Search

Generative adversarial networks (GANs) have emerged as a powerful generative model in computer vision. Given their impressive abilities in generating highly realistic images, they are also being used in novel ways in applications in the life sciences. This raises an interesting question when GANs are used in scientific or biomedical studies. Consider the setting where we are restricted to only using the samples from a trained GAN for downstream group difference analysis (and do not have direct access to the real data). Will we obtain similar conclusions? In this work, we explore if “generated” data, i.e., sampled from such GANs can be used for performing statistical group difference tests in cases versus controls studies, common across many scientific disciplines. We provide a detailed analysis describing regimes where this may be feasible. We complement the technical results with an empirical study focused on the analysis of cortical thickness on brain mesh surfaces in an Alzheimer’s disease dataset. To exploit the geometric nature of the data, we use simple ideas from spectral graph theory to show how adjustments to existing GANs can yield improvements. We also give a generalization error bound by extending recent results on Neural Network Distance. To our knowledge, our work offers the first analysis assessing whether the Null distribution in “healthy versus diseased subjects” type statistical testing using data generated from the GANs coincides with the one obtained from the same analysis with real data. The code is available at https://github.com/yyxiongzju/GLapGAN.

The demand and supply gap of energy is large and the people are pressured to bear increasing hours of load losing, unnecessary intake of energy makes subjects even worse. So, the importance and requirement for energy conservation is increased exponentially. This paper outlines a step towards the conservation of electricity in standard and electricity especially with the aid of employing efficient constructing system with automation and installation of renewable based Micro Distributed Generator (MDG). The Smart Grid (SG) is a technological transformation from conventional electric grid, electro-mechanically controlled system, to smart, intelligent, and electronically controlled system. There are about 20–30% losses present in the conventional electric grid due to substandard operations at generation, transmission and distribution side. MDG play important role to build an efficient and autonomous micro-grid. In this paper importance of microgrid and some renewable energy sources utilized in formation of microgrid are discussed.

This article focuses on the motion planning and control of an automated differential-driven two-wheeled E-puck robot using Generalized Regression Neural Network (GRNN) architecture in the Virtual Robot Experimentation Platform (V-REP) software platform among scattered obstacles. The main advantage of this GRNN over the feedforward neural network is that it provides accurate results in a short period with minimal error. First, the designed GRNN architecture receives real-time obstacle information from the Infra-Red (IR) sensors of an E-puck robot. According to IR sensor data interpretation, this architecture sends the left and right wheel velocities command to the E-puck robot in the V-REP software platform. In the present study, the GRNN architecture includes the MIMO system, i.e., multiple inputs (IR sensors data) and multiple outputs (left and right wheel velocities). The three-dimensional (3D) motion and orientation results of the GRNN architecture-controlled E-puck robot are carried out in the V-REP software platform among scattered and wall-type obstacles. Further on, compared with the feedforward neural network, the proposed GRNN architecture obtains better navigation path length with minimum error results.

The fault monitoring and diagnosis of industrial machineries are very significant in Industry 4.0 revolution but are often complicated and labour intensive. The application of artificial intelligence (AI) techniques have now been an important part of condition monitoring of the mechanical and electrical machines because of its fast computation, higher accuracy, and robustness in performance, reducing the dependence on experienced personnel with expert knowledge. This paper presents a review of applications of AI-based fault diagnosis techniques that have had demonstrated success when applied to various industrial machineries. The important literature published in the last twenty years (i.e., 2000 to 2020) have been reviewed and added. In this work, first, a brief of various AI techniques such as artificial neural networks (ANN), deep learning (DL), fuzzy logic (FL), and support vector machine (SVM) are added. The literature on AI-based diagnostics used for various industrial machines, such as induction motor, bearing, gear, and centrifugal pump, are added and discussed in detail. The observation, research gap, and new ideas have been discussed, followed by a conclusion.

Traditional machine learning methods assume that training and testing data must be from the same machine running condition (MRC) and drawn from the same distribution. However, in several real-time industrial applications, this assumption does not hold. The traditional methods work satisfactorily in steady-state conditions but fail in time-varying conditions. In order to utilize time-varying data in variable MRCs, this article proposes a novel low-level knowledge transfer framework using a deep neural network (DNN) model for condition monitoring of machines in variable running conditions. The low-level features have been extracted in time, frequency, and time–frequency domains. These features are extracted from the source data to train the DNN. The trained DNN-based parameters are then transferred to another DNN, which is modified according to the low-level features extracted from the target data. The proposed approach is validated through three case studies on: 1) the air compressor acoustic data set; 2) the Case Western Reserve University bearing data set; and 3) the intelligent maintenance system bearing data set. The prediction accuracy obtained for the above case studies is as high as 100%, 93.07%, and 100%, respectively, with fivefold cross-validation. These real-time results show considerable improvement in the prediction performance using the proposed approach. Note to Practitioners —Condition-based monitoring schemes are widely applicable to rotating machines in various industries since they operate in tough working situations, and consequently, unpredicted failures occur. These unpredicted failures may cause perilous accidents in the industries. CBM systems prevent such failures, which results in the reduction of equipment damage and, hence, increases machinery lifetime. Modern industries are so complex and generating huge data, and these data can be collected using sensors, but placing a large number of sensors is difficult and expensive for different but similar kinds of faults in industries. This also increases the cost due to additional sensors and circuits. In this article, the authors have proposed a novel low-level knowledge transfer framework using the deep neural network (DNN)-based method for condition monitoring of machines in variable running conditions. Low-level features have been extracted to reduce the computations of DNN drastically with improved performance. This article also considered additional faults in the target domain, which is more practical in real-time applications. The proposed scheme has been validated with three case studies on acoustic and vibration signatures.

