Pattern Recognition Classification Research Articles

An Analog Neural Network Computing Engine Using CMOS-Compatible Charge-Trap-Transistor (CTT)

An analog neural network computing engine based on CMOS-compatible charge-trap transistor (CTT) is proposed in this paper. CTT devices are used as analog multipliers. Compared to digital multipliers, CTT-based analog multiplier shows significant area and power reduction. The proposed computing engine is composed of a scalable CTT multiplier array and energy efficient analog-digital interfaces. By implementing the sequential analog fabric, the engine's mixed-signal interfaces are simplified and hardware overhead remains constant regardless of the size of the array. A proof-of-concept 784 by 784 CTT computing engine is implemented using TSMC 28-nm CMOS technology and occupies 0.68 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> . The simulated performance achieves 76.8 TOPS (8-bit) with 500 MHz clock frequency and consumes 14.8 mW. As an example, we utilize this computing engine to address a classic pattern recognition problem-classifying handwritten digits on MNIST database and obtained a performance comparable to state-of-the-art fully connected neural networks using 8-bit fixed-point resolution.

Analysis of Primary Care Computerized Medical Records (CMR) Data With Deep Autoencoders (DAE)

The use of deep learning is becoming increasingly important in the analysis of medical data such as pattern recognition for classification. The use of primary healthcare computational medical records (CMR) data is vital in prediction of infection prevalence across a population, and decision making at a national scale. To date, the application of machine learning algorithms to CMR data remains under-utilized despite the potential impact for use in diagnostics or prevention of epidemics such as outbreaks of influenza. A particular challenge in epidemiology is how to differentiate incident cases from those that are follow-ups for the same condition. Furthermore, the CMR data are typically heterogeneous, noisy, high dimensional and incomplete, making automated analysis difficult. We introduce a methodology for converting heterogeneous data such that it is compatible with a deep autoencoder for reduction of CMR data. This approach provides a tool for real time visualization of these high dimensional data, revealing previously unknown dependencies and clusters. Our unsupervised nonlinear reduction method can be used to identify the features driving the formation of these clusters that can aid decision making in healthcare applications. The results in this work demonstrate that our methods can cluster more than 97.84% of the data (clusters >5 points) each of which is uniquely described by three attributes in the data: Clinical System (CMR system), Read Code (as recorded) and Read Term (standardized coding). Further, we propose the use of Shannon Entropy as a means to analyse the dispersion of clusters and the contribution from the underlying attributes to gain further insight from the data. Our results demonstrate that Shannon Entropy is a useful metric for analysing both the low dimensional clusters of CMR data, and also the features in the original heterogeneous data. Finally, we find that the entropy of the low dimensional clusters are directly representative of the entropy of the input data (Pearson Correlation = 0.99, R2 = 0.98) and therefore the reduced data from the deep autoencoder is reflective of the original CMR data variability.

Energy Consumption Load Forecasting Using a Level-Based Random Forest Classifier

Energy consumers may not know whether their next-hour forecasted load is either high or low based on the actual value predicted from their historical data. A conventional method of level prediction with a pattern recognition approach was performed by first predicting the actual numerical values using typical pattern-based regression models, hen classifying them into pattern levels (e.g., low, average, and high). A proposed prediction with pattern recognition scheme was developed to directly predict the desired levels using simpler classifier models without undergoing regression. The proposed pattern recognition classifier was compared to its regression method using a similar algorithm applied to a real-world energy dataset. A random forest (RF) algorithm which outperformed other widely used machine learning (ML) techniques in previous research was used in both methods. Both schemes used similar parameters for training and testing simulations. After 10-time cross training validation and five averaged repeated runs with random permutation per data splitting, the proposed classifier shows better computation speed and higher classification accuracy than the conventional method. However, when the number of its desired levels increases, its prediction accuracy seems to decrease and approaches the accuracy of the conventional method. The developed energy level prediction, which is computationally inexpensive and has a good classification performance, can serve as an alternative forecasting scheme.

