Search

Graphology is a method of assessment of human personality from individual's handwriting style. The personality in terms of various attributes like fear, honesty, work habits, social skills etc. can be profiled using Graphology. The personality profiling is based on various features like line spacing, margins, slant, letter size, text density etc. Letter t and i are very important personality trait indicators in graphology. Letter t is an indicator of will in individuals. Letter i is an indicator of social connectivity in individuals. This work proposes a deep learning classifier to classify the specific type of t and i patterns. Convolutional neural network is trained to extract the t and i features. The features extracted from t and i handwriting patterns are classified using the ensemble of Deep learning classifier. The performance of the proposed ensemble classification is tested against different handwritten documents. The proposed solution is able to achieve about 90% classification accuracy

Vehicle segmentation, traffic density estimation and vehicle tracking are some of the important functionalities in smart traffic management. Segmentation of vehicles from traffic surveillance videos can help to realize many applications like speed monitoring, traffic estimation etc. Vehicle segmentation from surveillance videos becomes very challenging due to presence of occlusions, cluttered backgrounds and traffic density variations. In this work, we propose a deep learning adapted active net segmentation model for vehicle segmentation. The proposed solution involves three stages: adaptive background model based subtraction, active net sub netting with CNN, refined from CNN results with ETAN optimization. Adaptive background modeling is based on an adaptive gain function which is constructed from the pixels of the frames in the video. The gain function is able to compensate for the shadow and illumination problems affecting the vehicle segmentation. Deep learning assisted topological active net deformable model is able to provide higher segmentation accuracy in presence of occlusions, cluttered backgrounds and traffic density variations.

In medical imaging, segmentation plays a vital role towards the interpretation of X-ray images where salient features are extracted with the help of image segmentation. Without undergoing surgery, clinicians employ various modalities ranging from X-rays and CT-Scans to ultrasonography, and other imaging techniques to visualise and examine interior human body organ and structures. To ensure appropriate convergence, training a deep convolutional neural network (CNN) from scratch is tough since it requires more computational time, a big amount of labelled training data and a considerable degree of experience. Fine-tuning a CNN that has been pre-trained using, for instance, a huge set of labelled medical datasets, is a viable alternative. In this paper, a comparative study was done using pre-trained models such as VGG-19 and ResNet-50 as against training from scratch. To reduce overfitting, data augmentation and dropout regularization was used. With a recall of 92.03%, our analysis showed that the pre-trained models with proper finetuning was comparable with Iyke-Net, a CNN trained from scratch.

This article implements a new control method for a microgrid configuration comprising wind energy generation system, a solar PV array, a battery, local loads, and the utility grid, which provides a seamless transition of power during on-grid to off-grid operations and vice versa. This microgrid uses synchronous reluctance generator (SynRG) operated by a position (mechanical) sensor-less field-oriented control, for electric power generation from a variable speed wind turbine. The output of the solar PV array is connected to the dc-link of the microgrid through a dc–dc boost converter, whose output voltage is regulated at the desired value by the battery bank. For smooth synchronization and fast dynamic response, a fourth-order generalized integrator-based frequency-locked-loop with pre-filter (C-FGI-FLL) is implemented in this work. The use of C-FGI-FLL enhances the filtering capabilities of FGI-FLL under weak grid conditions, without affecting the computational burden. The C-FGI-FLL generates filtered input and its quadrature component, and these are used for the extraction of positive sequence components of the grid voltages. For faster dynamic response during load variations, C-FGI filters are used in each phase so that fundamental peaks of load currents are extracted accurately without any delay. The accurate and quick performance of this filter makes the system dynamics fast without deteriorating the performance. Simulation and experimental validations of this microgrid have been carried out both for dynamic and steady-state conditions.

Focus of the current study is to explore and analyse textual data in the form of incidents in pharmaceutical industry using topic modelling. Topic modelling applied in the current study is based on Latent Dirichlet Allocation. The proposed model is applied on a corpus containing 190 incidents to retrieve key words with highest probability of occurrence. It is used to form informative topics related to incidents.

In this article, a new control methodology is implemented for a microgrid consisting of wind-battery and the grid, which operates both under grid-connected mode and standalone mode. A synchronous reluctance generator (SynRG) with a new mechanical sensorless field-oriented control is used in the wind energy generation system (WEGS). The major issues present in the sensorless control of SynRG, due to the use of conventional integrator or low-pass filter during the flux estimation, are addressed in this article by implementing a fourth-order flux estimator. To compensate for the intermittent nature of the power generated from the WEGS, and to improve the reliability of the system pertaining to power availability, energy storage is included in the system. Controlled operation of the battery is ensured by using a bidirectional dc–dc converter, to enhance the longevity of the battery. Voltage source converters (VSCs) connected in a back-to-back configuration with a common dc-link is selected here so that a complete decoupling is achieved between the grid and WEGS. Grid-side VSC (GSC) is controlled, so that power quality at the point of common coupling is improved. The positive sequence components of grid voltages are extracted using observers to implement GSC control to avoid the adverse effects of unbalance and harmonic content in grid voltages. Design, modeling, and simulation of the system are done in MATLAB/SIMULINK platform, and a prototype developed in the laboratory is used for real-time validation.

