Description
Much like PyTorch, our MPC AI/ML framework nada-ai implements reusable standard AI modules. Pooling operations are one of these modules.
Average pooling is already implemented here but max pooling is not.
Use Case
Max pooling is a very common operation in AI models - particularly computer vision models.
Specific Requirements
Our work on Average pooling should already provide solid template to performing kernel-style operations.
The bit that would need to change is the actual operation that the kernel performs (average -> max) which can be found here.
Description
Much like PyTorch, our MPC AI/ML framework nada-ai implements reusable standard AI modules. Pooling operations are one of these modules.
Average pooling is already implemented here but max pooling is not.
Use Case
Max pooling is a very common operation in AI models - particularly computer vision models.
Specific Requirements
Our work on Average pooling should already provide solid template to performing kernel-style operations.
The bit that would need to change is the actual operation that the kernel performs (average -> max) which can be found here.