Deep convolutional neural networks for eigenvalue problems in mechanics

We show that deep convolutional neural networks (CNNs) can massively outperform traditional densely connected neural networks (NNs) (both deep or shallow) in predicting eigenvalue problems in mechanics. In this sense, we strike out in a new direction in mechanics computations with strongly predictive NNs whose success depends not only on architectures being deep but also being fundamentally different from the widely used to date. We consider a model problem: predicting the eigenvalues of one-dimensional (1D) and two-dimensional (2D) phononic crystals. For the 1D case, the optimal CNN architecture reaches 98% accuracy level on unseen data when trained with just 20 000 samples, compared to 85% accuracy even with 100 000 samples for the typical network of choice in mechanics research. We show that, with relatively high data efficiency, CNNs have the capability to generalize well and automatically learn deep symmetry operations, easily extending to higher dimensions and our 2D case. Most importantly, we show how CNNs can naturally represent mechanical material tensors, with its convolution kernels serving as local receptive fields, which is a natural representation of mechanical response. Strategies proposed are applicable to other mechanics' problems and may, in the future, be used to sidestep cumbersome algorithms with purely data-driven approaches based upon modern deep architectures.     

Acoustic Emission Recognition Based on a Two-Streams Convolutional Neural Network

The Convolutional Neural Network (CNN) is a widely used deep neural network. Compared with the shallow neural network, the CNN network has better performance and faster computing in some image recognition tasks. It can effectively avoid the problem that network training falls into local extremes. At present, CNN has been applied in many different fields, including fault diagnosis, and it has improved the level and efficiency of fault diagnosis. In this paper, a two-streams convolutional neural network (TCNN) model is proposed. Based on the short-time Fourier transform (STFT) spectral and Mel Frequency Cepstrum Coefficient (MFCC) input characteristics of two-streams acoustic emission (AE) signals, an AE signal processing and classification system is constructed and compared with the traditional recognition methods of AE signals and traditional CNN networks. The experimental results illustrate the effectiveness of the proposed model. Compared with single-stream convolutional neural network and a simple Long Short-Term Memory (LSTM) network, the performance of TCNN which combines spatial and temporal features is greatly improved, and the accuracy rate can reach 100% on the current database, which is 12% higher than that of single-stream neural network.     

Real-Time Violent Action Recognition Using Key Frames Extraction and Deep Learning

Violence recognition is crucial because of its applications in activities related to security and law enforcement. Existing semi-automated systems have issues such as tedious manual surveillances, which causes human errors and makes these systems less effective. Several approaches have been proposed using trajectory-based, non-object-centric, and deep-learning-based methods. Previous studies have shown that deep learning techniques attain higher accuracy and lower error rates than those of other methods. However, the their performance must be improved. This study explores the state-of-the-art deep learning architecture of convolutional neural networks (CNNs) and inception V4 to detect and recognize violence using video data. In the proposed framework, the keyframe extraction technique eliminates duplicate consecutive frames. This keyframing phase reduces the training data size and hence decreases the computational cost by avoiding duplicate frames. For feature selection and classification tasks, the applied sequential CNN uses one kernel size, whereas the inception v4 CNN uses multiple kernels for different layers of the architecture. For empirical analysis, four widely used standard datasets are used with diverse activities. The results confirm that the proposed approach attains 98% accuracy, reduces the computational cost, and outperforms the existing techniques of violence detection and recognition.     

