Efficient and Mathematically Robust Operations for Certified Neural Networks Inference
This work addresses certification challenges in neural network inference for safety-critical domains like urban air mobility, though it is incremental in focusing on hardware-level optimizations.
The paper tackles the problem of non-associativity in floating-point arithmetic for certified neural network inference, proposing alternative number representations and evaluating algorithms to improve hardware efficiency, with results showing significant gains from fixed-point arithmetic and optimal bit-width determination.
In recent years, machine learning (ML) and neural networks (NNs) have gained widespread use and attention across various domains, particularly in transportation for achieving autonomy, including the emergence of flying taxis for urban air mobility (UAM). However, concerns about certification have come up, compelling the development of standardized processes encompassing the entire ML and NN pipeline. This paper delves into the inference stage and the requisite hardware, highlighting the challenges associated with IEEE 754 floating-point arithmetic and proposing alternative number representations. By evaluating diverse summation and dot product algorithms, we aim to mitigate issues related to non-associativity. Additionally, our exploration of fixed-point arithmetic reveals its advantages over floating-point methods, demonstrating significant hardware efficiencies. Employing an empirical approach, we ascertain the optimal bit-width necessary to attain an acceptable level of accuracy, considering the inherent complexity of bit-width optimization.