Data classifier using learning-formed and clustered map

A sensor data classifier comprising:

• mapping means for mapping in a map space a plurality of prototypes that correspond to objects capable of being sensed by one or more sensors, the mapping means adjusting coupling weights between the plurality of prototypes using a plurality of input sensor data and forming a prototype map of the input sensor data through learning;
• and
• clustering means for calculating a predetermined measure defined between pairs of the prototypes and classifying the prototypes into a plurality of clusters based on the predetermined measure, the clusters representing a classification of the input sensor data, wherein
• the prototype map classified into clusters is supplied for a data classification process.

A computer implemented method for classifying sensor data according to one or more sensed characteristics in the sensor data by executing a learning process based on input sensor data and a target comprising organizing learning data using a self-organizing map and classifying the sensed characteristics in the input sensor data using a learning result, the method comprising the steps of:

• forming, through learning, a plurality of self-organizing maps corresponding to a plurality of learning parameters, each of the self-organizing maps providing a corresponding first organization of the input sensor data;
• and
• evaluating each of the plurality of self-organizing maps based on the target and adjusting the learning parameters based on one or more evaluation results, wherein
• a new self-organizing map is formed through learning with adjusted learning parameters, the new self-organizing map providing a corresponding second organization of the input sensor data that facilitates identification of representations of the target in the sensor data.

A computer implemented method for classifying sensor data, comprising:

• mapping in a map space a plurality of prototypes that correspond to objects capable of being sensed by one or more sensors;
• forming, through learning, a prototype map by adjusting coupling weights between the plurality of prototypes according to a plurality of sensor data inputs,
• calculating a predetermined measure defined between pairs of the prototypes and classifying the prototypes into a plurality of clusters based on the predetermined measure, the clusters representing a classification of the sensor data and facilitating identification of representations of the object in the sensor data, and
• supplying the prototype map classified into clusters to a data classification process for classifying representations of the object in the sensor data.

A computer executable medium including a sensor data classifying program which instructs a computer to execute a process to perform learning based on input sensor data using a self-organizing map and to classify the input sensor data according to a target represented by one or more sensed characteristics in the sensor data using results of the learning based on the target comprising organizing learning data, the program instructing the computer to execute the steps of:

• forming, through learning, a plurality of self-organizing maps corresponding to a plurality of learning parameters, each of the self-organizing maps providing a corresponding first organization of the input sensor data;
• and
• evaluating each of the plurality of self-organizing maps based on the target and adjusting the learning parameters based on one or more evaluation results, wherein
• a learning formation process for a new self-organizing map is performed with adjusted learning parameters, the new self-organizing map providing a corresponding second organization of the input sensor data that facilitates identification of representations of the target in the sensor data.

A computer executable medium including a sensor data classifying program which instructs a computer to execute the steps of:

• adjusting in a mapping procedure coupling weights between a plurality of prototypes in a map space using a plurality of input sensor data that includes one or more sensed characteristics and data representing a target;
• forming, through learning, a prototype map;
• a clustering procedure for calculating in clustering procedure a predetermined measure defined between pairs of the prototypes;
• classifying the prototypes into a plurality of clusters based on the predetermined, the clusters representing a classification of the input sensor data;
• and
• supplying the prototype map classified into clusters to a data classification process that facilitates identification of one or more representations of the target in the sensor data.