Fast, Scalable, Energy-Efficient Non-element-wise Matrix Multiplication on FPGA

Modern Neural Network (NN) architectures heavily rely on vast numbers of multiply-accumulate arithmetic operations, constituting the predominant computational cost. Therefore, this paper proposes a high-throughput, scalable and energy efficient non-element-wise matrix multiplication unit on FPGAs as a basic component of the NNs. We firstly streamline inter-layer and intra-layer redundancies of MADDNESS algorithm, a LUT-based approximate matrix multiplication, to design a fast, efficient scalable approximate matrix multiplication module termed "Approximate Multiplication Unit (AMU)". The AMU optimizes LUT-based matrix multiplications further through dedicated memory management and access design, decoupling computational overhead from input resolution and boosting FPGA-based NN accelerator efficiency significantly. The experimental results show that using our AMU achieves up to 9x higher throughput and 112x higher energy efficiency over the state-of-the-art solutions for the FPGA-based Quantised Neural Network (QNN) accelerators.

A Brief Survey of Recent Edge-Preserving Smoothing Algorithms on Digital Images

Edge preserving filters preserve the edges and its information while blurring an image. In other words they are used to smooth an image, while reducing the edge blurring effects across the edge like halos, phantom etc. They are nonlinear in nature. Examples are bilateral filter, anisotropic diffusion filter, guided filter, trilateral filter etc. Hence these family of filters are very useful in reducing the noise in an image making it very demanding in computer vision and computational photography applications like denoising, video abstraction, demosaicing, optical-flow estimation, stereo matching, tone mapping, style transfer, relighting etc. This paper provides a concrete introduction to edge preserving filters starting from the heat diffusion equation in olden to recent eras, an overview of its numerous applications, as well as mathematical analysis, various efficient and optimized ways of implementation and their interrelationships, keeping focus on preserving the boundaries, spikes and canyons in presence of noise. Furthermore it provides a realistic notion for efficient implementation with a research scope for hardware realization for further acceleration.