JPAT – Joint Proximity Association Template

Beaken Systems And Technology Solutions

Beatrice Braxton
Image Analysis Cloud Image Processing Data Mitigation Remote Sensing

An agnostic means of bidirectional communication between multilayer neural networks for the purpose of drawing out the best parts of each neural and then optimize the collaborative circuit for speed in execution. The Universal Approximation Theorem, by Cybenko and Hornik, states that ‘any multilayer neural network can be ultimately expressed as a single layer’. Our JPAT is this exact realization. We form bidirectional associations between layers and then zip these associations into one single layer. This save orders of magnitude in execution times. Also, there is no other tool that can draw associations between the layers of different networks, which allows for ‘best of layer’ associations among different neural networks; e.g. it may connect the frameworks of Le Net-5 and Contourlets designed for separate multispectral-hyperspectral-panchromatic-LIDAR generated data sets in Remote Sensing

Have something to say? Sign in to join the discussion.