Embedded Binarized Neural Networks

TL;DR

This paper introduces embedded Binarized Neural Networks (eBNNs), a memory-efficient approach enabling fast inference on tiny embedded devices by reordering computations and minimizing temporary storage.

Contribution

The paper presents a novel method to reduce memory usage in Binarized Neural Networks for embedded devices by reordering inference and using a single binary temporary, enabling efficient deployment.

Findings

eBNN achieves 95% accuracy on MNIST with only 15 KB memory.

eBNN reduces temporary storage by 32x compared to traditional BNNs.

Inference runtime is under 50 ms on a device with 15 KB memory.

Abstract

We study embedded Binarized Neural Networks (eBNNs) with the aim of allowing current binarized neural networks (BNNs) in the literature to perform feedforward inference efficiently on small embedded devices. We focus on minimizing the required memory footprint, given that these devices often have memory as small as tens of kilobytes (KB). Beyond minimizing the memory required to store weights, as in a BNN, we show that it is essential to minimize the memory used for temporaries which hold intermediate results between layers in feedforward inference. To accomplish this, eBNN reorders the computation of inference while preserving the original BNN structure, and uses just a single floating-point temporary for the entire neural network. All intermediate results from a layer are stored as binary values, as opposed to floating-points used in current BNN implementations, leading to a 32x reduction in required temporary space. We provide empirical evidence that our proposed eBNN approach allows efficient inference (10s of ms) on devices with severely limited memory (10s of KB). For example, eBNN achieves 95\% accuracy on the MNIST dataset running on an Intel Curie with only 15 KB of usable memory with an inference runtime of under 50 ms per sample. To ease the development of applications in embedded contexts, we make our source code available that allows users to train and discover eBNN models for a learning task at hand, which fit within the memory constraint of the target device.