PurposeSupply chain efficiency can be enhanced by integrating the activities in supply chain through digitalization. Advancements in digital technologies has facilitated in designing robust and dynamic supply chain by bringing in efficiency, transparency and reduction in lead times. This research tries to identify and prioritize the customer requirements and design requirements for effective integration of supply chain through digitalization.Design/methodology/approachThe key nine customer requirements and 16 design requirements applicable for an electronics company were shortlisted in consultation with the experts from the company and academia. An integrated analytic network process (ANP) and quality function deployment (QFD) methodology has been applied for prioritizing the customer and design requirements. The relative importance and interdependence of these requirements were identified and a House of Quality (HOQ) is constructed.FindingsThe HOQ constructed has prioritized and identified interrelationships among customer requirements and design requirements for effective supply chain digitalization. These findings could be effectively used by managers for planning the objectives on long-term, medium-term and short-term basis.Originality/valueThis study tries to bridge the gap of identifying and prioritizing the design and customer requirements for effective supply chain integration through digitalization. The results could aid practicing managers and academicians in decision-making on supply chain digitalization process.

An efficient control of a grid integrated microgrid with wind energy generating system (WEGS) and a battery storage, is implemented in this work. The quality of power injected to the utility is deteriorated with the level of unbalance and/or distortions present in the grid voltages. To compensate for the abnormalities, such as unbalance and distortions in the grid voltages, positive sequence components (PSCs) of grid voltages are extracted from their filtered $ \alpha \beta$ components. This work implements hybrid generalized integrator (HGI) filters for filtering the $ \alpha \beta$ components of grid voltages and also provides its 90 $\circ$ delayed signals, which are useful in generating PSCs. Phase and frequency of the grid currents are decided by the unit active templates derived from the PSCs of the grid voltages, therefore, high-quality grid currents are injected into the grid even for abnormal grid conditions. For faster dynamics during variations in load power, HGI filters are used for quick and accurate estimation of fundamental components of load currents. Intermittent power generation from WEGS is compensated for by integrating a battery bank into the system. Along with high-quality power injection, the grid side converter operates the microgrid in off-grid mode (during grid outages) and provides smooth grid integration once the grid reappears. A 3.7 kW speed sensorless synchronous reluctance generator is used in WEGS, which is controlled by the machine side converter. The control techniques are validated in MATLAB /Simulink platform, and experimental validation is done on a prototype developed in the laboratory.

This article presents a grid-integrated hybrid renewable energy sources based system comprising a solar photovoltaic (PV) array and a wind energy conversion system (WECS). The WECS uses a position-sensorless synchronous reluctance generator (SynRG) for the electric power generation from the wind turbine (WT), wherein, a sensorless field-oriented control (FOC) is made use of for the maximum power extraction (MPE). A second-order flux estimation (SFE) method along with frequency-locked loop (FLL) is utilized for the accurate flux estimation from the SynRG stator voltages and currents. A set of back-to-back connected three-phase two leg voltage source converter (VSC) topology is selected for the grid integration of WECS. This system has a common dc link where the solar PV array and the machine-side VSC (MSC) of the wind generator, are directly connected. The power output from the solar PV array and WECS, is shared between the grid and the local loads. The maximum power generation from SynRG in the WECS, is achieved by operating the SynRG at the speed estimated by the MPE algorithm. The maximum power is drawn from the solar PV array by adjusting the dc-link voltage, which is decided by the algorithm. For the proper power control and power quality improvement, the grid-side converter (GSC) is adequately controlled by implementing an observer-based control technique. The real-time validation of the system, is carried out using a developed laboratory prototype.

A new wind energy conversion system (WECS), with a position sensor-less synchronous reluctance generator (SynRG), integrated to the grid is proposed in this paper. Simple, brushless, compact structure of SynRG along with a full-scale converter makes it suitable for wind power applications. SynRG offers a fast dynamic response due to the absence of permanent magnets and windings in the rotor thus eliminating rotor losses completely. The torque pulsation from wind turbine (WT) are prevented from being transmitted to the grid by making use of a back-to-back (BTB) connected full-scale converters. The WECS and the grid are sharing a common DC-link through their respective voltage source converters (VSCs). The VSC connected to the machine side provides the required reactive power to the SynRG and operates it at a speed decided by the maximum power point (MPP) algorithm. A rotor position/speed sensor-less field-oriented control is implemented for controlling the SynRG. A fast, accurate, and oscillation-less adaptive observer based algorithm is introduced for the grid integration of WECS. MATLAB®/Simulink platform is used for the modeling and simulation of the system and a laboratory prototype is used for its experimental validation.

