Roadmap

Core

The list of items shown below are likely to be implemented due to their mainstream use, ability to increase learning speed, or ability to reduce computational resources.

• None

Nice-To-Have

The list of items shown below may not necessarily be implemented in the future. However, they could be prioritized with external demand, collaboration, or funding.

• Dilated Convolution Layers And Pooling Layers

  • Enables larger receptive field without more weight parameters.

  • Good in sparse-data settings.

  • Unknown use cases related to game environments.

• Generalized N-Dimensional Convolution Layers And Pooling Layers

  • Currently, we have up to 3 dimensional kernels.

  • Useful for pushing the boundaries of convolutional neural networks.

  • 4 dimensional kernels are used in videos. Unknown use cases for game environments.

• Less Bloated Function Blocks Design

  • Currently, function blocks’ differentiate() function have excessive amount of code being used. Additionally, we have suspicions that our initial code design decision might not have led to efficient backward propagation calculation.

  • However, the current design enable model parallelism and data parallelism. As such, we are debating or not if there are trade-off between code design (and its backward propagation calculation speed) and parallelism flexibility.