The system deadlock problem of the flexible manufacturing system (FMS) needs to be solved urgently. Deadlocks may occur when resources are repeatedly requested by multiple working processes, where such competition causes a lack of resources, and the system cannot get into subsequent processes smoothly. In this paper, we build FMS by using a particular class of Petri Net (PN) model called a system of simple sequential processes with resources (S3PR) net, which features a high utilization rate of resources. When parts of resources are held by related processes and these processes wait for others to release resources to move on, a circular waiting situation occurs. Circular waiting is one of the conditions for system deadlocks occurring, where the system under this situation has a high probability of causing system deadlocks. The resources being held have to be released to break the waiting states. This paper develops a novel control transition-based deadlock recovery policy by combining graph theory and reachability graph analysis technique. The proposed approach firstly builds a subnet of the PN model called resource flow graph (RFG) and finds out all its hold and request circuits (HRCs), which may lead to a circular waiting situation. In addition, we design a new algorithm to obtain the minimal set of control transitions further. We introduce three PN models of FMS to demonstrate the proposed method. Through experimental results and comparison between related works, it is known that the proposed deadlock recovery uses fewer controllers and has the same performance.

As the use of big data and its potential benefits become more widespread, public and private organizations around the world have realized the imperative of incorporating comprehensive and robust technologies into their business processes. In particular, companies are implementing more and more intelligent systems into their business processes, resulting in an exponential increase in the amount of data being collected and used in everyday life on a quarterly basis. Therefore, an algorithm that can efficiently explore the frequent patterns from big data will be able to productively analyze and utilize the generated data to better optimize resources and even develop new business models. Thus, in this urban data era, it is not only essential to make more efficient and accurate decisions to improve business profitability and customer satisfaction, but also a highly challenging core issue. In this study, a new approach, scalable FP mining, is proposed for the first time to achieve a certain level of performance and more efficient memory utilization despite large amounts of data. Experiment results for different data characteristics show that the best performance of the proposed method requires only 48.9% of the DP algorithm's execution time. In further experiments with increased exploration difficulty, the optimal execution time is only 37%. On specific associated databases, the proposed method maintains stable and excellent performance with 12.7% for different data characteristics. The experimental results with increasing exploration difficulty at each stage demonstrate the stability, robustness, and reliability of the proposed SFP, especially under high data complexity and exploration difficulty.

Given an undirected, unweighted graph with $n$ vertices and $m$ edges, the maximum cut problem is to find a partition of the $n$ vertices into disjoint subsets $V_1$ and $V_2$ such that the number of edges between them is as large as possible. Classically, it is an NP-complete problem, which has potential applications ranging from circuit layout design, statistical physics, computer vision, machine learning and network science to clustering. In this paper, we propose a quantum algorithm to solve the maximum cut problem for any graph $G$ with a quadratic speedup over its classical counterparts, where the temporal and spatial complexities are reduced to, respectively, $O(\sqrt{2^n/r})$ and $O(m^2)$. With respect to oracle-related quantum algorithms for NP-complete problems, we identify our algorithm as optimal. Furthermore, to justify the feasibility of the proposed algorithm, we successfully solve a typical maximum cut problem for a graph with three vertices and two edges by carrying out experiments on IBM's quantum computer.

After the fourth industrial evolution, precision and automatic manufacturing have become increasingly widely accepted in production. With highly variable productivity and flexibility, flexible manufacturing systems (FMS) lower production costs and increase efficiency. Due to its resource shareability, unexpected system deadlock may occur in some specific situations. Many existing works use deadlock prevention as the primary control methodology in research on system deadlock control, while this type of control policy would constrain the transportation resources and reduce the system’s liveness. This paper adopts a new transition-based deadlock recovery policy as the direct control strategy, which uses generating and comparing aiding matrix (GCAM) to determine the optimal control transition. We also improve the existing GCAM-based method by reducing the computational redundancy. This kind of control strategy and its benefit could be demonstrated through two typical systems of simple sequential processes with resource (S3PR) nets and their Petri nets model.

The amount of information nowadays is rapidly growing. Aside from valuable information, information that is unrelated to a target or is meaningless is also growing. Big data and broader digital technologies are considered the primary components of smart city governance and planning. Big data analysis is considered to define a new era in urban planning, research, and policy. Effective data mining and pattern detection techniques are becoming very important these days. Processing such a large amount of data entails the use of data mining, a technique that clarifies the association between valid information and excludes irrelevant data to implement a practical decision tree. A large amount of data affects processing time and I/O costs during data mining. This study proposes to distribute data among multiple clients and distribute a large amount of data computation equally to improve the resource cost problem of exploration. Following that, the main server consolidates the computation results and generates the survey results. Experiment results show that the proposed algorithm is superior, thus allowing a larger amount of data to be processed while producing high-quality results.

