Non-negative isomorphic neural networks for photonic neuromorphic accelerators
This work addresses a hardware-specific bottleneck for deploying AI models on energy-efficient photonic accelerators, representing an incremental improvement in domain-specific optimization.
The paper tackles the problem of deploying deep learning models on neuromorphic photonic accelerators, which cannot natively represent negative quantities, by introducing a methodology to obtain non-negative isomorphic equivalents of regular neural networks and a sign-preserving optimization approach, achieving comparable accuracy to regular networks.
Neuromorphic photonic accelerators are becoming increasingly popular, since they can significantly improve computation speed and energy efficiency, leading to femtojoule per MAC efficiency. However, deploying existing DL models on such platforms is not trivial, since a great range of photonic neural network architectures relies on incoherent setups and power addition operational schemes that cannot natively represent negative quantities. This results in additional hardware complexity that increases cost and reduces energy efficiency. To overcome this, we can train non-negative neural networks and potentially exploit the full range of incoherent neuromorphic photonic capabilities. However, existing approaches cannot achieve the same level of accuracy as their regular counterparts, due to training difficulties, as also recent evidence suggests. To this end, we introduce a methodology to obtain the non-negative isomorphic equivalents of regular neural networks that meet requirements of neuromorphic hardware, overcoming the aforementioned limitations. Furthermore, we also introduce a sign-preserving optimization approach that enables training of such isomorphic networks in a non-negative manner.