• Use a wide range of image-oriented layer types to implement cutting-edge computer-vision algorithms. »
  • Define network topologies with multiple inputs, outputs, and arbitrary directed acyclic graph connectivity structure. »
  • Work with image, categorical, and numeric inputs and outputs. »
  • Define networks with multiple loss functions to perform multitask learning. »
  • Easily evaluate trained networks using a variety of built-in classifier metrics. »
  • Train on out-of-core image datasets. »
  • Train networks on either CPUs or NVIDIA GPUs. »
  • Take advantage of the NVIDIA CUDA Deep Neural Network library (cuDNN) for optimal GPU performance. »
  • Import and export trained networks as "WLNet" files. »
  • Employ automatic tensor shape inference to write succinct network definitions. »

Related Examples

Related Functions

Related Guides

See Also New in 11

de es fr ja ko pt-br ru zh