Transmitting image data without losing significant information is challenging for any network application especially when large color images are transmitted through TCP communication protocol. This is due to network limitations such as buffer overflow, underflow and network traffic flow etc. This paper presents a new method for image size reduction such that the network can transmit data without much loss of information, and hence, quality. The proposed method obtains bit planes for the color input images, which results in eight binary planes. Unlike the existing bit plane based image size reduction methods, which assume that the most significant plane or some other planes contain useful information, the proposed method finds the plane that contains dominant information automatically. For each plane, the proposed method explores the saliency that finds dominant information based on Markov Chain Process and similarity estimation between neighbor pixels. To reduce computational burden, we use Canny edge maps of the saliency of the planes for feature extraction. We propose to explore ring-growing concept for the edge maps to study the spatial distribution of saliency, locally. The proposed method detects the plane based on statistics of saliency distribution. To validate the step of plane detection, we estimate transmission error caused during data transmission through TCP communication protocol for the images at sending and receiving ends. Experimental results on plane detection show that the proposed method is better than the existing methods in terms of detection rate. Our experiments on image transmission through TCP communication protocol show that the proposed method outperforms the existing methods in terms of error estimation and quality analysis. Furthermore, experiments are conducted to analyze packet loss in terms of number of duplicate acknowledgements and retransmission during packets transmission for the color, edge and plane to show that transmitting plane images improves network performance in terms of less number of duplicate acknowledgement, retransmission and time taken in seconds.

This paper presents a comparison of several artificial intelligence (AI) based binary classifiers for detecting the integrity of data obtained from Internet of Things (IoT) enabled wearable sensors. Detecting the integrity of data at the network edge facilitates the elimination of corrupted or unusable data, which translates to a lower amount of data stored and transmitted. This reduces the storage and power requirements of IoT devices without a reduction in functionality. In this work, we explore several machine learning-based classifiers to check the integrity of electrocardiogram (ECG) data. The feature vectors are derived from low complexity kurtosis and skewness based Signal Quality Indices (SQIs). From the experiments, it is found that a bagged ensemble of 3 neural networks achieves the highest detection accuracy of 99.47%. We also estimated the complexity and power consumed by the various classifier implementations and classifier fusion implementations. The energy consumed by the ensemble classifier was estimated to be around 0.039 nJ.

Stroke is widely considered as the second most common cause of mortality. The adverse consequences of stroke have led to global interest and work for improving the management and diagnosis of stroke. Various techniques for data mining have been used globally for accurate prediction of occurrence of stroke based on the risk factors that are associated with the electronic health care records (EHRs) of the patients. In particular, EHRs routinely contain several thousands of features and most of them are redundant and irrelevant that need to be discarded to enhance the prediction accuracy. The choice of feature-selection methods can help in improving the prediction accuracy of the model and efficient data management of the archived input features. In this paper, we systematically analyze the various features in EHR records for the detection of stroke. We propose a novel rough-set based technique for ranking the importance of the various EHR records in detecting stroke. Unlike the conventional rough-set techniques, our proposed technique can be applied on any dataset that comprises binary feature sets. We evaluated our proposed method in a publicly available dataset of EHR, and concluded that age, average glucose level, heart disease, and hypertension were the most essential attributes for detecting stroke in patients. Furthermore, we benchmarked the proposed technique with other popular feature-selection techniques. We obtained the best performance in ranking the importance of individual features in detecting stroke., Accepted in IEEE Access, 2020

This paper aims to reduce the power consumption of electrocardiography based wearable healthcare devices, by introducing power reduction approaches and considerations at system level design, where we have the highest potential to influence power. It focuses, in particular, on algorithm design and implementation, data acquisition, and transmission under constrained resources. A thorough investigation of the suitability of nine existing algorithms for on-sensor QRS feature detection is conducted, with respect to metrics such as sensitivity, positive predictivity, power consumption, parameter choice and time delay. Optimisation of data acquisition on CPU-based IoT systems is performed, and the current consumption is reduced by a factor of 3 using a combination of direct memory access (DMA) list approach and low-level register manipulations for task delegation. The acquisition data rate, sampling rate, buffer and batch size are also optimised. To reduce the power consumption by data transmission, the effect of on-sensor versus off-sensor processing is investigated. While focusing on CPU-based systems with experiments performed on a generic low-power wearable platform, the design optimisation and considerations proposed in this work could be extended to custom designs and allow further investigation into QRS detection algorithm optimisation for wearable devices.

Efficient Inter-Organizational Information Systems (IOIS) have become the backbone of modern supply chains. IOIS can be used to plan, coordinate, collaborate and integrate supply chains for attaining competitive advantage. The speed of innovative technology evolvement, lack of clarity, and delay in taking appropriate managerial and strategic decisions for adopting IOIS demand further research in this area. The robust advancement in digital technologies stresses a proper decision model for the IOIS adoption process. This paper provides a novel model for selecting the best IOIS alternative by considering the contents, scope, and critical decision-making factors affecting supply chain integration. Twelve decision-making factors affecting IOIS selection were identified and categorized under four significant dimensions: technological, operational, application, and innovative for effective decision-making. Study results reveal that project completion time is the most relevant criterion, followed by digital technology enablers and the financial resources required to select IOIS alternatives.

