RecLight: A Recurrent Neural Network Accelerator with Integrated Silicon Photonics
This addresses the need for efficient RNN acceleration in real-time scenarios like speech recognition, though it appears incremental as it builds on existing photonic accelerator concepts.
The paper tackled the problem of accelerating RNN inference for real-time applications by proposing RecLight, a photonic hardware accelerator, which achieved 37x lower energy-per-bit and 10% better throughput compared to state-of-the-art methods.
Recurrent Neural Networks (RNNs) are used in applications that learn dependencies in data sequences, such as speech recognition, human activity recognition, and anomaly detection. In recent years, newer RNN variants, such as GRUs and LSTMs, have been used for implementing these applications. As many of these applications are employed in real-time scenarios, accelerating RNN/LSTM/GRU inference is crucial. In this paper, we propose a novel photonic hardware accelerator called RecLight for accelerating simple RNNs, GRUs, and LSTMs. Simulation results indicate that RecLight achieves 37x lower energy-per-bit and 10% better throughput compared to the state-of-the-art.