A Dynamically Reconfigurable Accelerator Design Using a Sparse-Winograd Decomposition Algorithm for CNNs

Convolutional Neural Networks (CNNs) are widely used in many fields. Due to their high throughput and high level of computing characteristics, however, an increasing number of researchers are focusing on how to improve the computational efficiency, hardware utilization, or flexibility of CNN hardware accelerators. Accordingly, this paper proposes a dynamically reconfigurable accelerator architecture that implements a Sparse-Winograd F(2 2.3 3)-based high-parallelism hardware architecture. This approach not only eliminates the pre-calculation complexity associated with the Winograd algorithm, thereby reducing the difficulty of hardware implementation, but also greatly improves the flexibility of the hardware; as a result, the accelerator can realize the calculation of Conventional Convolution, Grouped Convolution (GCONV) or Depthwise Separable Convolution (DSC) using the same hardware architecture. Our experimental results show that the accelerator achieves a 3x–4.14x speedup compared with the designs that do not use the acceleration algorithm on VGG-16 and MobileNet V1. Moreover, compared with previous designs using the traditional Winograd algorithm, the accelerator design achieves 1.4x–1.8x speedup. At the same time, the efficiency of the multiplier improves by up to 142%.     

一种用细胞神经网络提取干涉条纹中心的新方法

提取干涉条纹的中心是干涉测量的关键环节,文中提出了一种基于细胞神经网络（CNN）提取干涉条纹中心的新方法.CNN是一种实时处理信号的大规模非线性模拟电路,同时它的局部联接特点使其适用于超大规模集成电路的实现.CNN具有并行运算的能力,可消除传统串行算法复杂性高、不能实时处理的缺点.对该方法进行了分析,给出了实例的仿真结果,证明该方法能快速准确地提取干涉条纹的中心,提高了干涉条纹的判别精度,从而增加了实验中干涉条纹处理的直观性和实时性.     

Convolutional and generative adversarial neural networks in manufacturing

Manufacturing is undergoing transformation driven by the developments in process technology, information technology, and data science. A future manufacturing enterprise will be highly digital. This will create opportunities for machine learning algorithms to generate predictive models across the enterprise in the spirit of the digital twin concept. Convolutional and generative adversarial neural networks have received some attention of the manufacturing research community. Representative research and applications of the two machine learning concepts in manufacturing are presented. Advantages and limitations of each neural network are discussed. The paper might be helpful in identifying research gaps, inspire machine learning research in new manufacturing domains, contribute to the development of successful neural network architectures, and getting deeper insights into the manufacturing data.     

Advancements in materials,devices, and integration schemes for a new generation of neuromorphic computers

The slowing pace of performance improvements in modern processors along with the breakdown of power scaling forecasts an imminent end to the traditional transistor scaling roadmap. Additionally, meeting the aggressive demands of proliferating applications in big-data processing, machine learning, artificial intelligence, and highly distributed edge computing requires radical advancements in materials, devices, and architectures for future processors. Neuromorphic computing has emerged as the most promising successor to conventional complementary metal oxide semiconductor (CMOS) devices and von Neumann architecture. This work reviews the status of neuromorphic research, compares the traditional CMOS approach with neuromorphic devices for implementing biologically inspired circuits, and provides an outlook into integration schemes for future brain-inspired computing hardware.     

Detecting Driver Distraction Using Deep-Learning Approach

Currently, distracted driving is among the most important causes of traffic accidents. Consequently, intelligent vehicle driving systems have become increasingly important. Recently, interest in driver-assistance systems that detect driver actions and help them drive safely has increased. In these studies, although some distinct data types, such as the physical conditions of the driver, audio and visual features, and vehicle information, are used, the primary data source is images of the driver that include the face, arms, and hands taken with a camera inside the car. In this study, an architecture based on a convolution neural network (CNN) is proposed to classify and detect driver distraction. An efficient CNN with high accuracy is implemented, and to implement intense convolutional networks for large-scale image recognition, a new architecture was proposed based on the available Visual Geometry Group (VGG-16) architecture. The proposed architecture was evaluated using the StateFarm dataset for driver-distraction detection. This dataset is publicly available on Kaggle and is frequently used for this type of research. The proposed architecture achieved 96.95% accuracy.     

