applications-of-deepROM

 Phone distance from telecominucation base stations

In this test case we were asked to determine the position of a client phone using only received signal intensity from three telecommunication base stations (fixed antennas). deepROM® was used for this aim.

To be assured of our product’s quality, a combination of TensorFlow and state of the art methods were used to learn the same data.

TensorFlow+State of the art deepROM®
Learning process –         Trial and error approach was used, 50 learning tries

–         Each try the results were analyzed, and modifications were made to the learning method

 

 Launch one learning

 

Validation error (%) 1.93 (best error of 50 tries) 1.82 (one try)
Number of links 22421 740

 

   
State of the art + TensorFlow deepROM®