This paper proposes the right and left motor velocity based multi-objective genetic algorithm controlled navigation method for Two-Wheeled Pioneer P3-DX Robot (TWPR) in Virtual Robot Experimentation Platform (V-REP) software scenarios. The first objective function is made by taking front, left, and right ultrasonic sensors data and right motor velocity. Similarly, the second objective function is designed using front, left, and right ultrasonic sensors data and left motor velocity. Next, the sensor data are considered independent variables or inputs. The velocities of the motors are chosen as dependent variables or outputs for making multi-objective fitness functions for genetic algorithm (GA). This multi-objective GA makes a sensor-actuator control architecture and helps the TWPR to avoid the obstacles in the simulated scenarios. Further, the successful simulation test results show that the multi-objective GA provided a collision-free smooth, and shortest path for TWPR compared to the previously developed benchmark single objective GA.

In recent years, intelligent condition-based monitoring of rotary machinery systems has become a major research focus of machine fault diagnosis. In condition-based monitoring, it is challenging to form a large-scale well-annotated dataset due to the expense of data acquisition and costly annotation. The generated data have a large number of redundant features which degraded the performance of the machine learning models. To overcome this, we have utilized the advantages of minimum redundancy maximum relevance ( mRMR ) and transfer learning with a deep learning model. In this work, mRMR is combined with deep learning and deep transfer learning framework to improve the fault diagnostics performance in terms of accuracy and computational complexity. The mRMR reduces the redundant information from data and increases the deep learning performance, whereas transfer learning, reduces a large amount of data dependency for training the model. In the proposed work, two frameworks, i.e., mRMR with deep learning and mRMR with deep transfer learning, have explored and validated on CWRU and IMS rolling element bearings datasets. The analysis shows that the proposed frameworks can obtain better diagnostic accuracy compared to existing methods and can handle the data with a large number of features more quickly.

The efficiency of the low-cost renewable energy source i.e. solar is very poor due to inadequate extraction of maximum power. By employing the proper maximum power point tracking algorithm, the efficiency can be increased. An innovative adaptive backstepping neural network controller is proposed in this paper to extract the maximum power from the solar panels by tracking the desired photovoltaic array voltage in real-time. The maximum power is extracted indirectly by tuning the PV voltage to the desired PV voltage where the maximum power is attained at the desired PV voltage point. The desired photovoltaic array voltage is obtained from the linear regression method. The change in photovoltaic current caused by varying irradiance and temperature is approximated using the Chebyshev polynomials. The quicker steady-state and transient responses are accomplished and the computational burden of the photovoltaic system control law is reduced because of the orthogonal property of Chebyshev polynomials. The asymptotically stable system is obtained by tuning the weights of the neurons in accordance with the Lyapunov stability analysis. Also, Lyapunov control function of the backstepping control design procedure finds a control law by an innovative cubic equation interpretation, instead of resolving the first derivative of the control law, that diminishes the ripples in the duty cycle to make its appropriateness in real-time. A prototype is developed to validate the robustness of this controller in maximum power extraction at a faster time and the results confirm that adaptive backstepping neural network controller surpasses the performances of conventional backstepping controller and constant voltage PID controller.

Several data-driven methods such as signal processing and machine learning exist separately to analyze non-linear and non-stationary data but their performance degrades due to insufficient information in the real-time application. In order to improve the performance, this paper proposes a novel feature extraction method using fusion of hand-crafted (low-level) features and high-level features, followed by feature extraction/selection on fused features. Local energy-based hand-crafted features have been derived from empirical mode decomposition, and high-level features have been extracted from the deep neural network. A method is also proposed for reduction of massive data points in the samples. The proposed scheme has studied the effect of variation in the number of extracted/selected features. The effectiveness of the proposed scheme is validated through three case studies: a) on acoustic dataset collected from the reciprocating type air compressor, b) on vibration dataset collected from deep groove ball bearing, and c) on steel plate faults dataset. The classification accuracy on acoustic dataset are obtained as high as 100.0%, 99.78%, and 99.78% using the random forest, linear support vector machine, and radial basis function support vector machine, respectively, with 5-fold cross-validation. Similarly, on vibration dataset obtained accuracies are 100.0%. The proposed scheme has been compared with ten conventional methods on five-fold cross-validation. These experimental results show considerable improvement in the prediction performance of machine conditions using the proposed scheme.

This paper describes the navigation of an automated Pioneer P3-DX wheeled robot between obstacles using particle swarm optimization (PSO) algorithm tuned feedforward neural network (FNN). This PSO algorithm minimizes the mean square error between the actual and predicted values of the FNN. In this work, 2 separate DC motors and 16 ultrasonic sensors have been used for making differential drive steering angle and for collecting the distance from obstacles, respectively. The proposed without tuned FNN and PSO-tuned FNN receives obstacle's distance as inputs form ultrasonic sensors and control the steering angle of a differential drive of automated Pioneer P3-DX wheeled robot as output. We have compared the results between without tuned FNN and PSO-tuned FNN, and it has been found that PSO-tuned FNN gives a better trajectory and takes less distance to reach the target. Virtual Robot Experimentation Platform software has been used to design the real-time simulation results. A comparative study between without tuned FNN and PSO-tuned FNN verifies the effectiveness of PSO-tuned FNN for automated Pioneer P3-DX wheeled robot navigation. Also, we have compared this winner PSO-tuned FNN to the previously developed PSO-optimized Fuzzy Logic Controller navigational technique to show the authenticity and real-time implementation of PSO-tuned FNN.