Combined regression and classification artificial neural networks for sideslip angle estimation and road condition identification

This paper presents an Artificial Neural Network based algorithm to estimate the sideslip angle of a vehicle in different road conditions. The method is based on two parallel tasks: (a) estimation of the sideslip angle in three road conditions, namely dry, wet and icy, by means of three separate regression networks, and (b) identification of the road condition with a pattern recognition classifier allowing to select the correct estimation output among the three regression network outputs. The regression and classification networks are trained by means of datasets recorded during driving sessions conducted on a high-performance instrumented vehicle in the three road conditions and with different maneuvers and driving cycles. The algorithm has been deployed on the same vehicle to test the performance on the real application. The results are presented in terms of comparison between the estimated angle and its direct measurement obtained by a high precision optical sensor installed onboard. The resulting accuracy of the method is about 98.6%.

Wheelchair Control Based on Facial Gesture Recognition

In this article, an approach for controlling a wheelchair using gestures from the user's face is presented, particularly some commands for the basic control operations required for driving a wheelchair are recognized. In order to recognize the face gestures an Artificial Neural Network which is trained since it is one of the most successful classifiers in Pattern Recognition. In particular, the authors' proposed method is useful for controlling a wheelchair when the user has restricted (or zero) mobility in some parts of the body such as: legs, arms or hands. According to their experimental results, the proposed approach provides a successful tool for controlling a wheelchair through a Natural User Interface based on machine learning.

Cross-validation of a classification method applied in a database of sEMG contractions collected in a body interaction videogame

paper presents the evaluation and cross-validation of four pattern recognition classifiers, with the objective of finding the best one for classify surface electromyography (sEMG) signals combined with information extracted from videogame’s variables. The classifiers, a linear classifier, a quadratic classifier, a k-nearest-neighbor classifier, and a support vector machine, were computed on a data matrix created with the recorder signal collected from 12 subjects in a body interaction videogame that used a sEMG as a control strategy for upper limbs virtual rehabilitation. Although the classifiers of sEMG signal had a widespread study, there is not evidence of how to deal with the information of sEMG signals combined with game variables. The classification task is related with discern each subject as “good player” or “bad player”, looking to following the performance of the videogame’s users through the game sessions. A cross-validation of 10 iterations was computed, PCA and Relieff were used as feature extraction and selection methods. The evaluation was developed using the percentage of accuracy, defined as the well-predicted points. The best accuracy in the classification task was found using an SVM with a misclassification parameter of 400 and an RBF kernel regularization parameter of 60. Base on this result, the SVMs showed to be the appropriate classifier to be used on sEMG signal combined with videogame variables and should be implemented to follow the user’s performance.

The Price of Artificial Intelligence

Summary Introduction : Whilst general artificial intelligence (AI) is yet to appear, today’s narrow AI is already good enough to transform much of healthcare over the next two decades. Objective : There is much discussion of the potential benefits of AI in healthcare and this paper reviews the cost that may need to be paid for these benefits, including changes in the way healthcare is practiced, patients are engaged, medical records are created, and work is reimbursed. Results : Whilst AI will be applied to classic pattern recognition tasks like diagnosis or treatment recommendation, it is likely to be as disruptive to clinical work as it is to care delivery. Digital scribe systems that use AI to automatically create electronic health records promise great efficiency for clinicians but may lead to potentially very different types of clinical records and workflows. In disciplines like radiology, AI is likely to see image interpretation become an automated process with diminishing human engagement. Primary care is also being disrupted by AI-enabled services that automate triage, along with services such as telemedical consultations. This altered future may necessarily see an economic change where clinicians are increasingly reimbursed for value, and AI is reimbursed at a much lower cost for volume. Conclusion : AI is likely to be associated with some of the biggest changes we will see in healthcare in our lifetime. To fully engage with this change brings promise of the greatest reward. To not engage is to pay the highest price.

