• Jeroen Bédorf

minds.ai's Molecule Property Predictor 'Netrin' optimised with Intel DL libraries

Netrin uses graph convolutional neural networks (GCNN) to predict properties of molecules and in doing so can be applied to drug discovery, potentially speeding up molecule screening and reducing wet lab costs. Training a GCNN for this purpose is a CPU- and memory-intensive process, and this solution brief from Intel describes the accelerated training that is possible with Intel Technologies.


The solution brief can be found here.