This article investigates the performance of serial-parallel (SP) configuration of photovoltaic (PV) array under artificial arrangement for partial shading conditions (PSCs). The concept of bypass diode (BPD) is a very attractive solution to reduce the shading effect on the PV module and therefore gives better performance in such shading environment types. Various bypass diode integration (BPD) placement topologies with the solar photovoltaic (PV) modules are being investigated to demonstrate the improved performance under PSCs in the current work. The sixteen PV modules arranged in SP configuration are associated with the BPD Topologies such as (a) Without bypass diode (W–BPD) (b), Single string- single bypass diode (SS–SBPD) (c) Single string- double bypass diode (SS–DBPD) (d) Series group- bypass diode (G–BPD) (e) Staggered group- bypass diode (SG–BPD) (f) Multi level-octal bypass diode (ML–OBPD) for performance investigation. The performance assessment for all PV array supported by BPD topologies has been investigated using current–voltage (I–V) and power–voltage (P–V) characteristics and comparing to show better power and voltage results at global maximum power point (GMPP), improved fill factor and minimized power losses etc. The results presented may be recommended for the appropriate PV array configuration interconnection of BPD. Overall, this article reports that the ML–OBPD based SP configured PV array is superior among the BPD topologies under the considered PSTCs.

Control of batch polymerization has been a challenging task. In this work, we have tried to use Reinforcement Learning (RL), and Deep Reinforcement Learning (DRL) based control on addressing the existing challenges. RL is a class of machine learning wherein an agent directly interacts with the environment and learns from its experience. The RL consist of an agent who takes an action, and the action changes the state of the environment. Based on old and new state agent gets a reward, which is reinforcement for its future actions. In this work, we have implemented RL and DRL based control for batch polymerization of Polymethyl methacrylate (PMMA). In both the controllers, the input variable considered was jacket temperature, while the reactor temperature was the output variable. Both the controllers have been found to achieve the given setpoint, while the DRL controller being faster than the RL controllers. Further, RL and DRL control with risk sensitivity were also carried out to accommodate the process constraints.

Accumulation of anomalous tissue in the head formed the Brain Tumor. Brain tumor is basically caused by the uncommon growth of the anomalous cells in the brain. Since the skull encloses the very rigid brain. Any unspecified growth inside the restricted space where the cells multiply abruptly and ceaselessly can create the major problem. After the 20th century, the number of the patient who have the brain tumor is increasing day by day for the aging population is a worldwide health problem. Basically Malformed mass called tumor can be identified into two types cancerous (malignant) or noncancerous (benign). The main objective of this research is to create the technique that reliably identifies the existence of a tumor that includes all cancerous tissues. Fish schooling Genetic algorithm is used to recognize a presence of cancerous part called tumor area from a soft tissue called brain MRI images and to predict that detected area whether it is a tumor or not. The proposed module consists of the steps such as Brain MRI dataset Pre-processing of MRI image, remove noise through image Enhancement, stripping the skull part, fish school search algorithm based indication are used in the research. According to the results acquire from the performance of Fish School Search Algorithm in detection of tumor achieves the better performance in terms of comparing the quality parameters such as accuracy, efficiency with the previous method used.

The emergence of deep learning has impacted numerous machine learning based applications and research. The reason for its success lies in two main advantages: 1) it provides the ability to learn very complex non-linear relationships between features and 2) it allows one to leverage information from unlabeled data that does not belong to the problem being handled. This paper presents a transfer learning procedure for cancer classification, which uses feature selection and normalization techniques in conjunction with s sparse auto-encoders on gene expression data. While classifying any two tumor types, data of other tumor types were used in unsupervised manner to improve the feature representation. The performance of our algorithm was tested on 36 two-class benchmark datasets from the GEMLeR repository. On performing statistical tests, it is clearly ascertained that our algorithm statistically outperforms several generally used cancer classification approaches. The deep learning based molecular disease classification can be used to guide decisions made on the diagnosis and treatment of diseases, and therefore may have important applications in precision medicine.

A number of results have recently demonstrated the benefits of incorporating various constraints when training deep architectures in vision and machine learning. The advantages range from guarantees for statistical generalization to better accuracy to compression. But support for general constraints within widely used libraries remains scarce and their broader deployment within many applications that can benefit from them remains under-explored. Part of the reason is that Stochastic gradient descent (SGD), the workhorse for training deep neural networks, does not natively deal with constraints with global scope very well. In this paper, we revisit a classical first order scheme from numerical optimization, Conditional Gradients (CG), that has, thus far had limited applicability in training deep models. We show via rigorous analysis how various constraints can be naturally handled by modifications of this algorithm. We provide convergence guarantees and show a suite of immediate benefits that are possible — from training ResNets with fewer layers but better accuracy simply by substituting in our version of CG to faster training of GANs with 50% fewer epochs in image inpainting applications to provably better generalization guarantees using efficiently implementable forms of recently proposed regularizers.