Partial Discharge Pattern Recognition Based on Synchrosqueezing Wavelet Transform and Multi-Scale Characteristic Parameters

Aiming at the high dimension of the characteristic of partial discharge and its high sensitivity to noise, firstly, the Synchrosqueezing wavelet transform is used to decompose the four typical partial discharge signals of transformers to overcome the defects of spectrum aliasing and energy leakage between real wavelet packet decomposition sub-bands.; Then, using the difference of energy and complexity of PD signals at different decomposition scales, the parameters of multi-scale energy and multi-scale energy spectrum entropy are used as the feature quantity of discharge type identification; Finally, the extracted features support vector machine classifier for discharge pattern recognition. Experimental results show, the proposed method can achieve better recognition than EMD and wavelet packet decomposition, and proves the effectiveness of the proposed method.

Super-Low Dose Lipopolysaccharide Dysregulates Neutrophil Migratory Decision-Making.

Neutrophils are the first responders to infection and play a pivotal role in many inflammatory diseases, including sepsis. Recent studies have shown that lipopolysaccharide (LPS), a classical pattern recognition molecule, dynamically programs innate immune responses. In this study, we show that pre-treatment with super-low levels of LPS [1 ng/mL] significantly dysregulate neutrophil migratory phenotypes, including spontaneous migration and altering neutrophil decision-making. To quantify neutrophil migratory decision-making with single-cell resolution, we developed a novel microfluidic competitive chemotaxis-chip (μC3) that exposes cells in a central channel to competing chemoattractant gradients. In this reductionist approach, we use two chemoattractants: a pro-resolution (N-Formyl-Met-Leu-Phe, fMLP) and pro-inflammatory (Leukotriene B4, LTB4) chemoattractant to model how a neutrophil makes a decision to move toward an end target chemoattractant (e.g., bacterial infection) vs. an intermediary chemoattractant (e.g., inflammatory signal). We demonstrate that naïve neutrophils migrate toward the primary end target signal in higher percentages than toward the secondary intermediary signal. As expected, we found that training with high dose LPS [100 ng/mL] influences a higher percentage of neutrophils to migrate toward the end target signal, while reducing the percentage of neutrophils that migrate toward the intermediary signal. Surprisingly, super-low dose LPS [1 ng/mL] significantly changes the ratios of migrating cells and an increased percentage of cells migrate toward the intermediary signal. Significantly, there was also an increase in the numbers of spontaneously migrating neutrophils after treatment with super-low dose LPS. These results shed light onto the directional migratory decision-making of neutrophils exposed to inflammatory training signals. Understanding these mechanisms may lead to the development of pro-resolution therapies that correct the neutrophil compass and reduce off-target organ damage.

Dynamic Movement of American English Monophthongs in the hVd Syllable: Evidence from Pattern Recognition Modeling with Cepstral/Spectal Parameters

Traditionally the steady-state central section of the vowel length has been assumed to characterize the vowel quality. However, since Peterson and Barney (1952), this position has been challenged especially for American English monophthongal vowels. In this paper, introduced are low-ordered 12 mel-scale frequency cepstral coefficients (MFCC), which can characterize the shape of the oral cavity filter for monophthongal vowel production in the mel-scale domain. Four pattern recognition classification models are fitted to the measurements of spectral and cepstral parameters at multiple sections of the vowel duration along with F0, Gender and Duration for the AE vowel signals in the hVd syllable in Hillenbrand et al. (1995). It turns out that pattern recognition classifiers with the cepstral properties outperform those with spectral properties, reaching the perception level of American English listeners’.