This paper presents the designing of low altitude long endurance (LALE) VTOL UAV aircraft for particularly given mission requirements. To fulfil the mission requirements of VTOL UAV, an analytical methodology is established. The analytical design and performance calculation is achieved by MATLAB(version 2020). On behalf of the designed value, the VTOL UAV aircraft is replicated into X-Plane software. The designed X-Plane VTOL UAV aircraft is used for the detailed simulation and validation of the calculated performance. In X-plane, detailed modeling of individual components is done, and flight tests are performed in different environments for endurance benchmark. A VTOL has five primary flight regimes, namely, take-off, forward transition, cruise, reverse transition, and landing; for all these flight regimes, the state of the charge of the battery is studied. Paper also has the interface of simulation strategy to control the fixed-wing VTOL UAV aircraft in manual simulation.

This special section presents seven state-of-the-art design techniques for AI hardware accelerators that lead to substantial improvements on energy efficiency, reconfigurability, and resource utilization.

In this paper we investigate the cues of COVID-19 on sustained phonation of Vowel-/i/, deep breathing and number counting data of the DiCOVA dataset. We use an ensemble of classifiers trained on different features, namely, super-vectors, formants, harmonics and MFCC features. We fit a two-class Weighted SVM classifier to separate the COVID-19 audio from Non-COVID-19 audio. Weighted penalties help mitigate the challenge of class imbalance in the dataset. The results are reported on the stationary (breathing, Vowel-/i/) and nonstationary( counting data) data using individual and combination of features on each type of utterance. We find that the Formant information plays a crucial role in classification. The proposed system resulted in an AUC score of 0.734 for cross validation, and 0.717 for evaluation dataset. Copyright © 2021 ISCA.

Top-down saliency models produce a probability map that peaks at target locations specified by a task/goal such as object detection. They are usually trained in a fully supervised setting involving pixel-level annotations of objects. We propose a weakly supervised top-down saliency framework using only binary labels that indicate the presence/absence of an object in an image. First, the probabilistic contribution of each image region to the confidence of a CNN-based image classifier is computed through a backtracking strategy to produce top-down saliency. From a set of saliency maps of an image produced by fast bottom-up saliency approaches, we select the best saliency map suitable for the top-down task. The selected bottom-up saliency map is combined with the top-down saliency map. Features having high combined saliency are used to train a linear SVM classifier to estimate feature saliency. This is integrated with combined saliency and further refined through a multi-scale superpixel-averaging of saliency map. We evaluate the performance of the proposed weakly supervised topdown saliency and achieve comparable performance with fully supervised approaches. Experiments are carried out on seven challenging datasets and quantitative results are compared with 40 closely related approaches across 4 different applications., Comment: 14 pages, 7 figures

The open-source event-driven ECG dataset is available at https://github.com/jedaiproject/Open-Source-Event-Driven-ECG-Dataset; International audience; In this article, non-uniformly sampled electrocardiogram (ECG) signals obtained from level-crossing analog-to-digital converters (LC-ADCs) are analyzed for event-driven classification and compression performance. The signal compression results show that it is important to assess the distortion in event-driven signals when simulating LC-ADC models, especially at lower resolutions and larger quantization steps. The effects of varying the LC-ADC parameters for the application of cardiac arrhythmia classifiers are also assessed using an artificial neural network (ANN) and the MIT-BIH Arrhythmia Database. In comparison with uniformly-sampled data, it is possible to achieve comparable classification accuracy at a much lower complexity with event-driven ECG signals. The results show the best event-driven model achieves over 97% accuracy with 79% reduction in ANN complexity with signal-to-distortion ratio (S/D)≥21dB. For S/D

Over the past several years, the electrocardiogram (ECG) has been investigated for its uniqueness and potential to discriminate between individuals. This paper discusses how this discriminatory information can help in continuous user authentication by a wearable chest strap which uses dry electrodes to obtain a single lead ECG signal. To the best of the authors' knowledge, this is the first such work which deals with continuous authentication using a genuine wearable device as most prior works have either used medical equipment employing gel electrodes to obtain an ECG signal or have obtained an ECG signal through electrode positions that would not be feasible using a wearable device. Prior works have also mainly dealt with using the ECG signal for identification rather than verification, or dealt with using the ECG signal for discrete authentication. This paper presents a novel algorithm which uses QRS detection, weighted averaging, Discrete Cosine Transform (DCT), and a Support Vector Machine (SVM) classifier to determine whether the wearer of the device should be positively verified or not. Zero intrusion attempts were successful when tested on a database consisting of 33 subjects.