An increase in the share of distributed generation (DG) in the global generation system is a direct indication of the development of available technologies. The extraction of natural energy resources and their use as DG has several advantages, such as the reduction in line losses, improved voltage profile and reliability, etc., but the incorrect installation of these power plants can also have some negative effects. The innovation in technology has motivated to extract the maximum benefit of natural energy resources. Due to this, the capacity and location of these energy resources should be carefully identified. The optimal placement of a distributed generation power plant, in the existing network, is analyzed in this article. The proposed methodology is inspired by the human immune system. In this methodology clonal selection principle of immune system is combined with particle swarm optimization. For checking the validity of the proposed method two test systems, IEEE 33-node radial distribution system and IEEE 14-node loop distribution system, are considered. Results show the validity of the proposed algorithm in radial as well as in loop distribution system.

A comparison of various analysis algorithms based for recommendation systems used in the market and businesses has their usage without considering the fact that if used with correct algorithm can increase its efficiency. Comparison is obtained in this paper after studying and researching various obtained algorithms present currently. This paper presents an introduction to the concept of recommendation systems which are recently working on different domain and then comparison is made, that too employed for scalability, reliability, faster process and efficiency.

Recent results in coupled or temporal graphical models offer schemes for estimating the relationship structure between features when the data come from related (but distinct) longitudinal sources. A novel application of these ideas is for analyzing group-level differences, i.e., in identifying if trends of estimated objects (e.g., covariance or precision matrices) are different across disparate conditions (e.g., gender or disease). Often, poor effect sizes make detecting the differential signal over the full set of features difficult: for example, dependencies between only a subset of features may manifest differently across groups. In this work, we first give a parametric model for estimating trends in the space of SPD matrices as a function of one or more covariates. We then generalize scan statistics to graph structures, to search over distinct subsets of features (graph partitions) whose temporal dependency structure may show statistically significant group-wise differences. We theoretically analyze the Family Wise Error Rate (FWER) and bounds on Type 1 and Type 2 error. Evaluating on US census data, we identify groups of states with cultural and legal overlap related to baby name trends and drug usage. On a cohort of individuals with risk factors for Alzheimer’s disease (but otherwise cognitively healthy), we find scientifically interesting group differences where the default analysis, i.e., models estimated on the full graph, do not survive reasonable significance thresholds.

We introduce a unified framework to jointly model images, text, and human attention traces. Our work is built on top of the recent Localized Narratives annotation frame-work [31], where each word of a given caption is paired with a mouse trace segment. We propose two novel tasks: (1) predict a trace given an image and caption (i.e., visual grounding), and (2) predict a caption and a trace given only an image. Learning the grounding of each word is challenging, due to noise in the human-provided traces and the presence of words that cannot be meaningfully visually grounded. We present a novel model architecture that is jointly trained on dual tasks (controlled trace generation and controlled caption generation). To evaluate the quality of the generated traces, we propose a local bipartite matching (LBM) distance metric which allows the comparison of two traces of different lengths. Extensive experiments show our model is robust to the imperfect training data and outperforms the baselines by a clear margin. More-over, we demonstrate that our model pre-trained on the pro-posed tasks can be also beneficial to the downstream task of COCO’s guided image captioning. Our code1 and project page2 are publicly available.

The unprecedented penetration of distributed energy resources (DERs) such as wind power generations (WPGs) poses tremendous challenges for for the planning and maintenance of power systems due to their intermittent and uncertain nature. This paper mainly focuses on comparing basic probabilistic load flow (PLF) techniques when WPGs are integrated into the existing power grid. Considering loads and WPGs as random inputs, the performance of the cumulant method (CM) and point estimation method (PEM) are analyzed with respect to Monte-Carlo method for higher precision and less computational time. Case-studies are carried out on sample 10-bus and SR 72-bus equivalent systems. Simulation results demonstrated that $2n+1$ PEM provides the best performance when dealing with high level of uncertainty associated with input variables.

This paper introduces a novel toolbox named BIDEAL for the generation of biclusters, their analysis, visualization, and validation. The objective is to facilitate researchers to use forefront biclustering algorithms embedded on a single platform. A single toolbox comprising various biclustering algorithms play a vital role to extract meaningful patterns from the data for detecting diseases, biomarkers, gene-drug association, etc. BIDEAL consists of seventeen biclustering algorithms, three biclusters visualization techniques, and six validation indices. The toolbox can analyze several types of data, including biological data through a graphical user interface. It also facilitates data preprocessing techniques i.e., binarization, discretization, normalization, elimination of null and missing values. The effectiveness of the developed toolbox has been presented through testing and validations on Saccharomyces cerevisiae cell cycle, leukemia cancer, mammary tissue profile, and ligand screen in B-cells datasets. The biclusters of these datasets have been generated using BIDEAL and evaluated in terms of coherency, differential co-expression ranking, and similarity measure. The visualization of generated biclusters has also been provided through a heat map and gene plot.

This paper studies an n-player non-cooperative game where each player has expected-value payoff function and chance-constrained strategy set. We consider the case where the row vectors defining the constraints are independent random vectors whose probability distributions are not completely known and belong to a certain distributional uncertainty set. The chance-constrained strategy sets are defined using a distributionally robust framework. We consider one density based uncertainty set and four two-moments based uncertainty sets. One of the considered uncertainty sets is based on a nonnegative support. Under the standard assumptions on the players’ payoff functions, we show that there exists a Nash equilibrium of a distributionally robust chance-constrained game for each uncertainty set. As an application, we study Cournot competition in electricity market and perform the numerical experiments for the case of two electricity firms.

