A Hard Voting Policy-Driven Deep Learning Architectural Ensemble Strategy for Industrial Products Defect Recognition and Classification

Abstract

Manual or traditional industrial product inspection and defect-recognition models have some limitations, including process complexity, time-consuming, error-prone, and expensiveness. These issues negatively impact the quality control processes. Therefore, an efficient, rapid, and intelligent model is required to improve industrial products’ production fault recognition and classification for optimal visual inspections and quality control. However, intelligent models obtained with a tradeoff of high accuracy for high latency are tedious for real-time implementation and inferencing. This work proposes an ensemble deep-leaning architectural framework based on a deep learning model architectural voting policy to compute and learn the hierarchical and high-level features in industrial artefacts. The voting policy is formulated with respect to three crucial viable model characteristics: model optimality, efficiency, and performance accuracy. In the study, three publicly available industrial produce datasets were used for the proposed model’s various experiments and validation process, with remarkable results recorded, demonstrating a significant increase in fault recognition and classification performance in industrial products. In the study, three publicly available industrial produce datasets were used for the proposed model’s various experiments and validation process, with remarkable results recorded, demonstrating a significant increase in fault recognition and classification performance in industrial products.

A novel multi-sensor hybrid fusion framework

“…This is because the hard voting strategy relies on the voting results of the majority of classifiers. Therefore, the majority of 'inferior votes' may outperform the minority of 'superior votes', thus affecting the overall performance [65].…”

A Fault Diagnosis Approach for Wind Turbine Gearbox Based on Ensemble Learning Model and Dung Beetle Optimization Algorithm

“…Nine basic models are constructed, which have the same neural network layer type and layer structure, with different activation functions. Then, an ensemble model is obtained by integrating the nine base models using hard voting [9]. Meanwhile, the target function is used as the fitness function for the optimization algorithm to select suitable combinations of base models and obtain a new ensemble model.…”

A Unified Efficient Deep Learning Architecture for Rapid Safety Objects Classification Using Normalized Quantization-Aware Learning

“…On the other hand, the spatially separable convolutions decompose convolution operations into two individual processes. In a typical convolution operation [24,25], if a 3 × 3 kernel filter is employed, a sample image can be convolved directly with the kernel. However, in spatially separable convolution, a 3 × 1 kernel is first used to convolve over the given image sample before a 1 × 3 kernel, which is more parameter efficient compared to the conventional convolution layers because of minimal matrix computations.…”