ARTIFICIAL INTELLIGENCE AIDS FOR COMMUNICATION

A SIGNIFICANT NUMBER OF PHYSICALLY DISABLED, NON-VOCAL PEOPLE COULD BENEFIT FROM ACCESS TO COMPUTERS, BUT CANNOT MAKE USE OF A COMPUTER KEYBOARD. BY USING AN EYE TRACKING SYSTEM, THESE INDIVIDUALS CAN PAINSTAKINGLY COMPOSE WORDS AND SENTENCES LETTER BY LETTER BY FOCUSING AN EYE ON ALPHABETIC IMAGES ON A COMPUTER SCREEN. THIS PROCESS IS HOWEVER, TEDIOUSLY SLOW AND PRONE TO CALIBRATION DIFFICULTIES. TO OVERCOME THESE PROBLEMS, WE WILL COUPLE THE EYE TRACKING SYSTEM WITH STATE-OF-THE- ART NEURAL NETWORK SOFTWARE TO ALLOW AUTOMATIC WORD COMPLETION AND WORD PREDICTION, AND SELF-ADJUSTMENT OF THE EYE TRACKER CALIBRATION MECHANISM. NEURAL NETWORKS ARE A VERY RECENT OUTGROWTH OF THE ARTIFICIAL INTELLIGENCE FIELD. THEY OFFER FAULT TOLERANT, ADAPTABLE, PARALLEL COMPUTATION. THEY ARE SELF-ADAPTING TO AN INDIVIDUAL USER AND CAN EVOLVE WITH HIM OVER TIME BY CONTINUOUSLY LEARNING HIS SPEECH PATTERNS. BY PREDICTING ENTIRE WORDS AT ONCE, NEURAL NETWORKS GREATLY INCREASE THE SPEED AND EASE OF COMMUNICATION WITHOUT ARTIFICIALLY RESTRICTING THE USER TO A FIXED VOCABULARY. THE NETWORKS CAN BE SPECIALIZED TO ACCOMMODATE A WIDE RANGE OF ACTIVITIES FROM DAILY COMMUNICATION NEEDS, TO SCHOOLWORK, BUSINESS, CREATIVE WRITING AND COMPUTER PROGRAMMING.