Convolutional Learnable-Group Weightless Neural Network

Weightless Neural Networks (WNNs) based on interconnected Lookup Tables (LUTs) have attracted attention for inference in extremely compact models, but achieving competitive accuracy under such tight resource budgets remains challenging. To address these issues, we introduce the Convolutional Learnable-Group Weightless Neural Network (CLGN). CLGN constructs convolutional layers using LUTs and incorporates a learnable GroupSum connection, thereby enhancing the accuracy of WNNs while maintaining low implementation resource consumption. Moreover, we propose a hierarchical training strategy to improve the training efficiency. We evaluate CLGN in two edge computing scenarios: (1) FPGA, where we evaluate accuracy, latency, throughput, power consumption, LUTs usage, and parameter size; and (2) Microprocessor, where we evaluate latency and memory usage. Compared with the state-of-the-art solutions, the proposed CLGN achieves superior accuracy while maintaining lower implementation resource consumption.

Small devices such as FPGA boards and edge processors often have limited memory and computing resources, making it difficult to run accurate machine learning models on them. This paper introduces CLGN, a compact neural network designed to reduce the resources needed for inference while keeping accuracy high. CLGN represents computation using simple lookup operations and adds a convolution-like structure so that it can better process image and audio data. It also uses a learnable grouping mechanism and a two-stage training method to improve accuracy and training efficiency. We evaluate CLGN on FPGA and microprocessor platforms, measuring accuracy, latency, throughput, power consumption, hardware usage, model size, and memory usage. The results show that CLGN achieves better accuracy than previous weightless neural network methods while using fewer implementation resources.