Typesafe Abstractions for Tensor Operations

TL;DR

This paper introduces a typesafe abstraction for tensor operations using Scala's type-level programming, enabling safer neural network layer implementations and laying groundwork for future deep learning frameworks on JVM.

Contribution

It presents a novel typesafe tensor abstraction leveraging Scala's HList, facilitating safer neural network layer implementations and potential deep learning framework development.

Findings

Typesafe tensor operations demonstrated with neural layers

Potential foundation for JVM-based deep learning frameworks

Enhanced safety in tensor computations

Abstract

We propose a typesafe abstraction to tensors (i.e. multidimensional arrays) exploiting the type-level programming capabilities of Scala through heterogeneous lists (HList), and showcase typesafe abstractions of common tensor operations and various neural layers such as convolution or recurrent neural networks. This abstraction could lay the foundation of future typesafe deep learning frameworks that runs on Scala/JVM.