Changing trends towards wind power generation requires new and advanced models for control implementation for improved power quality, better reliability and increased efficiency. This paper emphasizes the performance comparison of three grid connected configurations of wind energy conversion systems using squirrel cage induction generator. Control designs using different types of machine side converters such as a diode rectifier, a diode rectifier with boost converter and a voltage source converter have been modeled, simulated and analyzed. A complete comparison of these configurations is presented in terms of the real and reactive power control and the power quality. Maximum power point tracking has been implemented in the latter two control designs so that optimum power is harnessed for given wind velocities. The simulation results exemplify the outstanding performance of the system with the use of back to back connected voltage source converters.

Motion planning problems of a wheeled robot remain challenging for the researchers and are being solved by applying various nature-inspired algorithms. Therefore, this article implements the sunflower optimization algorithm and tries to solve the motion control and planning of a wheeled robot. A sunflower optimization algorithm is a newly invented algorithm, and it is inspired by the motion of sunflowers to capture solar radiation. In this study, the differential driven two-wheeled Pioneer P3-DX (P3D) robot has been chosen for experiments. The real-time ultrasonic sensors data (obstacles distance) information have been taken as inputs of the algorithm, and the steering control angle of the P3D robot is an output of the algorithm. The objective function of the algorithm has been made by taking these inputs and output data. Further, the implemented algorithm has been tested in the Virtual Robot Experimentation Platform (V-REP) software scenarios on differential driven two-wheeled P3D robot. Next, this implemented algorithm has been compared with previously developed particle swarm optimization algorithm to verify the effectiveness and authenticity of the implemented algorithm.

One strategy for adversarially training a robust model is to maximize its certified radius -- the neighborhood around a given training sample for which the model's prediction remains unchanged. The scheme typically involves analyzing a "smoothed" classifier where one estimates the prediction corresponding to Gaussian samples in the neighborhood of each sample in the mini-batch, accomplished in practice by Monte Carlo sampling. In this paper, we investigate the hypothesis that this sampling bottleneck can potentially be mitigated by identifying ways to directly propagate the covariance matrix of the smoothed distribution through the network. To this end, we find that other than certain adjustments to the network, propagating the covariances must also be accompanied by additional accounting that keeps track of how the distributional moments transform and interact at each stage in the network. We show how satisfying these criteria yields an algorithm for maximizing the certified radius on datasets including Cifar-10, ImageNet, and Places365 while offering runtime savings on networks with moderate depth, with a small compromise in overall accuracy. We describe the details of the key modifications that enable practical use. Via various experiments, we evaluate when our simplifications are sensible, and what the key benefits and limitations are., Comment: This paper has been accepted by CVPR 2021 as an oral presentation. An introduction video can be found: https://youtu.be/m1ya2oNf5iE

A major portion of global electricity consumption is done by residential sector. The nature of load variation is highly uncertain and has become a challenge to control thereby contributes a lot in peak load burden on utility. However, the concept of demand side management (DSM) enables the utility to manage the residential sector load in smart grid environment. The concept of DSM in the smart grid enables the ability to facilitate the flow of information between customers and utility that helps to rebuild the curve. In this paper, human behavior-based algorithm is developed to alter the load curve by managing residential sector load. The proposed algorithm is tested with Python coding and offers satisfactory results. The considerable reduction in peak load with increased utility earning is noticed with critical peak pricing scheme (CPP).

Generation Scheduling commonly known as Unit Commitment Problem is one of the most typical optimization problems in electrical engineering. In this paper, Particle Swarm Optimization Technique with few modifications is implemented as an Artificial Intelligence-based technique for the solution of this convoluted nonlinear optimization problem. The proposed technique is implemented on a standard IEEE ten generator system. The entire work is performed in MATLAB environment. The results have been compared and found better than the other available techniques available in literature.

The shortage of N95 masks have spurred efforts on developing safe and scientifically-validated decontamination and reuse protocols that are easily scalable and universally applicable even in low-resource settings. We report on the development and implementation of a heat-based N95 mask decontamination system in a large hospital setting (Tata Memorial Hospital, Mumbai, India) with over 8000 N95 masks from about 1400 individual users decontaminated and in reuse till date. We describe the challenges and constraints in choosing a proven, scalable, and easy-to-implement decontamination solution. We discuss the heat treatment and particle filtration efficiency measurement experiments done to validate a decontamination treatment protocol at a target temperature of 70oC for a duration of 60 minutes, and the scaling up of this method using a standard hot drying cabinet at the hospital. The logistics of ensuring optimal utilization of the decontamination facility without compromising on basic safety principles are detailed. Our method relies on equipment available in standard hospitals, is simple to set-up, scalable, and can be easily replicated in low-resource settings. We further believe such limited reuse strategies, even in times of abundant N95 mask availability, would not only be cost-saving but also be environmentally responsible in reducing the amount of medical waste.

