Accelerating Transformers with Fourier-Based Attention for Efficient On-Device Inference | IEEE Conference Publication | IEEE Xplore