Action Recognition has been studied for many years. In recent years, there are some methods using 3D-CNN (C3D, I3D, R2 + 1D), which have high accuracy, but it is hard to train and quite time-consuming due to the network architecture of extracting spatial–temporal features and the huge action dataset. Since 2D-CNN has a pre-trained model with high accuracy and speed in object recognition, there is also a method of fine-tune it on Recurrent neural network (RNN), Long Short-Term Memory (LSTM) network and other network that can extract temporal features, but due to the poor performance of fine-tune, although the speed is increased, the accuracy has dropped significantly. Therefore, this research wants to use the high accuracy of 3D-CNN to distill 2D-CNN produce a great pre-trained model for action recognition and combine it with Attention Mechanism LSTM to make model on fine-tune on other action dataset can accelerate and achieve the accuracy of approximating 3D-CNN.

In Industry 4.0, automation is a critical requirement for mechanical production. This study proposes a computer vision-based method to capture images of rotating tools and detect defects without the need to stop the machine in question. The study uses frontal lighting to capture images of the rotating tools and employs scale-invariant feature transform (SIFT) to identify features of the tool images. Random sample consensus (RANSAC) is then used to obtain homography information, allowing us to stitch the images together. The modified YOLOv4 algorithm is then applied to the stitched image to detect any surface defects on the tool. The entire tool image is divided into multiple patch images, and each patch image is detected separately. The results show that the modified YOLOv4 algorithm has a recall rate of 98.7% and a precision rate of 97.3%, and the defect detection process takes approximately 7.6 s to complete for each stitched image.

Instead of the mass production industry in the past, over these decades, precision and automatic manufacturing have become more and more widely accepted in the production area. With highly variable productivity and flexibility, Flexible Manufacturing Systems (FMS) can decrease production costs and increase efficiency. Due to its resource sharing, unexpected system deadlock may occur in some situations. In research on system deadlock control, lots of existing literature use deadlock prevention as the primary control method, while it could block resources transporting and downscale the system reachability graph. This paper adopts a deadlock recovery policy as the direct control strategy based on control transition technology. This kind of control strategy and its benefit could be demonstrated through classical systems of simple sequential processes with resources (S 3 PR) nets and their Petri nets model.

Currently, the majority of industrial metal processing involves the use of taps for cutting. However, existing tap machines require relocation to specialized inspection stations and only assess the condition of the cutting edges for defects. They do not evaluate the quality of the cutting angles and the amount of removed material. Machine vision, a key component of smart manufacturing, is commonly used for visual inspection. Taps are employed for processing various materials. Traditional tap replacement relies on the technician’s accumulated empirical experience to determine the service life of the tap. Therefore, we propose the use of visual inspection of the tap’s external features to determine whether replacement or regrinding is needed. We examined the bearing surface of the tap and utilized single images to identify the cutting angle, clearance angle, and cone angles. By inspecting the side of the tap, we calculated the wear of each cusp. This inspection process can facilitate the development of a tap life system, allowing for the estimation of the durability and wear of taps and nuts made of different materials. Statistical analysis can be employed to predict the lifespan of taps in production lines. Experimental error is 16 μm. Wear from tapping 60 times is equivalent to 8 s of electric grinding. We have introduced a parameter, thread removal quantity, which has not been proposed by anyone else.

In this paper, we propose a bio-molecular algorithm with O( ${n}$ 2 ) biological operations, O( $2^{n-1}$ ) DNA strands, O( ${n}$ ) tubes and the longest DNA strand, O( ${n}$ ), for inferring the value of a bit from the only output satisfying any given condition in an unsorted database with $2^{n}$ items of ${n}$ bits. We show that the value of each bit of the outcome is determined by executing our bio-molecular algorithm ${n}$ times. Then, we show how to view a bio-molecular solution space with $2^{{\textit {n-1}}}$ DNA strands as an eigenvector and how to find the corresponding unitary operator and eigenvalues for inferring the value of a bit in the output. We also show that using an extension of the quantum phase estimation and quantum counting algorithms computes its unitary operator and eigenvalues from bio-molecular solution space with $2^{{\textit {n-1}}}$ DNA strands. Next, we demonstrate that the value of each bit of the output solution can be determined by executing the proposed extended quantum algorithms ${n}$ times. To verify our theorem, we find the maximum-sized clique to a graph with two vertices and one edge and the solution ${b}$ that satisfies ${b}^{2} \equiv 1$ (mod 15) and $1 < {b} < (15/2)$ using IBM Quantum’s backend.