Computational tools in the form of software packages are burgeoning in the field of medical imaging and biomedical research. These tools enable biomedical researchers to analyze a variety of data using modern machine learning and statistical analysis techniques. While these publicly available software packages are a great step towards a multiplicative increase in the biomedical research productivity, there are still many open issues related to validation and reproducibility of the results. A key gap is that while scientists can validate domain insights that are implicit in the analysis, the analysis itself is coded in a programming language and that domain scientist may not be a programmer. Thus, there is no/limited direct validation of the program that carries out the desired analysis. We propose a novel solution, building upon recent successes in natural language understanding, to address this problem. Our platform allows researchers to perform, share, reproduce and interpret the analysis pipelines and results via natural language. While this approach still requires users to have a conceptual understanding of the techniques, it removes the burden of programming syntax and thus lowers the barriers to advanced and reproducible neuroimaging and biomedical research.

A sizable body of work on relative attributes provides evidence that relating pairs of images along a continuum of strength pertaining to a visual attribute yields improvements in a variety of vision tasks. In this paper, we show how emerging ideas in graph neural networks can yield a solution to various problems that broadly fall under relative attribute learning. Our main idea is the observation that relative attribute learning naturally benefits from exploiting the graph of dependencies among the different relative attributes of images, especially when only partial ordering is provided at training time. We use message passing to perform end to end learning of the image representations, their relationships as well as the interplay between different attributes. Our experiments show that this simple framework is effective in achieving competitive accuracy with specialized methods for both relative attribute learning and binary attribute prediction, while relaxing the requirements on the training data and/or the number of parameters, or both.

Increase in the emission of greenhouse gases, global warming, government policies, and raise in the awareness about the benefits of the renewable energy resources are the key factors for increase in the penetration of renewable energy resources in the existing system. This integration is mainly done in the form of Distributed Generation (DG). The Integration of DG in distribution system converts passive network into an active network. For extracting maximum benefits of integration of DG, optimal size of DG should be integrated at optimum location. In this paper, performance analysis of a distribution system is studied after the optimal placement of DG. For optimal placement, Particle Swarm Optimization (PSO) is used. Performance analysis is done for multi-DG placement. For detailed study, IEEE 33-bus radial distribution system is considered as test system.

Algorithmic decision making based on computer vision and machine learning methods continues to permeate our lives. But issues related to biases of these models and the extent to which they treat certain segments of the population unfairly, have led to legitimate concerns. There is agreement that because of biases in the datasets we present to the models, a fairness-oblivious training will lead to unfair models. An interesting topic is the study of mechanisms via which the de novo design or training of the model can be informed by fairness measures. Here, we study strategies to impose fairness concurrently while training the model. While many fairness based approaches in vision rely on training adversarial modules together with the primary classification/regression task, in an effort to remove the influence of the protected attribute or variable, we show how ideas based on well-known optimization concepts can provide a simpler alternative. In our proposal, imposing fairness just requires specifying the protected attribute and utilizing our routine. We provide a detailed technical analysis and present experiments demonstrating that various fairness measures can be reliably imposed on a number of training tasks in vision in a manner that is interpretable.

Inverted bottleneck layers, which are built upon depthwise convolutions, have been the predominant building blocks in state-of-the-art object detection models on mobile devices. In this work, we investigate the optimality of this design pattern over a broad range of mobile accelerators by revisiting the usefulness of regular convolutions. We discover that regular convolutions are a potent component to boost the latency-accuracy trade-off for object detection on accelerators, provided that they are placed strategically in the network via neural architecture search. By incorporating regular convolutions in the search space and directly optimizing the network architectures for object detection, we obtain a family of object detection models, MobileDets, that achieve state-of-the-art results across mobile accelerators. On the COCO object detection task, MobileDets outperform MobileNetV3+SSDLite by 1.7 mAP at comparable mobile CPU inference latencies. MobileDets also outperform MobileNetV2+SSDLite by 1.9 mAP on mobile CPUs, 3.7 mAP on Google EdgeTPU, 3.4 mAP on Qualcomm Hexagon DSP and 2.7 mAP on Nvidia Jetson GPU without increasing latency. Moreover, MobileDets are comparable with the state-of-the-art MnasFPN on mobile CPUs even without using the feature pyramid, and achieve better mAP scores on both EdgeTPUs and DSPs with up to 2x speedup. Code and models are available in the TensorFlow Object Detection API: https://github.com/tensorflow/models/tree/master/research/object_detection., Comment: Accepted at CVPR 2021; Code and models are available in the TensorFlow Object Detection API: https://github.com/tensorflow/models/tree/master/research/object_detection

In today’s world, automation has become an inseparable part of our lives. Simultaneously, food is the basic unit on which humans survive and when technology and food are thought together, it becomes important to heed to the hygiene, quality and cost. As of today’s scenario, 34% food available in the globe is packaged which makes it crucial to look forward towards the shortcomings and the advancements in the production, packaging, inventory, supply chain, package handling, quality management and other related aspects. In the conventional packaging system, power failure and human error affect the efficiency of the system, miss-match of the processes and waste of time which is non-negotiable for any FMCG. This paper encompasses engineering solution to the mentioned challenges. It includes barcode sensor assisted with a load cell, a decision-making device—Arduino Mega 2560, programmable logic controller(s) (PLC) for the identification and sorting of CLD(s) or RSC(s) along a retracting conveyor. For the packaging solution, a revolving disc type lid closing system and a compact taping mechanism have been introduced which can reduce dependency on manpower for CLD packing. GSM module in the feedback to the microcontroller is placed, such that any deviation from the default parameters will be sensed automatically, and area of problem occurrence (AOC) is easily identified. In addition to above advents, this paper also talks about the SCADA implemented HMI(s) in order to meet the industry standards of monitoring and control along with the overall condition monitoring and data acquisition for research and analysis.