Internet of Things (IoT) enabled wearable sensors for health monitoring are widely used to reduce the cost of personal healthcare and improve quality of life. The sleep apnea-hypopnea syndrome, characterized by the abnormal reduction or pause in breathing, greatly affects the quality of sleep of an individual. This paper introduces a novel method for apnea detection (pause in breathing) from electrocardiogram (ECG) signals obtained from wearable devices. The novelty stems from the high resolution of apnea detection on a second-by-second basis, and this is achieved using a 1-dimensional convolutional neural network for feature extraction and detection of sleep apnea events. The proposed method exhibits an accuracy of 99.56% and a sensitivity of 96.05%. This model outperforms several lower resolution state-of-the-art apnea detection methods. The complexity of the proposed model is analyzed. We also analyze the feasibility of model pruning and binarization to reduce the resource requirements on a wearable IoT device. The pruned model with 80\% sparsity exhibited an accuracy of 97.34% and a sensitivity of 86.48%. The binarized model exhibited an accuracy of 75.59% and sensitivity of 63.23%. The performance of low complexity patient-specific models derived from the generic model is also studied to analyze the feasibility of retraining existing models to fit patient-specific requirements. The patient-specific models on average exhibited an accuracy of 97.79% and sensitivity of 92.23%. The source code for this work is made publicly available., Accepted for discussion at the IEEE International Symposium on Circuits and Systems (ISCAS) 2021

Wearable Artificial Intelligence-of-Things (AIoT) devices demand smart gadgets that are both resource and energy-efficient. In this paper, we explore efficient implementation of binary convolutional neural network employing function merging and block reuse techniques. The hardware implemented in field programmable gate array (FPGA) platform can classify ventricular beat in electrocardiogram achieving accuracy of 97.5%, sensitivity of 85.7%, specificity of 99.0%, precision of 92.3%, and F1-score of 88.9% while consuming only 10.5-µW of dynamic power dissipation.

Smart traffic management is being proposed for better management of traffic infrastructure and regulate traffic in smart cities. With surge of traffic density in many cities, smart traffic management becomes utmost necessity. Vehicle categorization, traffic density estimation and vehicle tracking are some of the important functionalities in smart traffic management. Vehicles must be categorized based on multiple levels like type, speed, direction of travel and vehicle attributes like color etc. for efficient tracking and traffic density estimation. Vehicle categorization becomes very challenging due to occlusions, cluttered backgrounds and traffic density variations. In this work, a traffic adaptive multi-level vehicle categorization using deep learning is proposed. The solution is designed to solve the problems in vehicle categorization in terms of occlusions, cluttered backgrounds.

This present study proposes a current controlled modulation method for brushless DC motors for electric vehicle. A quasi-square current control technique is used. Conventional controllers such as PI, PID controllers are used first, then the system is evaluated using Fuzzy Logic controller which further improves the system and also fuzzy controller is far better than other controllers due to the use of linguistic terms.

A real video surveillance is the most effective and powerful way to monitor a particular remote area and fixed-wing Unmanned aerial vehicles (UAVs) are the most popular UAV to complete this job. This paper proposed a hardware integration method of the vision system with a communication setup, which is mounted on our fixed-wing UAV. During the practical implementation of real time video surveillance systems, a lot of challenges like delay, image quality, range, etc., are encountered. Handling of these challenges are discussed in this paper.

Current deep learning-based texture recognition methods extract spatial orderless features from pre-trained deep learning models that are trained on large-scale image datasets. These methods either produce high dimensional features or have multiple steps like dictionary learning, feature encoding and dimension reduction. In this paper, we propose a novel end-to-end learning framework that not only overcomes these limitations, but also demonstrates faster learning. The proposed framework incorporates a residual pooling layer consisting of a residual encoding module and an aggregation module. The residual encoder preserves the spatial information for improved feature learning and the aggregation module generates orderless feature for classification through a simple averaging. The feature has the lowest dimension among previous deep texture recognition approaches, yet it achieves state-of-the-art performance on benchmark texture recognition datasets such as FMD, DTD, 4D Light and one industry dataset used for metal surface anomaly detection. Additionally, the proposed method obtains comparable results on the MIT-Indoor scene recognition dataset. Our codes are available at https://github.com/maoshangbo/DRP-Texture-Recognition. This work was conducted within the Rolls-Royce@NTU Corporate Lab under the project DACS 2.1: Artificial Intelligence (AI) for Smart Image Understanding with support from the Industry Alignment Fund (IAF) Singapore under the Corp Lab@University Scheme.

In this paper, hardware design and implementation to realize the effect of 3D sound with time-varying FIR filters are presented. 3D sound is a type of audio that encapsulates and recreates the effect identical to the way our ears normally experience. The spatial location of sound results in its three dimensional aspect. To synthesize it from a stereo recording, Head Related Transfer Functions (HRTFs), which describe the spectral behaviour of sounds coming from a particular direction are used. FIR filters derived from this transfer function are applied to the incoming sound, yielding spatial effect. The system was implemented using 180 nm technology libraries targeting an Application Specific Integrated Circuit (ASIC) and the functionality was validated in real-time on FPGA.

Wearable Internet of Things (IoT) devices are getting ubiquitous for continuous physiological data acquisition and health monitoring. This paper investigates an electrocardiogram (ECG) based biometric user authentication technique for IoT edge devices. A convolutional neural network (CNN) based deep learning technique for user authentication is proposed. The proposed technique achieves an authentication accuracy of 99.63% when tested with 290 subjects from Physionet PTB ECG database. To limit the complexity of the technique for IoT edge nodes, we applied optimisation techniques such as binarisation and approximation of the CNN weights. Accuracy-vs-time-complexity trade-off analysis is performed and results are presented for different optimisations. Our evaluations shows that the complexity-optimised method achieves 98.88% authentication accuracy with acceptable CPU cycles consumed.