In recent years, knowledge discovery in databases provides a powerful capability to discover meaningful and useful information.For numerous real-life applications, frequent pattern mining and association rule mining have been extensively studied.In traditional mining algorithms, data are centralized and memory-resident.As a result of the large amount of data, bandwidth limitation, and energy limitations when applying these methods to distributed databases, especially in this era of big data, the performance is not effective enough.Hence, data mining on distributed environments has emerged as an important research area.To improve the performance, we propose a set of algorithms based on FP growth that discover FPs that are capable of providing fast and scalable service in distributed computing environments and a brief data structure to store items and counts to minimize the data for transmission on the network.To ensure completeness and execution capability, DistEclat and BigFIM were considered for the experiment comparison.Experiments show that the proposed method has superior cost-effectiveness for processing massive datasets and good capabilities under various experiment conditions.The proposed method on average required only 33% of the execution time and 45% of the transmission cost of DistEclat.Compared to BigFIM, The proposed method on average required 23.3% of the execution time and 14.2% of the transmission cost of BigFIM.

This study proposes an integrated deep network consisting of a detection and identification module for person search. Person search is a very challenging problem because of the large appearance variation caused by occlusion, background clutter, pose variations, etc., and it is still an active research issue in the academic and industrial fields. Although various studies have been proposed, following the protocols of the person re-identification (ReID) benchmarks, most existing works take cropped pedestrian images either from manual labelling or a perfect detection assumption. However, for person search, manual processing is unavailable in practical applications, thereby causing a gap between the ReID problem setting and practical applications. One fact is also ignored: an imperfect auto-detected bounding box or misalignment is inevitable. We design herein a framework for the practical surveillance scenarios in which the scene images are captured. For person search, detection is a necessary step before ReID, and previous studies have shown that the precision of detection results has an influence on person ReID. The detection module based on the Faster R-CNN is used to detect persons in a scene image. For identifying and extracting discriminative features, a multi-class CNN network is trained with the auto-detected bounding boxes from the detection module, instead of the manually cropped data. The distance metric is then learned from the discriminative features output by the identification module. According to the experimental results of the test performed in the scene images, the multi-class CNN network for the identification module can provide a 62.7% accuracy rate, which is higher than that for the two-class CNN network.

According to the World Health Organization global status report on road safety, traffic accidents are the eighth leading cause of death in the world, and nearly one-fifth of the traffic accidents were cause by driver distractions. Inspired by the famous two-stream convolutional neural network (CNN) model, we propose a driver behavior analysis system using one spatial stream ConvNet to extract the spatial features and one temporal stream ConvNet to capture the driver's motion information. Instead of using three-dimensional (3D) ConvNet, which would suffer from large parameters and the lack of a pre-trained model, two-dimensional (2D) ConvNet is used to construct the spatial and temporal ConvNet streams, and they were pre-trained by the large-scale ImageNet. In addition, in order to integrate different modalities, the feature-level fusion methodology was applied, and a fusion network was designed to integrate the spatial and temporal features for further classification. Moreover, a self-compiled dataset of 10 actions in the vehicle was established. According to the experimental results, the proposed system can increase the accuracy rate by nearly 30% compared to the two-stream CNN model with a score-level fusion.

Video frame interpolation algorithm aims at synthesizing intermediate frame(s) sequence between two consecutive frames, these intermediate frames are both temporally and spatially coherent with input frames and each other. Video frame interpolation is a classic problem in computer vision and has many applications, e.g., frame rate upscaling and slow-motion effect.Most existing approaches are single-frame interpolation and have shown impress performance. However, these approa-ches, which cannot be directly used to synthesize multiple frames at one time, sometimes are user-unfriendly. On the other hand, existing multiple-frame interpolation approaches sometimes lead to higher storage space or computational cost. Therefore, we propose an adaptive variable frame inter-polation method, which evaluates the amount of frame to be generated according to the estimated motion, to reduce the storage space and improve the generation efficiency. In addition, we add edge loss to the loss function, expecting better results.