Technical diagnostic system in the maintenance of turbomachinery for ammonia synthesis in the process industries

Technical maintenance of machines and equipment in processing industry requires elaborate technical diagnostics systems to recognize the current state and forecast their future state. Creating such a system is a complex task due to multiple factors, with aging in aggressive exploitation environment being an important one. Statistical pattern recognition systems are very suitable to solve problems of technical diagnostics as they produce quantitative estimates of the states. We present the use of a hybrid Bayesian pattern recognition classifier that utilizes statistical and fuzzy paradigms and expresses the measurement information with four types of features (discrete, pseudo-discrete, multi-normal and independent continuous). It uses frequentist and subjective information (from training samples and expert opinion respectively) to identify the unknown parameters of the conditional likelihood density functions of each technical state. We discuss possible sources to collect learning information, and different methods to represent it. The classifier uses three different methods for parameter estimation of the conditional likelihood densities using data fusion. The classification is realised as a discriminant non-linear machine, which incorporates fuzzy approaches at different levels. We develop a novel algorithm for fault prediction without dynamic learning with four possible types of answers. A detailed example of technical diagnostics system for classification and prediction of states of turbomachinery for ammonia synthesis is presented. For the journal bearing diagnostics, we introduce modification of the hybrid Bayesian classifier using pseudo-priors to incorporate rule-based knowledge and improve the classification.

Microscopic Machine Vision Based Degradation Monitoring of Low-Voltage Electromagnetic Coil Insulation Using Ensemble Learning in a Membrane Computing Framework

In this paper, a novel microscopic machine vision system is proposed to solve a degradation monitoring problem of low-voltage electromagnetic coil insulation in practical industrial fields, where an ensemble learning approach in a compound membrane computing framework is newly introduced. This membrane computing framework is constituted by eight layers, 29 membranes, 72 objects, and 35 rules. In this framework, multiple machine learning methods, including classical pattern recognition methods and novel deep learning methods, are tested and compared. First, the most optimal feature extraction approaches are selected. Then, the selected approaches are fused together to achieve an even better monitoring performance. Third, a large number of experiments are used to evaluate and prove the usefulness and potential of the proposed system, where a mean accuracy of 61.4% is achieved on 1035 validation images of six degradation states with single state matching, and mean accuracies of 61.0% and 77.4% are achieved on 622 test images of six degradation states with single state matching and state range matching, respectively. Finally, a mechanical device is designed to apply the system to real industrial tasks.

DEGRADATION STAGE RECOGNITION OF BEARING WITHIN LIFE CYCLE

Bearings are important components in nuclear plant system, which are widely applied in different types of pump, electromotor and generator. In long term operation status, bearing performance would degenerate due to fatigue and bad lubrication. The damage of bearings would influence the performance and even the safety of devices. Therefore, it is extremely essential to obtain the operational condition information of bearings during the entire life cycle. In order to monitor and identify the bearing conditions in the entire life cycle, a degradation stage recognition method is proposed in this study. The recognition method consists of 4 main steps: firstly, extract a part of samples from the four stages (normal, early fault, medium fault, terminal fault) of bearing life cycle data set; secondly, extract the signal feature with multiscale permutation entropy technique; then, using the dimensional reduced sample features to train a pattern recognition classifier; finally, applying the classifier to identify the signal samples from the life cycle data set. With this automatic recognition method, the performance and degradation stages in the entire life cycle of bearings can be efficiently and effectively identified without additional manual assistance. The feasibility of this method depends on both of the effective recognition technique and the completion of data for classifier training. Thence, the processing technique improvement and the sustained operation data collection are indispensable for extending the method application to other components and devices.

