresponds element-for-element to the subset A.sub.0, is supplied from the user terminal for use as teacher patterns to train the neural network of said neuroprocessing center, and
• said neurocomputer in said user terminal receiving a subset A.sub.1 of the set A consisting of additional patterns other than the original patterns included in the subset A.sub.0 and generating a subset B.sub.1 of modified patterns corresponding element-for-element to the subset A.sub.1 by applying the modification rules.

A neuroprocessing system comprising:

• a neuroprocessing center available as a public facility for a user, to execute neuroprocessing as requested by the user, said neuroprocessing center including
• a neural network;
• and
• means for learning a weight value of said neural network;
• and
• a user terminal comprising a neural network co-operating with said neuroprocessing center to execute the neuroprocessing for generating modified graphic patterns, to transmit training data from said user terminal to said neuroprocessing center to train said neural network in said neuroprocessing center by updating the weight value, said neuroprocessing center sending the weight value to said user terminal value supplied from said neuroprocessing center in said neural for use in executing neuroprocessing to generate the modified graphic patterns.

A method of providing a neuroprocessing service, comprising the steps of:

• (a) inputting a set of original graphic patterns;
• (b) inputting user modified patterns including a training patterns corresponding to a subset of the original graphic patterns;
• (c) estimating at least one modification rule based on differences between the training patterns and the subset of the original graphic patterns;
• and
• (d) applying the at least one modification rule to the original graphic patterns, to obtain modified graphic patterns.

A neuroprocessing system comprising:

• a user terminal for use by a user, to supply training data;
• and
• a neuroprocessing center to receive as the training data a graphic pattern modified by the user, said neuroprocessing center including
• a neural network having weight values and producing an output to the user via said user terminal;
• and
• training means for training said neural network using the training data from said user terminal to obtain an updated weight value of said neural network in accordance with the training data, said neuroprocessing center executing neuroprocessing to generate a modified graphic pattern through said neural network with the updated weight value.

A neuroprocessing system comprising:

• a user terminal for supplying training data;
• a neuroprocessing center for receiving the training data, said neuroprocessing center including
• a neural network having a weight value, for receiving said training data and producing an output to the user via said user terminal;
• and
• training means for training said neural network and for updating the weight value of said neural network in accordance with the training data, said neuroprocessing center executing neuroprocessing for generating a modified graphic pattern through said neural network with the weight value.

A method of providing a neuroprocessing service, comprising the steps of:

• (a) storing a set of original graphic patterns at neuroprocessing;
• (b) inputting user modified patterns, including a training patterns corresponding to a subset of the original graphic patterns, at a user terminal;
• (c) transmitting the training patterns from the user terminal to the neuroprocessing center;
• (d) estimating at least one modification rule based on differences between the training patterns and the subset of the original graphic patterns;
• (e) applying the at least one modification rule to the original graphic patterns, to obtain modified graphic patterns;
• (f) transmitting the modified graphic patterns from the neuroprocessing center to the user terminal;
• (g) training a neural network in the user terminal using the modified graphic patterns;
• and
• (h) identifying the user modified patterns using the neural network trained in step (g).