The advent of low-cost wearable Internet of Things (IoT) sensors has made it possible to continuously acquire physiological signals such as electrocardiogram (ECG) for long durations. Techniques for automated analysis is essential for deriving intelligence from such a large quantity of data. This paper presents a 1-dimenslonal convolutional neural network (CNN) for heartbeat classification from ECG signals obtained from an ambulatory device. The proposed technique can classify heartbeats into 5 classes as specified in AAMI standard and was tested using the Physionet MIT-BIH Arrhythmia database. To address the imbalance of classes in the dataset we used the SMOTE algorithm to augment the dataset. The network was trained using the augmented data and achieved an accuracy of 98.12%, sensitivity of 98.07%, and a specificity of 98.29%.

This paper proposes an algorithm to process data of high-precision sensor array with data compression. To eliminate the jitter in pressure sensing, the Speeded-up Robust Features(SURF), Fast Approximate Nearest Neighbor Search Library(FLANN), RANdom SAmple Consensus(RANSAC), Gaussian filter and motion compensation are implemented. Moreover, data compression is conducted based on JPEG-LS and Run-length coding. As a result, the proposed algorithm achieves the jitter eliminating and high compression ratio.

Carpenters are key members and carpentry trades have a major role in the construction industry. Different hand and power tools are used to complete the jobs in this area. Various skill parameters are required for the proper usage of different tools, which are used in the construction industry. Based on the survey, choose one of the important tools used in this field that has conducted the experimentation. This paper describes the skill parameters involved in the proper use of the jigsaw tool and conducted an experimental study to analyze those skills varied between novices and experts. Based on the expert’s data, we could easily teach the novice for the proper usage of the tool. The expert’s data are set as a reference data for skilling the novices. This paper also discusses a comparison of expert’s data with the novice for proposing an assistive training system for jigsaw tools, which will accelerate the learning of use of the jigsaw tool.

This innovative practice category work in progress paper describes a project in a final-year un-dergraduate course on implementing a neural accelerator on an FPGA for edge computing. In our university, an undergraduate course on Embedded Hardware System Design introduces students to advanced hardware design techniques with the goal of integrating the created hard-ware into a complete system. Students learn concepts such as high-level synthesis (HLS), logic synthesis and physical design, with an emphasis on FPGA-based designs. They also understand bus systems such as Advanced eXtensible Interface (AXI) to interconnect the various components. The concepts are put into practice through a project. A series of labs provide scaffolding to students through the course of implementing the project. These labs take stu-dents systematically through an introduction to hardware-software co-design, hardware design, creation and interfacing of custom co-processors and HLS. Important hardware design and optimization concepts, as well as managing the data interaction between hardware and software were reinforced through the project. The project also provided many students with the opportunity to be introduced to neural networks and machine learning (ML). Quantitative and qualitative results from a survey indicate that students gained a lot of knowledge and experience through the course of the project. The current form of the project streams in data from a local computer to which the FPGA is connected. Future work includes true and direct cloud connectivity and improved use cases for making it a true Internet of Things (IoT) project.

Real-time implementations of image processing algorithms on embedded platforms are gaining importance. In this paper, we propose an Application Specific Integrated Circuit (ASIC) architecture for the perfect reconstruction of images using wavelets with a view to extending this to denoising and feature extraction of images. An architecture that implements the Dual-Tree Wavelet Transform is presented. The architecture features a 128×128 single-port block memory and its addressing schemes, a simple upsampling/downsampling method and a folding and adding mechanism. It is implemented using 180nm technology. The results show perfect reconstruction of 128×128 grayscale images with up to 1-bit error in pixel values when compared to the corresponding input images.

Many commercial and forensic applications of speech demand the extraction of information about the speaker characteristics, which falls into the broad category of speaker profiling. The speaker characteristics needed for profiling include physical traits of the speaker like height, age, and gender of the speaker along with the native language of the speaker. Many of the datasets available have only partial information for speaker profiling. In this paper, we attempt to overcome this limitation by developing a new dataset which has speech data from five different Indian languages along with English. The metadata information for speaker profiling applications like linguistic information, regional information, and physical characteristics of a speaker are also collected. We call this dataset as NITK-IISc Multilingual Multi-accent Speaker Profiling (NISP) dataset. The description of the dataset, potential applications, and baseline results for speaker profiling on this dataset are provided in this paper., 5pages, Initial version submitted to Interspeech2020

Ki-67 labelling index is a biomarker which is used across the world to predict the aggressiveness of cancer. To compute the Ki-67 index, pathologists normally count the tumour nuclei from the slide images manually; hence it is timeconsuming and is subject to inter pathologist variability. With the development of image processing and machine learning, many methods have been introduced for automatic Ki-67 estimation. But most of them require manual annotations and are restricted to one type of cancer. In this work, we propose a pooled Otsu's method to generate labels and train a semantic segmentation deep neural network (DNN). The output is postprocessed to find the Ki-67 index. Evaluation of two different types of cancer (bladder and breast cancer) results in a mean absolute error of 3.52%. The performance of the DNN trained with automatic labels is better than DNN trained with ground truth by an absolute value of 1.25%.