A MobileNets Convolutional Neural Network for GIS Partial Discharge Pattern Recognition in the Ubiquitous Power Internet of Things Context: Optimization, Comparison, and Application

With the construction and promotion of the Ubiquitous Power Internet of Things (UPIoT), it is an increasingly urgent challenge to comprehensively improve the recognition accuracy of the gas-insulated switchgear (GIS) partial discharge (PD), and to incorporate the model into UPIoT intelligent terminals supported by edge computing in embedded systems. Therefore, this paper proposes a novel MobileNets convolutional neural network (MCNN) model to identify the GIS PD patterns. We first construct the PD pattern recognition classification datasets by means of experiments and FDTD simulation, and also preprocess images via binarization processing. After constructing the MCNN model, depthwise separable convolutions and an inverse residual structure are adopted to deal with the vanishing gradient of the deep convolutional neural network (DCNN) in the GIS PD pattern recognition process. Then, through the graphics standardization process, the MCNN model is trained and tested. The whole training process is visualized by Tensorboard. Compared with other deep learning models and traditional machine learning methods, MCNN particularly stands out in recognition accuracy and time consumption with a 96.5% overall recognition rate and merely 7.3 seconds in training time. This research explores how to optimize the model by improving the recognition accuracy, and by reducing its computing load, storage space and energy consumption for better incorporation into intelligent terminals in the UPIoT context.

Hepatitis C: From inflammatory pathogenesis to anti-inflammatory/hepatoprotective therapy.

Hepatitis C virus (HCV) infection commonly causes progressive liver diseases that deteriorate from chronic inflammation to fibrosis, cirrhosis and even to hepatocellular carcinoma. A long-term, persistent and uncontrolled inflammatory response is a hallmark of these diseases and further leads to hepatic injury and more severe disease progression. The levels of inflammatory cytokines and chemokines change with the states of infection and treatment, and therefore, they may serve as candidate biomarkers for disease progression and therapeutic effects. The mechanisms of HCV-induced inflammation involve classic pathogen pattern recognition, inflammasome activation, intrahepatic inflammatory cascade response, and oxidative and endoplasmic reticulum stress. Direct-acting antivirals (DAAs) are the first-choice therapy for effectively eliminating HCV, but DAAs alone are not sufficient to block the uncontrolled inflammation and severe liver injury in HCV-infected individuals. Some patients who achieve a sustained virologic response after DAA therapy are still at a long-term risk for progression to liver cirrhosis and hepatocellular carcinoma. Therefore, coupling with anti-inflammatory/hepatoprotective agents with anti-HCV effects is a promising therapeutic regimen for these patients during or after treatment with DAAs. In this review, we discuss the relationship between inflammatory mediators and HCV infection, summarize the mechanisms of HCV-induced inflammation, and describe the potential roles of anti-inflammatory/hepatoprotective drugs with anti-HCV activity in the treatment of advanced HCV infection.

Data-driven techniques for fault diagnosis in power generation plants based on solid oxide fuel cells

The development of fault diagnosis systems able to early detect and identify any malfunctioning is of great importance towards the diffusion of energy conversion plants based on solid oxide fuel cells. Because the traditional model-based schemes for the diagnosis demonstrated a poor fault identification capability (especially when many operating conditions and fault sizes are combined), hybrid schemes were proposed that, incorporating pattern recognition classifiers, achieved significant performance improvements. However, like in the model-based case, the hybrid schemes require the running of the plant model in parallel to the real plant functioning. In this study, a data-driven fault diagnosis scheme is presented, in which the plant model is used only off-line, for the training of the classifier. Both support vector machine and random forest classifiers are proposed and tested. It is demonstrated that, when a support vector machine classifier is deployed, the data-driven system outperforms the hybrid systems in both fault detection and fault identification. High-level performance is achieved despite the large number of combinations among working conditions and fault sizes used for testing the system.