The tiny size of defects and the noise information in the Ball-Grid-Array (BGA) Chip images have been challenging the image-processing based visual inspection systems in the Integrated Circuit (IC) manufacturing industry. Moreover, the gradient vanishing and high time-consuming problems of deep neural network models are a big obstacle for its application to solve industrial projects. This paper focuses on proposing a modified version of the YOLOv3 model, a leading object detection and classification method in terms of speed; and its application to deal with the problem of detecting and classifying defects on BGA Chip Images. There are five modifications constructed on (4) YOLOv3 architecture that are aimed to enhance the ability of feature extraction towards small objects, strengthen the flow of information inside the network and eliminate the problem of redundant information. With the application of this model into BGA Chip Defects estimation, 49 patches (320x320 pixels), extracted from a single high-resolution BGA Chip image (1, 450x1,450 pixels), are continuously fed into the modified YOLOv3 model to detect and classify the inner defects. As a result, the problem of BGA Chip Defects estimation is solved with the highest performance achieves an average precision of 86% at IoU (Intersection over Union) of 0.75 and an average recall of 99%.

Anime line sketch colorization is to fill a variety of colors the anime sketch, to make it colorful and diverse. The coloring problem is not a new research direction in the field of deep learning technology. Because of coloring of the anime sketch does not have fixed color and we can't take texture or shadow as reference, so it is difficult to learn and have a certain standard to determine whether it is correct or not. After generative adversarial networks (GANs) was proposed, some used GANs to do coloring research, achieved some result, but the coloring effect is limited. This study proposes a method use deep residual network, and adding discriminator to network, that expect the color of colored images can consistent with the desired color by the user and can achieve good coloring results.

Data mining is a set of methods used to mine hidden information from data. It mainly includes frequent pattern mining, sequential pattern mining, classification, and clustering. Frequent pattern mining is used to discover the correlation among various sets of items within large databases. The rapid upward trend in data size slows the mining of frequent patterns. Numerous studies have attempted to develop algorithms that operate in distributed computing environments to accelerate the mining process. FLR-mining (Fast, Load balancing and Resource efficient mining algorithm) is one of the fastest methods of mining with efficient consideration of load balancing and resources. FLR-mining can automatically determine the appropriate number of computing nodes. However, FLR-mining and existing methods assume that the network bandwidth is constant. In practical distributed and many-task computing systems, this assumption fails because there are packet collisions caused by many mining tasks that run in a simultaneous manner. Therefore, a method that can consider the varying network bandwidth is necessary. In this study, we propose a method that can rapidly mine frequent patterns under the varying network bandwidth. The proposed method can also determine the appropriate number of computing nodes to efficiently utilize computing resources and achieve load balancing. Through empirical evaluation, the proposed method is shown to deliver excellent performance in terms of execution efficiency and load balancing.

Super-resolution is the use of low-resolution images to reconstruct corresponding high-resolution images. This technology is used in many places such as medical fields and monitor systems. The traditional method is to interpolate to fill in the information lost when the image is enlarged. The initial use of deep learning is SRCNN, which is divided into three steps, extracting image block features, feature nonlinear mapping and reconstruction. Both PSNR and SSIM have significant progress compared with traditional methods, but there are still some details in detail restoration. defect. SRGAN will generate anti-network applications to SR problems. The method is to improve the image magnification by more than 4 times, which is easy to produce too smooth. In this study, we hope to improve the EnhanceNet by training with different loss functions and different types of images to achieve better reconstruction results.

In this study, we propose a viewpoint-invariant person re-identification scheme with pose priors and weighted local features. We divide the pose angle into three classes: (0 degrees, 180 degrees), (45 degrees, 135 degrees), and 90 degrees. Each of the classes has a weighted map. In addition, the texture-based feature, histogram of oriented gradients, is extracted to predict pose angle using support vector machine. Moreover, two additional features, salient color names and local binary patterns (LBP), are extracted. The former feature is computed using a weighted map with Gaussian distribution. The latter feature is computed using a weighted map based on the predicted pose angle. Then, the image representation is concatenated with salient color names and LBP. Finally, we adopt cross-view quadratic discriminant analysis for person re-identification.

Association rules mining has attracted much attention among data mining topics because it has been successfully applied in various fields to find the association between purchased items by identifying frequent patterns (FPs). Currently, databases are huge, ranging in size from terabytes to petabytes. Although past studies can effectively discover FPs to deduce association rules, the execution efficiency is still a critical problem, particularly for big data. Progressive size working set (PSWS) and parallel FP-growth (PFP) are state-of-the-art methods that have been applied successfully to parallel and distributed computing technology to improve mining processing time in many-task computing, thereby bridging the gap between high-throughput and high-performance computing. However, such methods cannot mine before obtaining a complete FP-tree or the corresponding subdatabase, causing a high idle time for computing nodes. We propose a method that can begin mining when a small part of an FP-tree is received. The idle time of computing nodes can be reduced, and thus, the time required for mining can be reduced effectively. Through an empirical evaluation, the proposed method is shown to be faster than PSWS and PFP.