A low complexity interpolated finite impulse response (IFIR) based 18 band ANSI S1.11 1/3 octave digital filter bank for hearing aid is proposed and implemented using 65 nm UMC technology in this study. ANSI S1.11 specifications for 1/3 octave Class-2 filters are chosen as design criteria in the proposed method. The strict requirements at lower bands make the hardware implementation of ANSI S1.11 filter bank complex and difficult in a power critical application like hearing aid. In the proposed technique, the maximum margin available in the filter specifications is utilised in both upper and lower band edge specifications to reduce the filter order. IFIR technique was used to reduce the computations at lower bands which can be implemented in hardware efficiently without altering the sampling frequency of the input signal. Compared to other recent architectures, >50% reduction in total number of filter coefficients is achieved and the group delay is kept

Generally, surveillance refers to a single, known and mostly static point of interest that is observed for a predetermined amount of time. Similarly, reconnaissance implies a large area to be covered thereby requiring rapid mobility and capability to observe multiple points of interest. Real-time video surveillance is a good way to realize both surveillance and reconnaissance, which involves multiple challenges and complexities; viz., computational efficiency, latency, image quality, etc. The concept of utilizing a fixed-wing VTOL Unmanned Aerial Vehicle (UAV) is more appropriate than the conventional fixed-wing UAV or multi-rotors, in terms of the quality of visual imageries, increased operational range, reduced time to target and thereby reduced mission times, minimal dependencies on infrastructure, and so on. This research studies the application of a fixed-wing VTOL UAV for real-time low-latency monitoring systems for reconnaissance; thereby quantifying various benefits, including the practical performance of such a video surveillance system. The simulation required for this research is realized using a Robot Operating System (ROS) and the final model is validated using both hardware and software.

A Powered Paraglider, also known as Paramotor, has a ram-air inflated canopy in the shape of an aerofoil from which a payload, commonly called as Gondola, housing both propulsion system and control mechanism is suspended. It can lift heavy loads, is quick to setup for rapid launch, and is compact and light-weight, thereby making it ideal for military operations like tactical surveillance and cargo deployment. Paramotors are suitable for scenarios where stable and low speed flying capabilities are necessary. This paper presents a software architecture for guidance and control of light weight small scale Paramotors. For heading and altitude tracking, the system uses feedback compensated control laws. First, linear models are derived that describe both the Paramotor's longitudinal and lateral dynamics. Then, a six degree-of-freedom model is used to describe dynamics, weight, aerodynamic forces on payload and parafoil, aerodynamic moments, effect of apparent forces and moments, moments generated on the centre of mass by the forces exerted at the payload and parafoil. Then system identification based on simplified linear lateral and longitudinal models is used. These simplified linear models are used for designing control laws using classical frequency domain techniques. MATLAB/Simulink was used to simulate the performance of the proposed Paramotor controllers. It was found that the described approach is robust enough for designing control strategies to maintain stability in event of disturbances.

The paper describes the methodology and the design process of a Model Predictive Controller (MPC) that has been applied to solve Missile-on-Missile Interception. The MPC combines the control and guidance laws to create an ideal control law with much higher performance and guarantees strike with a wide range of target and missile parameters. The MPC is also able to handle actuator delay and hence is more practically suited than traditional guidance/control schemes. Also, MPC satisfies sensor/seeker angle constraints, actuator constraints, angle of attack constraints, and fin deflection angle constraints, thus making it more versatile and powerful than any analytical method-based control/guidance schemes. The only disadvantage is the higher computational cost requirements, which will be easily mitigated in the future with the advancements of powerful microprocessors. The performance of the MPC is demonstrated in simulations where the missile is a standard supersonic missile but the target is a maneuvering hypersonic missile with initial separation of just a few kilometers.

A deep hybrid fuzzy neural Hammerstein-Wiener model (FNHW), is proposed in this paper. The implication and inference of a neuro-fuzzy is based on the fuzzy rulebase that has been formed during traning. It requires the training data to be able to adequately represent entire system behaviors. However, the test data may vary with distribution shift in time series domain. Furthers, the training data may be derived from steady-state while the test data which is in the form of dynamically changing represented by drastic data shift under certain scenario such as financial crisis. A hybrid approach is proposed to employ neuro-fuzzy system to make prediction on steady-state data in parallel with the Hammerstein-Wiener model to predict the dynamically changing behavior. This is implemented by MLP as the control unit to decide the scale of contribution that each system is made to the final prediction. By doing this, the soundness of rulebase inference from neuro-fuzzy system on the steady-state data is achieved as well as inheriting the good approximation accuracy and excellent asymptotic tracking advantages of Hamerstein-Wiener model on the dynamically changing data. The effectiveness of proposed model is evaluated on two financial stock price prediction dataset. Experimental results showed that FNHW outperforms other neuro-fuzzy methods for both dataset.

