Integrating Language and Cognition in Grounded Adaptive Agents

EOARD Grant # 053060

 


 

 

Results:

  1. Scale up of the MFT algorithm (Stage 4 - Using three value for each Concept-Model and three values for each object). See illustration

 

Description:

          In stage one we constructed a series of experiment where the Concept Model is represented by a vector of three numbers and the Object is also represented by a vector of three numbers. (Consider RBG values)

  • Test 1: As a first experiment we used five random generated concept-models (from 0 to 1) and five objects

 

     Click for full size image

Figure 1: The time evolution of five modelling fields using a dynamic fuzziness scheme, categorizing five objects (vector of three).


  • Test 2: As a first experiment we used five random generated concept-models (from 0 to 1) and ten objects

 

     Click for full size image

Figure 2: The time evolution of five modelling fields using a dynamic fuzziness scheme, categorizing ten objects (vector of three).


  • Test 3: As a first experiment we used ten random generated concept-models (from 0 to 1) and ten objects

 

     Click for full size image

Figure 3: The time evolution of ten modelling fields using a dynamic fuzziness scheme, categorizing ten objects (vector of three).

 


  • Test 4: As a first experiment we used ten random generated concept-models (from 0 to 1) and fifty objects

 

     Click for full size image and here for the data

Figure 3: The time evolution of ten modelling fields using a dynamic fuzziness scheme, categorizing fifty objects (vector of three).

 

 

Up - Back