Defect recognition is an important task in intelligent manufacturing. Due to the subjectivity of human annotation, the collected defect data usually contains a lot of noise and unpredictable uncertainties, which have a great negative influence on defect recognition. It is a significant challenge to discover an effective defect recognition model with satisfactory uncertainty processing ability. A natural way is to automatically search for an efficient deep model, which can be realized by neural architecture search (NAS). To achieve this, we propose an efficient fuzzy NAS framework for defect recognition, where the searched architecture can effectively handle uncertain information from the given datasets. Specifically, we first design a fuzzy search space and the related encoding strategy for fuzzy NAS. Then, we propose a comparator-based evolutionary search approach, where an online end-to-end comparator is learned to directly determine the selection of candidate architectures from the evolutionary population. The comparator works in an end-to-end way and it transforms the complex ranking problem of evaluating architectures into a simple classification task, which overcomes the rank disorder issue suffered from traditional performance predictors. A series of experimental results demonstrate that the architecture with fewer #Params (1.22 M) search by fuzzy neural architecture search framework for defect recognition method achieves higher accuracy (92.26%) compared to the state-of-the-art results (i.e., DARTS-PV) on the ELPV dataset, as well as competitive results (accuracy = 76.4%, #Params = 1.04 M) on the CODEBRIM dataset. Experimental results show the effectiveness and efficiency of our proposed method in handling uncertain problems.

Detecting and classifying vehicles as objects from images and videos is challenging in appearance-based representation, yet plays a significant role in the substantial real-time applications of Intelligent Transportation Systems (ITSs). The rapid development of Deep Learning (DL) has resulted in the computer-vision community demanding efficient, robust, and outstanding services to be built in various fields. This paper covers a wide range of vehicle detection and classification approaches and the application of these in estimating traffic density, real-time targets, toll management and other areas using DL architectures. Moreover, the paper also presents a detailed analysis of DL techniques, benchmark datasets, and preliminaries. A survey of some vital detection and classification applications, namely, vehicle detection and classification and performance, is conducted, with a detailed investigation of the challenges faced. The paper also addresses the promising technological advancements of the last few years.

Analysis of hyperspectral imagery (HSI) is a critical aspect of remote sensing in precision agriculture, for which effective dimensionality reduction (DR) strategies for the inherent complexity and uncertainty of the data are highly necessary. The fusion of fuzzy logic with DR techniques offers potential promises to refine enough feature information from the classification system, which may potentially compromise critical information. However, graph-based deep learning, especially the use of graph attention networks (GATs), has advanced the field by ensuring enhanced spatial and spatial-type feature relationships, which alleviates classification errors. Although graph convolution networks and their predecessors have greatly advanced the use of spatial relationship features, deep learning algorithms, such as three-dimensional convolutional neural networks (3D-CNNs), have resorted to reducing the need for and reliance on high samples through spatial feature learning and consideration. Therefore, to advance the field, a novel approach, Enhanced Transformation enabled Fuzzy Graph network (ETFG), was designed, which is a combination of deep fuzzy-based DR, enhanced through the use of 3D-CNN and GAT, with the application of principal component analysis (PCA) for optimized DR. The ETFG model entailed two major processing stages, where the initial stage involved classifying the raw data cube using the 3D-CNN, and then the results processed an algorithm enriched by lightweight GAT-based modules. The ETFG model combines fuzzy C-means clustering and optimized DR using PCA, contributing the best of PCA and GAT capabilities for optimized classification. At high-performance optimal DR, the ETFG model offered optimal multispectral imaging and the analysis and classification of hyperspectral data, which is promising enough to advance the field's needs.

A fixed interconnection parallel architecture is characterized by a graph, with vertices corresponding to processing nodes and edges representing communication links. An ordered set R of nodes in a graph G is said to be a resolving set of G if every node in G is uniquely determined by its vector of distances to the nodes in R. Each node in R can be thought of as the site for a sonar or loran station, and each node location must be uniquely determined by its distances to the sites in R. A fault-tolerant resolving set R for which the failure of any single station at node location v in R leaves us with a set that still is a resolving set. The metric dimension (resp. fault-tolerant metric dimension) is the minimum cardinality of a resolving set (resp. fault-tolerant resolving set). In this paper, we study the metric and fault-tolerant dimension of certain families of interconnection networks. In particular, we focus on the fault-tolerant metric dimension of the butterfly, the Benes and a family of honeycomb derived networks called the silicate networks. Our main results assert that three aforementioned families of interconnection have an unbounded fault-tolerant resolvability structures. We achieve that by determining certain maximal and minimal results on their fault-tolerant metric dimension.

Graph signal processing deals with signals whose domain, defined by a graph, is irregular. The total $\pi $ -electron energy or simply the $\pi $ -electronic energy, as calculated within the Hückel tight-binding molecular orbital approximation, is one of the important quantum-theoretical characteristic of conjugated molecules. In this paper, we propose an efficient computer-assisted computational method to determine eigenvalues-based distance descriptors for chemical compounds which are then used to learn to quantitative relationship between the activity/property and the structure (QSAR/QSPR) of compound. Comparisons with other similar methods show that our proposed method possesses less algorithmic and computational complexities and is more computationally diverse. The proposed method is used to determine predictive potential of eigenvalues-based distance descriptors for measuring the $\pi $ -electronic energy of benzenoid hydrocarbons. Importantly, we propose three new chemical matrices and, unexpectedly, results show that the spectral descriptors defined based on new chemical matrices outperform all the well-known descriptors in the literature. Specifically, our proposed second atom-bond connectivity Estrada index show the best correlation coefficient of 0.9997. Applications of our computational method to certain infinite families of carbon nanotubes and carbon nanocones are presented. The obtained results can potentially be used to determine the $\pi $ -electronic energy of these nanotubes and nanocones theoretically with higher accuracy and negligible error.