An efficient control of a grid integrated microgrid system with wind energy generating system (WEGS) and battery storage, is proposed in this work. The quality of power injected to the utility is deteriorated with the level of unbalance and/or distortions present in the grid voltages. To compensate for the abnormalities, such as unbalance and distortions in grid voltages, positive sequence components (PSCs) of grid voltages are extracted from their filtered $\alpha\beta$ components. This work proposes a hybrid generalized integrator (HGI) filter for filtering the $\alpha\beta$ components of grid voltages and also provides its 90circ delayed signals, which are useful in generating PSCs. Phase and frequency of grid currents are decided by the unit active templates derived from PSCs of grid voltages, therefore, highquality grid currents are injected into the grid even for abnormal grid conditions. For faster dynamics during variations in load power, HGI filters are used for quick and accurate estimation of fundamental components of load currents. Intermittent power generation from WEGS is compensated for by integrating a battery bank into the system. Along with high-quality power injection, grid side converter (GSC) also maintains unity power factor at the point of common coupling (PCC). A 3. 7kW speed sensorless synchronous reluctance generator (SynRG) is used in WEGS, which is controlled by the machine side converter (MSC). The SynRG along with MSC control extracts maximum available power from the wind turbine (WT). Proposed control techniques are validated in Matlab/Simulink platform, and experimental validation is done on a developed laboratory prototype.

IoT wearable devices are widely expected to reduce the cost and risk of personal healthcare. However, ambulatory data collected from such devices are often corrupted or contaminated with severe noises. Signal Quality Indicators (SQIs) can be used to assess the quality of data obtained from wearable devices, such that transmission/ storage of unusable data can be prevented. This article introduces a novel and generalized SQI which can be implemented on an edge device for detecting the quality of any quasi-periodic signal under observation, regardless of the type of noise present. The application of this SQI on Electrocardiogram (ECG) signals is investigated. From the analysis carried out, it was found that the proposed generalized SQI is suitable for quality assessment of ECG signals and exhibits a linear behavior in the medium to high SNR regions under all noise conditions considered. The proposed SQI was used for acceptability testing of ECG records in CinC Physionet 2011 challenge dataset and found to be accurate for 90.4% of the records while having minimal computational complexity., Comment: Accepted at ISCAS 2020

This brief presents an ultra-low power single chip solution for electrocardiography (ECG) signal acquisition and processing in wearable ECG sensors. The chip contains a low noise preamplifier with embedded band-pass function, a programmable gain buffer, a 12-bit successive approximation ADC, a novel morphological filter based QRS detector, 8-Kb on-chip SRAM, a control unit and MCU interfaces. The chip was designed and implemented in 0.35- ${\mu }\text{m}$ standard CMOS process. The analog core operates from 0.8 V to 1.8 V, while the digital circuits and SRAM operate from 1.5 V to 3.6 V. The chip has a total core area of 5.74 mm2 and consumes $2.3~{\mu }\text{W}$ . Small size and low power consumption make this chip suitable for usage in wearable ECG sensors. Apart from presenting the measurement results, we also successfully demonstrate a prototype wearable ECG device, for long term cardiac monitoring using the proposed ECG-on-chip.

Our visual system is able to extract information on facial attractiveness from groups of faces that contain both coarse and detailed information. This raises the question: What information is extracted from a face group? Is the attractiveness perception of multiple faces driven by high or low spatial frequency that can highlight the local or global information of the faces, respectively? In the first experiment, we adapted participants to four unattractive faces with full bandwidth (FB), high spatial frequency (HSF), and low spatial frequency (LSF). We observed significant aftereffects in the HSF faces adaptation condition, which suggests that the perception of multiple unattractive faces is largely driven by HSF information. In the second experiment, we found a similar but different pattern in the direct-rating tasks, suggesting distinct perception mechanisms in unattractive versus attractive faces. In the third experiment, both the adaptation and direct-rating paradigms suggested that perception of multiple attractive faces is largely driven by LSF information. Overall, results from the three experiments together found that perception of multiple attractive and unattractive faces depends on visual information from different spatial frequencies, suggesting distinctive mechanisms in processing attractive and unattractive groups of faces. Ministry of Education (MOE) Nanyang Technological University This study was supported by the Social Science Foundationof Jiangsu Province (17JYC006), the City & Universitystrategy-Soochow University Leading Research Team inHumanities and Social Sciences (H. Y.), Nanyang Techno-logical University Undergraduate Research Experience onCAmpus (URECA) programme (W. C.), College of Arts,Humanities and Social Sciences Incentive Scheme (H. X.),Nanyang Technological University (H. X.), and SingaporeMinistry of Education Academic Research Fund (AcRF)Tier 1 (H. X.).

We introduce Kappa, an electronic tangible interface development kit designed to promote computational thinking for school students. Kappa enables students to create interactive games, animated stories and musical instruments using physical objects. We discuss the design considerations, the working mechanism and the need for such a toolkit in technology education. We also discuss results and observations from a pilot deployment of the kit in a school in urban India in a project-based learning context.