crystalnet
crystalnet copied to clipboard
crystalnet -- a mini core AI library (being refactored, see https://github.com/lgarithm/stdnn-ops)
An affine operator is the composition of a bilinear operator and an additive operator. y = l(w, x) + b = A(w, b) (x)
In a future (not near so far), the core library will be spitted into ## a library for basic tensor data type - should follow the style of `std::vector` -...
A computation graph is a DAG, where each node is an operation. An operation can only be performed when all of its predecessors are finished. Operations are supposed to be...
* Intel SGX A good [paper](https://css.csail.mit.edu/6.858/2017/readings/costan-sgx.pdf) that introduces SGX * Federated learning [TensorFlow](https://www.tensorflow.org/federated/federated_learning) introduces federated learning recently. * [Federated learning application](https://arxiv.org/pdf/1811.03604.pdf)
## Datasets * [X] mnist * [ ] fashion mnist * [ ] cifar10 * [ ] cifar100 * [ ] tiny imagenet * [ ] imagenet (ILSVRC) * [...
* http://en.cppreference.com/w/cpp/error/exception
*
It should be memory efficient. The element of each iteration should be exactly the same shape, i.e. dynamic batch size will not be supported ATM.