Integration of renewable generations with electrical power systems has gained considerable attention in recent years due to environmental and economic benefits. However, this integration introduces additional uncertainties into the existing system and requires appropriate uncertainty modeling for power systems. Typically the uncertainties in power systems are modeled using probabilistic or possibilistic approaches. A combined probabilistic-possibilistic approach is necessary when some uncertain variables are probabilistic and others are possibilistic. This paper presents a complete review of uncertainty categorization and several techniques to address the uncertainty in power systems, along with the merits and weaknesses of each technique. The challenges have been highlighted for future research directions. Analytical and approximate methods are reviewed in this paper when wind power generations are integrated into the existing power grid. Considering the uncertainties of wind power generation and system load demands, the basic probabilistic methods such as Monte-Carlo simulation, cumulant, and 2 n +1 point estimation methods are implemented. To explore the capability and shortcoming of these basic methods, a 72-bus equivalent system of Indian southern region power grid is taken into consideration. The results obtained using Monte-Carlo simulation method are treated as a benchmark to analyze the performance of the cumulant and 2 n +1 point estimation methods.

This paper proposes a novel adaptive Type-2 fuzzy filter for removing salt and pepper noise from the images. The filter removes noise in two steps. In the first step, the pixels are categorized as good or bad based on their primary membership function (MF) values in the respective filter window. In this paper, two approaches have been proposed for finding threshold between good or bad pixels by designing primary MFs. a) MFs with distinct Means and same Variance and b) MFs with distinct Means and distinct Variances. The primary MFs of the Type-2 fuzzy set is chosen as Gaussian membership functions. Whereas, in the second step, the pixels categorized as bad are denoised. For denoising, a novel Type-1 fuzzy approach based on a weighted mean of good pixels is presented in the paper. The proposed filter is validated for several standard images with the noise level as low as 20% to as high as 99%. The results show that the proposed filter performs better in terms of peak signal-noise-ratio values compared to other state-of-the-art algorithms.

Among all the motors Induction motor (IM) plays a vital role in industry and the demand for their reliability and safe operation is increasing day by day. They are reliable but they do wear out if not maintained timely which in turn will lead to excessive loss of revenue and also man power and this motivates us to develop an intelligent methodology for diagnostic of incipient faults in Induction Motor. This paper focuses on the development of optimum Deep Learning based diagnostic technique to detect the mechanical and electrical faults in Induction Motor. Here, mechanical and electrical faults of different severity level are being tested on machine fault simulator using acquired current and vibrational signals. In this work, optimum model of Deep Neural Network (DNN) based on critical vibration and current features is developed and finally used to timely and effectively detect that which kind of faults the given IM is dealing with. The results are added and discussed in the result and discussions.

In various applications of defence and research such as demagnetization of naval vehicles and satellites, removal of residual magnetic field plays an essential role. For such use cases, conventional way of demagnetization is the use of Helmholtz coils and Merritt coils. These coil systems produce uniform magnetic field, but the volume of the uniform region is relatively smaller. Therefore, for the applications which require patch of uniform magnetic field for longer longitudinal length of the region such as deperming process of Submarine, Aircraft Carrier; these coil systems become infeasible to implement practically due to their enormous size. Another impediment in implementation of above-mentioned coil system is that when size of coil system becomes larger, the requirement current also increases for the same degree of results. This increasing amount of current poses issues of more complexities in electrical control circuits, heat dissipation and cost. Another disadvantage with such system is that a slight shift from the optimized position causes a larger change in uniformity, and thus, such systems are not robust with shift in location. This paper proposes a novel method for the design of a higher-order equidistant coil system which overcomes these flaws and presents a technique to compute the ampere-turns requirement of each coil for generation of higher-degree homogeneous magnetic field. The system so designed has more feasible design parameters as compared to the conventional systems for high uniformity applications. The proposed system is sturdier and fault tolerant against any deviation or error from the ideal design parameters. The theoretical results are matched against the finite element simulation software Opera which are in close agreement.

Summary For large-scale testing with graph-associated data, we present an empirical Bayes mixture technique to score local false-discovery rates (FDRs). Compared to procedures that ignore the graph, the proposed Graph-based Mixture Model (GraphMM) method gains power in settings where non-null cases form connected subgraphs, and it does so by regularizing parameter contrasts between testing units. Simulations show that GraphMM controls the FDR in a variety of settings, though it may lose control with excessive regularization. On magnetic resonance imaging data from a study of brain changes associated with the onset of Alzheimer’s disease, GraphMM produces greater yield than conventional large-scale testing procedures.

Hybrid generation systems are very essential to meet the demand of power and rural electrification. Autonomous hybrid energy system integrates the different sources of power such as solar, wind, fuel cell etc. This paper develops a model of autonomous energy system to control frequency. The aim of the paper is to design a Proportional-Integral (PI) based autonomous hybrid energy system for minimizing the frequency deviation in the wind, diesel, aqua electrolyzer, fuel cell, battery energy storage system and compare its performance and robustness with firefly algorithm. Firefly algorithm is used to optimize the gain of PI controller for reducing the deviation in frequency. For checking the effectiveness of the developed control schemes for enhancement of the hybrid power system small scale stability, variation in generation and load both are considered. Time domain simulations are performed under generation and load varying conditions. In past classic PI controllers are tuned manually which is complicated. Results show that soft computing techniques Firefly algorithm methods have proved their excellence to minimize frequency deviation.

Modern deep networks have proven to be very effective for analyzing real world images. However, their application in medical imaging is still in its early stages, primarily due to the large size of three-dimensional images, requiring enormous convolutional or fully connected layers - if we treat an image (and not image patches) as a sample. These issues only compound when the focus moves towards longitudinal analysis of 3D image volumes through recurrent structures, and when a point estimate of model parameters is insufficient in scientific applications where a reliability measure is necessary. Using insights from differential geometry, we adapt the tensor train decomposition to construct networks with significantly fewer parameters, allowing us to train powerful recurrent networks on whole brain image volume sequences. We describe the "orthogonal" tensor train, and demonstrate its ability to express a standard network layer both theoretically and empirically. We show its ability to effectively reconstruct whole brain volumes with faster convergence and stronger confidence intervals compared to the standard tensor train decomposition. We provide code and show experiments on the ADNI dataset using image sequences to regress on a cognition related outcome.

The worlds increasing demand of electricity and lessening of fossil fuels has created a condition for serious concern to search for new inexhaustible resources to fulfill the future electrical energy demand. In this study a techno-economic comparison of a grid connected residential load is made with the grid connected residential load along with solar PV panels based on Net-Metering system. The design implementation and analysis is accomplished by using HOMER Software. This paper also considers the feasibility analysis of solar radiations at the described location. It is observed here that the system under consideration gives minimum expenditure when connected with solar PV panels. Simulations result shows that the total net present cost along with cost of energy and annual operational cost is reduced significantly when compared with the existing grid system. In this system the grid has same rates for selling and buying of electrical energy and a payback period analysis is also carried out with the proposed system.

There is a growing interest in strategies that can help us understand or interpret neural networks -- that is, not merely provide a prediction, but also offer additional context explaining why and how. While many current methods offer tools to perform this analysis for a given (trained) network post-hoc, recent results (especially on capsule networks) suggest that when classes map to a few high level ``concepts'' in the preceding layers of the network, the behavior of the network is easier to interpret or explain. Such training may be accomplished via dynamic/EM routing where the network ``routes'' for individual classes (or subsets of images) are dynamic and involve few nodes even if the full network may not be sparse. In this paper, we show how a simple modification of the SGD scheme can help provide dynamic/EM routing type behavior in convolutional neural networks. Through extensive experiments, we evaluate the effect of this idea for interpretability where we obtain promising results, while also showing that no compromise in attainable accuracy is involved. Further, we show that the minor modification is seemingly ad-hoc, the new algorithm can be analyzed by an approximate method which provably matches known rates for SGD.

The analysis of medical images and to find meaningful patterns in it is a cumbersome task, even with the use of techniques of Computer Vision when the dataset is very large. In such a situation deep learning is a handy tool, because of its ability to learn and extract meaningful patterns and features from the images. The use of multiple modalities of training data to train system has been in practice for conventional machine learning algorithms. Here, in this paper, we are going to present a Deep Learning based architecture for extraction of features from large training set of medical images. The deep learning model is tested against conventional techniques by performing Multimodal fusion, Shared Learning and Cross Learning on it. It was found out that Deep Learning model performs superior than the conventional techniques in multimodal fusion and shared learning settings.

There is much interest in computer vision to utilize commodity hardware for gaze estimation. A number of papers have shown that algorithms based on deep convolutional architectures are approaching accuracies where streaming data from mass-market devices can offer good gaze tracking performance, although a gap still remains between what is possible and the performance users will expect in real deployments. We observe that one obvious avenue for improvement relates to a gap between some basic technical assumptions behind most existing approaches and the statistical properties of the data used for training. Specifically, most training datasets involve tens of users with a few hundreds (or more) repeated acquisitions per user. The non i.i.d. nature of this data suggests better estimation may be possible if the model explicitly made use of such “repeated measurements” from each user as is commonly done in classical statistical analysis using so-called mixed effects models. The goal of this paper is to adapt these “mixed effects” ideas from statistics within a deep neural network architecture for gaze estimation, based on eye images. Such a formulation seeks to specifically utilize information regarding the hierarchical structure of the training data — each node in the hierarchy is a user who provides tens or hundreds of repeated samples. This modification yields an architecture that offers state of the art performance on various publicly available datasets improving results by 10-20%.

This paper introduces a novel pooling method namely fuzzy based pooling for image classification. Herein, a pooling method for bolstering the performance of conventional convolutional neural network (CNN) has been proposed. Conventional architecture of CNN uses pooling operation for dimension reduction which sometimes results in the loss of information. In this paper, a novel pooling method using fuzzy logic is introduced for dimension reduction. The proposed pooling method reduces the spatial size of convolved features in two steps. In the first step, the convolved features within a window to be pooled are processed using type-2 fuzzy logic for identifying the dominant features. Then, type-1 fuzzy logic with a weighted average of the dominant features within a window is used to reduce the spatial size. The proposed method is bench marked against conventional pooling techniques for MNIST dataset of handwritten digits recognition and CIFAR-10 dataset of RGB images. The accuracy shows that the proposed fuzzy based pooling performs better than the standard pooling techniques such as max and average pooling which helps to improve the performance of CNN.