R²N: A Novel Deep Learning Architecture for Rain Removal from Single Image

Visual degradation of captured images caused by rainy streaks under rainy weather can adversely affect the performance of many open-air vision systems. Hence, it is necessary to address the problem of eliminating rain streaks from the individual rainy image. In this work, a deep convolution neural network (CNN) based method is introduced, called Rain-Removal Net (R²N), to solve the single image de-raining issue. Firstly, we decomposed the rainy image into its high-frequency detail layer and low-frequency base layer. Then, we used the high-frequency detail layer to input the carefully designed CNN architecture to learn the mapping between it and its corresponding de-rained high-frequency detail layer. The CNN architecture consists of four convolution layers and four deconvolution layers, as well as three skip connections. The experiments on synthetic and real-world rainy images show that the performance of our architecture outperforms the compared state-of-the-art de-raining models with respects to the quality of de-rained images and computing efficiency.     

Identification of Anomaly Scenes in Videos Using Graph Neural Networks

Generally, conventional methods for anomaly detection rely on clustering, proximity, or classification. With the massive growth in surveillance videos, outliers or anomalies find ingenious ways to obscure themselves in the network and make conventional techniques inefficient. This research explores the structure of Graph neural networks (GNNs) that generalize deep learning frameworks to graph-structured data. Every node in the graph structure is labeled and anomalies, represented by unlabeled nodes, are predicted by performing random walks on the node-based graph structures. Due to their strong learning abilities, GNNs gained popularity in various domains such as natural language processing, social network analytics and healthcare. Anomaly detection is a challenging task in computer vision but the proposed algorithm using GNNs efficiently performs the identification of anomalies. The Graph-based deep learning networks are designed to predict unknown objects and outliers. In our case, they detect unusual objects in the form of malicious nodes. The edges between nodes represent a relationship of nodes among each other. In case of anomaly, such as the bike rider in Pedestrians data, the rider node has a negative value for the edge and it is identified as an anomaly. The encoding and decoding layers are crucial for determining how statistical measurements affect anomaly identification and for correcting the graph path to the best possible outcome. Results show that the proposed framework is a step ahead of the traditional approaches in detecting unusual activities, which shows a huge potential in automatically monitoring surveillance videos. Performing autonomous monitoring of CCTV, crime control and damage or destruction by a group of people or crowd can be identified and alarms may be triggered in unusual activities in streets or public places. The suggested GNN model improves accuracy by 4% for the Pedestrian 2 dataset and 12% for the Pedestrian 1 dataset compared to a few state-of-the-art techniques.     

Feature Fusion-Based Deep Learning Network to Recognize Table Tennis Actions

A system for classifying four basic table tennis strokes using wearable devices and deep learning networks is proposed in this study. The wearable device consisted of a six-axis sensor, Raspberry Pi 3, and a power bank. Multiple kernel sizes were used in convolutional neural network (CNN) to evaluate their performance for extracting features. Moreover, a multiscale CNN with two kernel sizes was used to perform feature fusion at different scales in a concatenated manner. The CNN achieved recognition of the four table tennis strokes. Experimental data were obtained from 20 research participants who wore sensors on the back of their hands while performing the four table tennis strokes in a laboratory environment. The data were collected to verify the performance of the proposed models for wearable devices. Finally, the sensor and multi-scale CNN designed in this study achieved accuracy and F1 scores of 99.58% and 99.16%, respectively, for the four strokes. The accuracy for five-fold cross validation was 99.87%. This result also shows that the multi-scale convolutional neural network has better robustness after five-fold cross validation.     

Deep learning for photoacoustic tomography from sparse data

The development of fast and accurate image reconstruction algorithms is a central aspect of computed tomography. In this paper, we investigate this issue for the sparse data problem in photoacoustic tomography (PAT). We develop a direct and highly efficient reconstruction algorithm based on deep learning. In our approach, image reconstruction is performed with a deep convolutional neural network (CNN), whose weights are adjusted prior to the actual image reconstruction based on a set of training data. The proposed reconstruction approach can be interpreted as a network that uses the PAT filtered backprojection algorithm for the first layer, followed by the U-net architecture for the remaining layers. Actual image reconstruction with deep learning consists in one evaluation of the trained CNN, which does not require time-consuming solution of the forward and adjoint problems. At the same time, our numerical results demonstrate that the proposed deep learning approach reconstructs images with a quality comparable to state of the art iterative approaches for PAT from sparse data.     