A 2.2-&lt;inline-formula&gt; &lt;tex-math notation="LaTeX"&gt;$\mu$ &lt;/tex-math&gt; &lt;/inline-formula&gt;W Cognitive Always-On Wake-Up Circuit for Event-Driven Duty-Cycling of IoT Sensor Nodes

We report an always-on event-driven asynchronous wake-up circuit with trainable pattern recognition capabilities to duty-cycle power-constrained Internet-of-Things (IoT) sensor nodes. The wake-up circuit is based on a level-crossing analog-to-digital converter (LC-ADC) employed as a feature-extraction block with automatic activity-sampling rate scaling behavior. A novel asynchronous digital logic classifier for sequential pattern recognition is presented. It is driven by the LC-ADC activity and trained to minimize classification errors due to falsely detected events. As proof-of-concept, a prototype of the wake-up circuit is fabricated in 130nm CMOS technology within 0.054 mm2 of active area, covering up to 2.6 kHz of input signal bandwidth. The prototype has been first validated by interfacing it with a commercial accelerometer to classify hand gestures in real-time, reaching 81% of accuracy with only 2.2 $\mu \text{W}$ at 1-V supply. To highlight the flexibility of the design, a second application, detecting pathologic electrocardiogram beats is also discussed.

A CNN-SVM combined model for pattern recognition of knee motion using mechanomyography signals

The commonly used classifiers for pattern recognition of human motion, like backpropagation neural network (BPNN) and support vector machine (SVM), usually implement the classification by extracting some hand-crafted features from the human biological signals. These features generally require the domain knowledge for researchers to be designed and take a long time to be tested and selected for high classification performance. In contrast, convolutional neural network (CNN), which has been widely applied to computer vision, can learn to automatically extract features from the training data by means of convolution and subsampling, but CNN training usually requires large sample data and has the overfitting problem. On the other hand, SVM has good generalization ability and can solve the small sample problem. Therefore, we proposed a CNN-SVM combined model to make use of their advantages. In this paper, we detected 4-channel mechanomyography (MMG) signals from the thigh muscles and fed them in the form of time series signals to the CNN-SVM combined model for the pattern recognition of knee motion. Compared with the common classifier performing the classification with hand-crafted features, the CNN-SVM combined model could automatically extract features using CNN, and better improved the generalization ability of CNN and the classification accuracy by means of combining the SVM. This study would provide reference for human motion recognition using other time series signals and further expand the application fields of CNN.

A Latent Class Pattern Recognition and Data Quality Assessment of Non-Commute Long-Distance Travel in California

This study analyzes 8-week long-distance travel records from the California Household Travel Survey for completeness and identifies general types of non-commute long-distance tours using Latent Class Analysis. Likely due to the difficulty of gathering data of this kind, there has been relatively limited study of non-commute long-distance travel, despite the substantial contribution to many households’ greenhouse gas emissions and travel expenses. The California Household Travel Survey includes a valuable long-distance 8-week travel dataset, but this study identifies several possible shortcomings in the dataset. Of particular importance is a severe underreporting of shorter trips, which may result from a mix of respondent forgetfulness and survey fatigue. Despite the issues with the data, latent class cluster analysis was able to identify five distinct, informative patterns of long-distance travel. This analysis shows that long-distance tours for vacation, business travel, medical, and shopping are substantially distinct in terms of their travel characteristics and correspond to different combinations of other activities in the tour, and they are done by different types of households. The method used here to identify the typology of long-distance travel can be easily expanded to include a variety of other explanatory variables of this type of behavior in more focused data collection settings.

Improving Myoelectric Pattern Recognition Robustness to Electrode Shift by Autoencoder.

It is evident that the electrode shift will result in a degradation of myoelectric pattern recognition classification accuracy, which is inevitable during the prosthetic socket donning and doffing. To cope with this limitation, we propose an unsupervised feature extraction method called sparse autoencoder (SAE) to extract the robust spatial structure and correlation of high density (HD) electromyography (EMG). The algorithm is evaluated on nine intact-limbed subjects and one amputee. The experimental results show that SAE achieves lower classification error without shift, and significantly decrease the sensitivity to electrode shift with ±1 cm compared with the timedomain and autoregressive features (TDAR). Furthermore, SAE is not sensitive to the shift direction that is perpendicular to the muscle fibers. The promising results of this study make great contribution to promoting the applications of pattern recognition based myoelectric control system in real-world condition.