Pervasive 3D Graphical User Interface Configured for Machine Learning

Abstract

A three-dimensional graphical user interface (3D GUI) configured to be used by a computer, a display system, an electronic system, or an electro-mechanical system. The 3D GUI provides an enhanced user-engaging experience while enabling a user to manipulate the motion of an object of arbitrary size and a multiplicity of independent degrees of freedom, using sufficient degrees of freedom to represent the motion. The 3D GUI includes the functionality of machine learning (ML) and the support vector machine (SVM) and convolutional neural network (CNN) which provides intelligent control of robot kinematics and computer graphics as well as the ability of the user to more quickly learn the more subtle applications of 3D computer graphics.

We claim: 
     
         1 . A system comprising:
 a memory and at least one processor coupled to the memory in a computer, a display system, an electronic system, or an electro-mechanical system, configured to present on a display device a three-dimensional graphical user interface (3D GUI);   wherein said 3D GUI is configured to allow maneuvering an object in a 3D space represented by said 3D GUI by a motion of at least three independent degrees of freedom, said motion being characterized by either linear or non-linear motion vectors, or both; and   said space being augmented by additional dimensions for characterizing features,   wherein said linear and non-linear motion vectors represent translational and rotational motion respectively and are capable of being generated by a single gestural motion of a navigational device on a reference surface without applying the input of other motion detection devices.   
     
     
         2 . The system of  claim 1  further comprising a neural network module that is loaded into the memory of said system or implemented as a separated device/subsystem, including a graphic processing unit, GPU; or an application specified integrated circuit, ASIC, electronically linking to said system;
 wherein said neural network module carries a specific artificial intelligence function that, through a method of of machine learning or an equivalent learning method, said neural network module is able to classify a plurality of 3D objects presentable by said 3D GUI; 
 wherein at least one property of said 3D objects is identified by said computer or said separated device/subsystem as a feature vector; and wherein 
 the status of said feature vector can be configured by said computer or said separated device/subsystem, making said system able to control an output signal, or the kinematics of an object undergoing a motion. 
 
     
     
         3 . The system of  claim 2 , wherein:
 when in software formation, said neural network module may be incorporated by said system with a plurality of other software modules in a layered configuration;   wherein said software modules are stored in the memory of said system or in said separated device/subsystem, and wherein   each of said software modules is dedicated to a unique functionality of said system, at least two of said unique functionalities are associated to providing the perspectives of said 3D GUI and a robotic kinematics.   
     
     
         4 . The system of  claim 2 :
 wherein said neural network module is characterized as an SVM (support vector machine), CNN (convolutional neural network), or a machine learning method that has a net effect equivalent to that of said SVM or said CNN;   
     
     
         5 . The system of  claim 2 :
 wherein a first exemplary set of said 3D objects, either represented by a plurality of graphical vectors or a set of image data in said 3D GUI, denote a unique group of interactive beings, such beings including a bouquet of flowers configured to interact with a butterfly;   wherein said first exemplary set of 3D objects are classified by said neural network module; and wherein,   using a machine learning process, said first exemplary set of 3D objects are identified by said neural network module as a plurality of distinct species, based on the information provided by said feature vector;   wherein a cursor is configured by said 3D GUI to act as a second set of 3D objects including a plurality of butterflies, configured to interact with said first exemplary set of 3D objects according to some of its feature vectors that are identifiable by said 3D GUI.   
     
     
         6 . The system of  claim 5 , wherein said interactive beings denote a cluster of plants including a bouquet of flowers, a group of animals, including a group of bees, a set of biological entities, including cells in a medical image, or a few typical cartoon characters including Tinker Bell and Winny the Pooh, that have their own personalities or some unique properties identifiable by said 3D GUI. 
     
     
         7 . A computer-implemented method for three dimensional (3D) graphical rendering of objects on a display, comprising the steps of:
 rendering a plurality of three dimensional graphical vectors referenced to at least one vanishing point(s);   wherein a position of said vanishing point(s) can be manipulated by an artificial intelligence technique; and   dividing said plurality of three dimensional graphical vectors tracked by said method into one or more classes, each of which forms a margin with one another that is configured to be recognized by an AI (artificial intelligence) method;   wherein, when said margin reaches different values, said method recognizes that occurrence and generates graphical rendering effects, or supports levels of interaction between a user and said method.   
     
     
         8 . The computer-implemented method of  claim 7  wherein said artificial intelligence method is a support vector machine (SVM). 
     
     
         9 . The computer-implemented method of  claim 7  wherein said artificial intelligence technique is a convolutional neural network (CNN), in which at least one of its output signals is not decided by an optimized value of the margin of support vector machine (SVM). 
     
     
         10 . A computer-implemented neural signal processing system configured to reduce the processing load or time carried out by said computer for classifying a plurality of neural signals while maintaining the accuracy of results of said processing within a user acceptable range, comprising;
 using the classification functionality of a support vector machine (SVM) stored as a module in a 3D GUI, either in hardware or software formation, creating a plurality of multidimensional feature vectors based on a set of raw input data comprising a 3D image, or a 3D vector graphic, or acoustic data in multiple frequency channels, or a vector field, all of whose profiles can be mapped to a 2D image frame;   designating a plurality of vanishing points in said 2D image frame, such that the apparent degrees of freedom of said raw input data after being mapped to said 2D image frame follow a consistent trend of decreasing toward one of said vanishing points,   
       by manipulating the positions of said vanishing points in said 2D image frame automatically, or by an in-situ manual process using 3D navigational device that provides means of changing said 2D image frame by more than three degrees of freedom, the total dimension or size of the vector space constructed by said plurality of multidimensional feature vectors can be manipulated and reduced, which subsequently causes the processing load of said computer in said computer-implemented said neural network system to be reduced correspondingly while the accuracy of result of said neural network system is till maintained at a level acceptable to the user. 
     
     
         11 . A computer-implemented method for neural network signal processing using a computer configured to utilize a three dimensional graphical user interface (3D GUI) shown on a display, said method comprising the steps of;
 using the classification functionality of a support vector machine downloaded in a module in said 3D GUI, creating a plurality of multidimensional feature vectors based on a set of raw input data comprising a 3D image, or a 3D vector graphic, or acoustic data in multiple frequency channels, or a vector field, all of whose profiles can be mapped to a 2D image frame; designating a plurality of vanishing points in said 2D image frame, such that the apparent degrees of freedom of said raw input data after being mapped to said 2D image frame follow a consistent trend of decreasing toward one of said vanishing points,   
       manipulating the positions of said vanishing points in said 2D image frame automatically, or by an in-situ manual process using 3D navigational device that provides means of changing said 2D image frame by more than three degrees of freedom, whereby the total dimension or size of the vector space constructed by said plurality of multidimensional feature vectors can be manipulated and reduced from 3D to 2.5D, which reduction subsequently causes a processing load of said computer in said computer-implemented neural network processing system to be reduced correspondingly while the accuracy of result of said neural network system is till maintained at a level acceptable to the user. 
     
     
         12 . The computer-implemented method of  claim 11 :
 wherein said 3D GUI communicates with a 3D navigational device that is controllably moving along a tinted 2D reference surface, accessing said set of raw input data and, by touching a surface element of said 3D navigational device, altering intensities of a system of illumination within said 3D navigational device thereby changing the 3D address of said vanishing point(s), causing the total dimension or size of said vector space constructed by said plurality of multidimensional feature vectors to be reduced.