Sentence Similarity Measurement with Convolutional Neural Networks Using Semantic and Syntactic Features

Calculating the semantic similarity of two sentences is an extremely challenging problem. We propose a solution based on convolutional neural networks (CNN) using semantic and syntactic features of sentences. The similarity score between two sentences is computed as follows. First, given a sentence, two matrices are constructed accordingly, which are called the syntax model input matrix and the semantic model input matrix; one records some syntax features, and the other records some semantic features. By experimenting with different arrangements of representing the syntactic and semantic features of the sentences in the matrices, we adopt the most effective way of constructing the matrices. Second, these two matrices are given to two neural networks, which are called the sentence model and the semantic model, respectively. The convolution process of the neural networks of the two models is carried out in multiple perspectives. The outputs of the two models are combined as a vector, which is the representation of the sentence. Third, given the representation vectors of two sentences, the similarity score of these representations is computed by a layer in the CNN. Experiment results show that our algorithm (SSCNN) surpasses the performance MPCPP, which noticeably the best recent work of using CNN for sentence similarity computation. Comparing with MPCNN, the convolution computation in SSCNN is considerably simpler. Based on the results of this work, we suggest that by further utilization of semantic and syntactic features, the performance of sentence similarity measurements has considerable potentials to be improved in the future.     

Convolutional Neural Network for Histopathological Osteosarcoma Image Classification

Osteosarcoma is one of the most widespread causes of bone cancer globally and has a high mortality rate. Early diagnosis may increase the chances of treatment and survival however the process is time-consuming (reliability and complexity involved to extract the hand-crafted features) and largely depends on pathologists’ experience. Convolutional Neural Network (CNN—an end-to-end model) is known to be an alternative to overcome the aforesaid problems. Therefore, this work proposes a compact CNN architecture that has been rigorously explored on a Small Osteosarcoma histology Image Dataaseet (a high-class imbalanced dataset). Though, during training, class-imbalanced data can negatively affect the performance of CNN. Therefore, an oversampling technique has been proposed to overcome the aforesaid issue and improve generalization performance. In this process, a hierarchical CNN model is designed, in which the former model is non-regularized (due to dense architecture) and the later one is regularized, specifically designed for small histopathology images. Moreover, the regularized model is integrated with CNN’s basic architecture to reduce overfitting. Experimental results demonstrate that oversampling might be an effective way to address the imbalanced class problem during training. The training and testing accuracies of the non-regularized CNN model are 98% & 78% with an imbalanced dataset and 96% & 81% with a balanced dataset, respectively. The regularized CNN model training and testing accuracies are 84% & 75% for an imbalanced dataset and 87% & 86% for a balanced dataset.     

Development of automatic glioma brain tumor detection system using deep convolutional neural networks

We have developed six convolutional neural network (CNN) models for finding optimal brain tumor detection system on high-grade glioma and low-grade glioma lesions from voluminous magnetic resonance imaging human brain scans. Glioma is the most common form of brain tumor. The models are constructed based on the different combinations and settings of hyperparameters with conventional CNN architecture. The six models are two layers with five epochs, five layers with dropout, five layers with stopping criteria (FLSC), FLSC and dropout (FLSCD), FLSC and batch normalization (FLSCBN), and FLSCBN and dropout. The models were trained and tested with BraTS2013 and whole brain atlas data sets. Among them, FLSCBN model yielded the best classification results for brain tumor detection. Experimental results revealed that our deep learning approach was better than the conventional state-of-art methods.     

