Roadmap 2021/2022
drahnr opened this issue · comments
Bigger work items:
- Recursive Neural Networks / Long Term Short Term Memory #87
- OpenCL Backend [juice/coaster-nn]
- Autodiff [juice/coaster-nn]
- Accuracy enum/marker structs instead of type usage [juice/greenglas/coaster]
Low hanging fruits:
- Honour biases [juice] #6
- Gradient calculation for ND-convolution native backend [coaster-nn]
- Regression examples [juice-examples]
- Add preprocessing filters [greenglas]
The full presentation of November 30th, 2017:
bernhard_schuster__rust_machine_learning_with_juice.pdf
So 2018 was a pretty busy year so far and I am about to change my job, so not as much as I would have wished happened so far.
Another thing I've been eyeing lately is this device
https://www.aaeon.com/en/p/iot-gateway-maker-boards-ai-core using its C API https://movidius.github.io/ncsdk/ncapi/ncapi2/c_api/readme.html
If somebody is interested I am happy to wrap the SDK in some rust, it would be awesome to integrated this (or something similiar) into coaster-nn.
2018 long passed, 2019 has gone by too without much action but the work items are still the same :)