Search papers, labs, and topics across Lattice.
This paper introduces a novel convolutional block based on grouped convolutions to improve the scalability of LUT-based precomputed 1D-CNNs for deployment on FPGAs. They propose an algorithm to guide hyperparameter selection for this block, optimizing resource efficiency. Evaluated on atrial fibrillation detection using ECG recordings from the MIT-BIH database, the resulting hardware accelerators achieve a F1-Score of up to 95% while requiring minimal resources (2,844 LUTs, no DSPs or BRAM) on an AMD Spartan 7 S15.
Achieve 95% F1-score in atrial fibrillation detection with a tiny 1D-CNN accelerator that fits on a low-end FPGA, using only LUTs and no DSPs or BRAM.
1D-CNNs play a crucial role for time-series analysis on tiny smart sensor systems, e.g. for biosignal analysis, predictive maintenance, or structural health monitoring. LUTbased precomputation has emerged as an interesting optimization technique to implement such neural networks on FPGAs. The core idea is to precompute all possible outputs of a neural network layer and store them directly in the lookup tables of the FPGAs. This enables highly resource-efficient networks with ultra-low latency but suffers from poor scalability. Previous work has explored using depthwise-separable convolutions to improve scalability. In this paper, we generalize this approach to consider additional forms of grouped convolutions. Based on this, we propose a novel type of convolutional block and an algorithm to guide the choice of hyper parameters for this block. We evaluate our approach on a medical time-series dataset for predicting atrial fibrillation using the MIT-BIH database (ECG recordings). The resulting hardware accelerators are small enough to be deployed on an AMD Spartan 7 S15. They achieve a F1-Score of up to 95% while only requiring 2,844 LUTs and no DSPs or BRAM.