Efficiency parameters estimation in gemstones cut design using artificial neural networks

This paper deals with the problem of estimating cut results for faceted gemstones. The proposed approach applies artificial neural networks for a faceted gemstones analysis tool that could be further developed for incorporation in a computer-aided-design (CAD) context. Basic concepts concerning gemstone processing are introduced and the design of computational tools using neural networks is discussed. The model presented proposes two criteria to assess the efficiency of lapidary designs for rock crystal quartz: brilliance and yield. Closing the article, 62 different lapidary models were used to train and test the neural network tool.     

AI Cannot Understand Memes: Experiments with OCR and Facial Emotions

The increasing capabilities of Artificial Intelligence (AI), has led researchers and visionaries to think in the direction of machines outperforming humans by gaining intelligence equal to or greater than humans, which may not always have a positive impact on the society. AI gone rogue, and Technological Singularity are major concerns in academia as well as the industry. It is necessary to identify the limitations of machines and analyze their incompetence, which could draw a line between human and machine intelligence. Internet memes are an amalgam of pictures, videos, underlying messages, ideas, sentiments, humor, and experiences, hence the way an internet meme is perceived by a human may not be entirely how a machine comprehends it. In this paper, we present experimental evidence on how comprehending Internet Memes is a challenge for AI. We use a combination of Optical Character Recognition techniques like Tesseract, Pixel Link, and East Detector to extract text from the memes, and machine learning algorithms like Convolutional Neural Networks (CNN), Region-based Convolutional Neural Networks (RCNN), and Transfer Learning with pre-trained denseNet for assessing the textual and facial emotions combined. We evaluate the performance using Sensitivity and Specificity. Our results show that comprehending memes is indeed a challenging task, and hence a major limitation of AI. This research would be of utmost interest to researchers working in the areas of Artificial General Intelligence and Technological Singularity.

     

Artificial neural networks for pattern recognition

This tutorial article deals with the basics of artificial neural networks (ANN) and their applications in pattern recognition. ANN can be viewed as computing models inspired by the structure and function of the biological neural network. These models are expected to deal with problem solving in a manner different from conventional computing. A distinction is made between pattern and data to emphasize the need for developing pattern processing systems to address pattern recognition tasks. After introducing the basic principles of ANN, some fundamental networks are examined in detail for their ability to solve simple pattern recognition tasks. These fundamental networks together with the principles of ANN will lead to the development of architectures for complex pattern recognition tasks. A few popular architectures are described to illustrate the need to develop an architecture specific to a given pattern recognition problem. Finally several issues that still need to be addressed to solve practical problems using ANN approach are discussed. This paper is mostly a consolidation of work reported by several researchers in the literature, some of which is cited in the references. The author has borrowed several ideas and illustrations from the references quoted in this paper.     

Deep Transfer Learning Based Detection and Classification of Citrus Plant Diseases

Citrus fruit crops are among the world’s most important agricultural products, but pests and diseases impact their cultivation, resulting in yield and quality losses. Computer vision and machine learning have been widely used to detect and classify plant diseases over the last decade, allowing for early disease detection and improving agricultural production. This paper presented an automatic system for the early detection and classification of citrus plant diseases based on a deep learning (DL) model, which improved accuracy while decreasing computational complexity. The most recent transfer learning-based models were applied to the Citrus Plant Dataset to improve classification accuracy. Using transfer learning, this study successfully proposed a Convolutional Neural Network (CNN)-based pre-trained model (EfficientNetB3, ResNet50, MobiNetV2, and InceptionV3) for the identification and categorization of citrus plant diseases. To evaluate the architecture’s performance, this study discovered that transferring an EfficientNetb3 model resulted in the highest training, validating, and testing accuracies, which were 99.43%, 99.48%, and 99.58%, respectively. In identifying and categorizing citrus plant diseases, the proposed CNN model outperforms other cutting-edge CNN model architectures developed previously in the literature.