Method, system, and apparatus for generating and training a digital signal processor for evaluating graph data

Abstract

Embodiments of the present disclosure provide methods, systems, apparatuses, and computer program products for generating, training, and utilizing a digital signal processor (DSP) to evaluate graph data that may include irregular grid graph data. An example DSP that may be generated, trained, and used may include a set of hidden layers, wherein each hidden layer of the set of hidden layers comprises a set of heterogeneous kernels (HKs), and wherein each HK of the set of HKs includes a corresponding set of filters selected from the constructed set of filters and associated with one or more initial Laplacian operators and corresponding initial filter parameters.

1 .- 20 . (canceled) 
     
     
         21 . An apparatus comprising one or more processors and a memory comprising computer-readable instructions, wherein the computer-readable instructions are executable by the one or more processors to cause the apparatus to:
 receive irregular grid graph data; and   generate, using an optimized digital signal processor (DSP), a predicted result based at least in part on the irregular grid graph data, wherein:
 the optimized DSP comprises a set of hidden layers, 
 a hidden layer of the set of hidden layers comprises a set of heterogeneous kernels (HKs), and 
 an HK of the set of HKs comprises a corresponding set of filters selected from a constructed set of filters and associated with one or more initial Laplacian operators and corresponding initial filter parameters. 
   
     
     
         22 . The apparatus of  claim 21 , wherein the irregular grid graph data comprises a set of nodes with one or more node pairs connected by a set of edges in Euclidean space. 
     
     
         23 . The apparatus of  claim 22 , wherein the set of nodes comprise:
 (i) a known graph dataset comprising one or more known graph nodes corresponding to one or more known predictions, and   (ii) an unknown graph dataset comprising one or more unknown graph nodes without one or more corresponding prediction.   
     
     
         24 . The apparatus of  claim 23 , wherein the optimized DSP is previously trained using the known graph dataset. 
     
     
         25 . The apparatus of  claim 23 , wherein the predicted result comprises a machine learning based prediction for an unknown graph node of the unknown graph dataset. 
     
     
         26 . The apparatus of  claim 21 , wherein the corresponding set of filters comprise:
 (i) at least one first filter associated with a K-order Chebyshev filter type,   (ii) at least one second filter associated with a first order renormalized filter type, and   (iii) at least one third filter associated with a K-order topology adaptive filter type.   
     
     
         27 . The apparatus of  claim 21 , wherein the HK is previously generated based at least in part on a weighted combination of the corresponding set of filters. 
     
     
         28 . The apparatus of  claim 21 , wherein the hidden layer of the set of hidden layers is previously generated based at least in part on a weighted combination of the set of HKs. 
     
     
         29 . The apparatus of  claim 21 , wherein the optimized DSP is previously generated by updating the one or more initial Laplacian operators and the corresponding initial filter parameters in an iterative manner to optimize an objective function. 
     
     
         30 . The apparatus of  claim 29 , wherein the objective function is a loss function. 
     
     
         31 . The apparatus of  claim 29 , wherein the objective function is a reward function. 
     
     
         32 . The apparatus of  claim 21 , wherein the optimized DSP further comprises a discriminant layer. 
     
     
         33 . A computer-implemented method comprising:
 receiving, by one or more processors, irregular grid graph data; and   generating, by the one or more processors and using an optimized digital signal processor (DSP), a predicted result based at least in part on the irregular grid graph data, wherein:
 the optimized DSP comprises a set of hidden layers, 
 a hidden layer of the set of hidden layers comprises a set of heterogeneous kernels (HKs), and 
 an HK of the set of HKs comprises a corresponding set of filters selected from a constructed set of filters and associated with one or more initial Laplacian operators and corresponding initial filter parameters. 
   
     
     
         34 . The computer-implemented method of  claim 33 , wherein the corresponding set of filters comprise:
 (i) at least one first filter associated with a K-order Chebyshev filter type,   (ii) at least one second filter associated with a first order renormalized filter type, and   (iii) at least one third filter associated with a K-order topology adaptive filter type.   
     
     
         35 . The computer-implemented method of  claim 33 , wherein the HK is previously generated based at least in part on a weighted combination of the corresponding set of filters. 
     
     
         36 . A non-transitory computer readable medium comprising executable portions configured to:
 receive irregular grid graph data; and   generate, using an optimized digital signal processor (DSP), a predicted result based at least in part on the irregular grid graph data, wherein:
 the optimized DSP comprises a set of hidden layers, 
 a hidden layer of the set of hidden layers comprises a set of heterogeneous kernels (HKs), and 
 an HK of the set of HKs comprises a corresponding set of filters selected from a constructed set of filters and associated with one or more initial Laplacian operators and corresponding initial filter parameters. 
   
     
     
         37 . The non-transitory computer readable medium of  claim 36 , wherein the optimized DSP is previously generated by updating the one or more initial Laplacian operators and the corresponding initial filter parameters in an iterative manner to optimize an objective function. 
     
     
         38 . The non-transitory computer readable medium of  claim 37 , wherein the objective function is a loss function. 
     
     
         39 . The non-transitory computer readable medium of  claim 37 , wherein the objective function is a reward function. 
     
     
         40 . The non-transitory computer readable medium of  claim 36 , wherein the optimized DSP further comprises a discriminant layer.