Despite this constraint accuracy of the constrained (TrueNorth) model that deployed on-chip converges to the accuracy of the accuracy of an unconstrained (software) model after ~500 epochs for the detection network, as shown in Fig. To achieve this on neuromorphic hardware, the training model constrains the coefficients of individual neurons to. In this work, we demonstrate for the first-time, spinal image segmentation performed using a deep learning network implemented on neuromorphic hardware of the IBM TrueNorth Neurosynaptic System and validate the performance of our network by comparing it to human-generated segmentations of spinal vertebrae and disks. Adapting deep learning models to execute image segmentation tasks on such chips, requires specialized training and validation. However, given that they closely emulate the human brain, on-chip hardware and digital memory limitations also constrain them. Neuromorphic processors that replicate the structure of the brain are better-suited to train and execute deep learning models for image segmentation by relying on massively-parallel processing. These algorithms are executed on conventional von Neumann processor architectures or GPUs. Abstract Deep convolutional neural networks have found success in semantic image segmentation tasks in computer vision and medical imaging.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |