Showing papers by "Andreas Spanias published in 2019"

Positive And Unlabeled Learning Algorithms And Applications: A Survey

Abstract: This paper will address the Positive and Unlabeled learning problem (PU learning) and its importance in the growing field of semi-supervised learning. In most real-world classification applications, well labeled data is expensive or impossible to obtain. We can often label a small subset of data as belonging to the class of interest. It is frequently impractical to manually label all data we are not interested in. We are left with a small set of positive labeled items of interest and a large set of unknown and unlabeled data. Learning a model for this is the PU learning problem.In this paper, we explore several applications for PU learning including examples in biological/medical, business, security, and signal processing. We then survey the literature for new and existing solutions to the PU learning problem.

Solar Array Fault Detection using Neural Networks

Abstract: In this paper, we describe a Cyber-Physical system approach to fault detection in Photovoltaic (PV) arrays. More specifically, we explore customized neural network algorithms for fault detection from monitoring devices that sense data and actuate at each individual panel. We develop a framework for the use of feedforward neural networks for fault detection and identification. Our approach promises to improve efficiency by detecting and identifying eight different faults and commonly occurring conditions that affect power output in utility scale PV arrays.

Connection Topology Optimization in Photovoltaic Arrays using Neural Networks

Abstract: A cyber-physical system (CPS) approach for optimizing the output power of photovoltaic (PV) energy systems is proposed. In particular, a novel connection topology reconfiguration strategy for PV arrays to maximize power output under partial shading conditions using neural networks is put forth. Depending upon an irradiance/shading profile of the panels, topologies, namely series parallel (SP), total cross tied (TCT) or bridge link (BL) produce different maximum power points (MPP). The connection topology of the PV array that provides the maximum power output is chosen using a multi-layer perceptron. The simulation results show that empirically an output power increase of 12% can be achieved through reconfiguration. The method proposed can be implemented in any CPS PV system with switching capabilities and is simple to implement.

Analysis and Design of Robust Max Consensus for Wireless Sensor Networks

Abstract: A novel distributed algorithm for estimating the maximum of the node initial state values in a network, in the presence of additive communication noise is proposed. Conventionally, the maximum is estimated locally at each node by updating the node state value with the largest received measurements in every iteration. However, due to the additive channel noise, the estimate of the maximum at each node drifts at each iteration and this results in nodes diverging from the true max value. Max-plus algebra is used as a tool to study this ergodic process. The subadditive ergodic theorem is invoked to establish a constant growth rate for the state values due to noise, which is studied by analyzing the max-plus Lyapunov exponent of the product of noise matrices in a max-plus semiring. The growth rate of the state values is upper bounded by a constant which depends on the spectral radius of the network and the noise variance. Upper and lower bounds are derived for both fixed and random graphs. Finally, a two-run algorithm robust to additive noise in the network is proposed and its variance is analyzed using concentration inequalities. Simulation results supporting the theory are also presented.

PV Array Fault Detection using Radial Basis Networks

Abstract: An increase in grid-connected photovoltaic arrays creates a need for efficient and reliable fault detection. In this paper, machine learning strategies for fault detection are presented. An Artificial Neural Network was studied with the goal of detecting three photovoltaic module conditions. In addition, an unsupervised approach was successfully implemented using the -means clustering algorithm, successfully detecting arc and ground faults. To distinguish and localize additional faults such as shading and soiling, a supervised approach is adopted using a Radial Basis Function Network. A solar array dataset with voltage, current, temperature, and irradiance was examined. This dataset had labeled data with normal conditions and faults due to soiling and shading. A radial basis network was trained to classify faults, resulting in an error rate below 2% on synthetic data with realistic levels of noise.

Designing an Effective Metric Learning Pipeline for Speaker Diarization

Abstract: State-of-the-art speaker diarization systems utilize knowledge from external data, in the form of a pre-trained distance metric, to effectively determine relative speaker identities to unseen data. However, much of recent focus has been on choosing the appropriate feature extractor, ranging from pre-trained i–vectors to representations learned via different sequence modeling architectures (e.g. 1D-CNNs, LSTMs, attention models), while adopting off-the-shelf metric learning solutions. In this paper, we argue that, regardless of the feature extractor, it is crucial to carefully design a metric learning pipeline, namely the loss function, the sampling strategy and the discriminative margin parameter, for building robust diarization systems. Furthermore, we propose to adopt a fine-grained validation process to obtain a comprehensive evaluation of the generalization power of metric learning pipelines. To this end, we measure diarization performance across different language speakers, and variations in the number of speakers in a recording. Using empirical studies, we provide interesting insights into the effectiveness of different design choices and make recommendations.

Analysis and Design of Robust Max Consensus for Wireless Sensor Networks

Abstract: A novel distributed algorithm for estimating the maximum of the node initial state values in a network, in the presence of additive communication noise is proposed. Conventionally, the maximum is estimated locally at each node by updating the node state value with the largest received measurements in every iteration. However, due to the additive channel noise, the estimate of the maximum at each node drifts at each iteration and this results in nodes diverging from the true max value. Max-plus algebra is used as a tool to study this ergodic process. The subadditive ergodic theorem is invoked to establish a constant growth rate for the state values due to noise, which is studied by analyzing the max-plus Lyapunov exponent of the product of noise matrices in a max-plus semiring. The growth rate of the state values is upper bounded by a constant which depends on the spectral radius of the network and the noise variance. Upper and lower bounds are derived for both fixed and random graphs. Finally, a two-run algorithm robust to additive noise in the network is proposed and its variance is analyzed using concentration inequalities. Simulation results supporting the theory are also presented.

Distributed Bayesian Estimation with Low-rank Data: Application to Solar Array Processing

Abstract: In this paper, we present a distributed array processing algorithm to analyze the power output of solar photo-voltaic (PV) installations, leveraging the low-rank structure inherent in the data to estimate possible faults. Our multi-agent algorithm requires near-neighbor communications only and is also capable of jointly estimating the common low rank cloud profile and local shading of panels. To illustrate the workings of our algorithm, we perform experiments to detect shading faults in solar PV installations within a single ZIP code. Additionally, we also derive a Bayesian lower bound on the shading parameter’s mean squared estimation error. The results are promising and show that we can successfully estimate the fraction of partial shading in solar installations that can usually go unnoticed.

A Linear Systems Perspective on Intrusion Detection for Routing in Reconfigurable Wireless Networks

Abstract: Reconfigurable wireless networks, such as ad hoc or wireless sensor networks, do not rely on fixed infrastructure. Nodes must cooperate in the multi-hop routing process. This dynamic and open nature make reconfigurable networks vulnerable to routing attacks that could degrade significantly network performance. Intrusion detection systems consist of a set of techniques designed to identify hostile behavior. In this paper, there are several approaches for intrusion detection in reconfigurable network routing such as collaborative, statistical, or machine learning-based techniques. In this paper, we introduce a new approach to intrusion detection for reconfigurable network routing based on linear systems theory. Using this approach, we can discriminate routing attacks by considering the system's z-plane poles. The z-plane can be thought of as a two dimensional feature space that arises naturally. It is independent of the number of network attack detection metrics and does not require extra dimensionality reduction. Two different host-based intrusion detection techniques, inspired by this new linear systems perspective, are presented and analyzed through a case study. The case study considers the effects of attack severity and node mobility to the attack detection performance. High attack detection accuracy was obtained without increasing packet overhead for both techniques by analyzing locally available information.

Adaptive Video Subsampling For Energy-Efficient Object Detection

Abstract: Energy-efficient computer vision is vitally important for embedded and mobile platforms where a longer battery life can allow increased deployment in the field. In image sensors, one of the primary causes of energy expenditure is the sampling and digitization process. Smart subsampling of the image-array in a manner that is task-specific, can result in significant savings of energy. We present an adaptive algorithm for video subsampling, which is aimed at enabling accurate object detection, while saving sampling energy. The approach utilizes objectness measures, which we show can be accurately estimated even from sub-sampled frames, and then uses that information to determine the adaptive sampling for the subsequent frame. We show energy savings of 18 - 67% with only a slight degradation in object detection accuracy in experiments. These results motivated us to further explore energy-efficient subsampling using advanced techniques such as, reinforcement learning and Kalman filtering. The experiments using these techniques are underway and provide ample support for adaptive subsampling as a promising avenue for embedded computer vision in the future.

Reconstruction-Free Compressive Vision for Surveillance Applications

Abstract: Compressed sensing (CS) allows signals and images to be reliably inferred from undersampled measurements. Exploiting CS allows the creation of new types of high-performance sensors includi...

A Modified Logistic Regression for Positive and Unlabeled Learning

Abstract: The positive and unlabeled learning problem is a semi-supervised binary classification problem. In PU learning, only an unknown percentage of positive samples are known, while the remaining samples, both positive and negative, are unknown. We wish to learn a decision boundary that separates the positive and negative data distributions. In this paper, we build on an existing popular probabilistic positive unlabeled learning algorithm and introduce a new modified logistic regression learner with a variable upper bound that we argue provides a better theoretical solution for this problem. We then apply this solution to both simulated data and to a simple image classification problem using the MNIST dataset with significantly improved results.

Articulation constrained learning with application to speech emotion recognition

Abstract: Speech emotion recognition methods combining articulatory information with acoustic features have been previously shown to improve recognition performance. Collection of articulatory data on a large scale may not be feasible in many scenarios, thus restricting the scope and applicability of such methods. In this paper, a discriminative learning method for emotion recognition using both articulatory and acoustic information is proposed. A traditional l1-regularized logistic regression cost function is extended to include additional constraints that enforce the model to reconstruct articulatory data. This leads to sparse and interpretable representations jointly optimized for both tasks simultaneously. Furthermore, the model only requires articulatory features during training; only speech features are required for inference on out-of-sample data. Experiments are conducted to evaluate emotion recognition performance over vowels /AA/, /AE/, /IY/, /UW/ and complete utterances. Incorporating articulatory information is shown to significantly improve the performance for valence-based classification. Results obtained for within-corpus and cross-corpus categorical emotion recognition indicate that the proposed method is more effective at distinguishing happiness from other emotions.

Audio Source Separation via Multi-Scale Learning with Dilated Dense U-Nets

Abstract: Modern audio source separation techniques rely on optimizing sequence model architectures such as, 1D-CNNs, on mixture recordings to generalize well to unseen mixtures. Specifically, recent focus is on time-domain based architectures such as Wave-U-Net which exploit temporal context by extracting multi-scale features. However, the optimality of the feature extraction process in these architectures has not been well investigated. In this paper, we examine and recommend critical architectural changes that forge an optimal multi-scale feature extraction process. To this end, we replace regular $1-$D convolutions with adaptive dilated convolutions that have innate capability of capturing increased context by using large temporal receptive fields. We also investigate the impact of dense connections on the extraction process that encourage feature reuse and better gradient flow. The dense connections between the downsampling and upsampling paths of a U-Net architecture capture multi-resolution information leading to improved temporal modelling. We evaluate the proposed approaches on the MUSDB test dataset. In addition to providing an improved performance over the state-of-the-art, we also provide insights on the impact of different architectural choices on complex data-driven solutions for source separation.

Introducing Machine Learning in Undergraduate DSP Classes

Abstract: Machine Learning (ML) and Artificial Intelligence (AI) algorithms are enabling several modern smart products and devices. Furthermore, several initiatives such as smart cities and autonomous vehicles utilize AI and ML computational engines. The current and emerging applications and the growing industrial interest in AI necessitate introducing ML algorithms at the undergraduate level. In this paper, we describe a series of activities to introduce ML in undergraduate digital signal processing (DSP) classes. These activities include a computational comparative study of ML algorithms for spoken digit recognition using spectral representations of speech. We choose spectral representations and features for speech as those concepts associate with the core topics in DSP such as FFT and autoregressive spectra. Our primary objective is to introduce undergraduate DSP students to feature extraction and classification using appropriate signal analysis and ML tools. An online module on ML along with a computer exercise are developed and assigned as a semester project in the DSP class. The exercise is developed in Python and also on the online JDSP HTML5 environments. An assessment study of the modules and computer exercises are also part of this effort.

Formation-aware Cloud Segmentation of Ground-based Images with Applications to PV Systems

Abstract: Ground-based sky imaging has won popularity due to its higher temporal and spatial resolution when compared with satellite or air-borne sky imaging systems. Cloud identification and segmentation is the first step in several areas, such as climate research and lately photovoltaic power generation forecast. Cloud-sky segmentation involves several variables including sun position and type and altitude of clouds. We proposed a training-free cloud/sky segmentation based on a threshold that adapts to the cloud formation conditions. Experimental results show that the proposed method reaches higher detection accuracy against state-of-the-art algorithms; additionally, qualitative results over hemispherical high dynamic range (HDR) sky images are provided. The proposed cloud segmentation method can be applied to shading prediction for photovoltaic (PV) systems.

Invenio: Discovering Hidden Relationships Between Tasks/Domains Using Structured Meta Learning

Abstract: Exploiting known semantic relationships between fine-grained tasks is critical to the success of recent model agnostic approaches. These approaches often rely on meta-optimization to make a model robust to systematic task or domain shifts. However, in practice, the performance of these methods can suffer, when there are no coherent semantic relationships between the tasks (or domains). We present Invenio, a structured meta-learning algorithm to infer semantic similarities between a given set of tasks and to provide insights into the complexity of transferring knowledge between different tasks. In contrast to existing techniques such as Task2Vec and Taskonomy, which measure similarities between pre-trained models, our approach employs a novel self-supervised learning strategy to discover these relationships in the training loop and at the same time utilizes them to update task-specific models in the meta-update step. Using challenging task and domain databases, under few-shot learning settings, we show that Invenio can discover intricate dependencies between tasks or domains, and can provide significant gains over existing approaches in terms of generalization performance. The learned semantic structure between tasks/domains from Invenio is interpretable and can be used to construct meaningful priors for tasks or domains.

Graph Filtering with Multiple Shift Matrices

Abstract: We propose a novel graph filtering method for semi-supervised classification that adopts multiple graph shift matrices to obtain more flexibility in dealing with misleading features. The resulting optimization problem is solved with a computationally efficient alternating minimization approach. In simulation experiments, we implement both conventional and our proposed graph filters as semi-supervised classifiers on real and synthetic datasets to demonstrate advantages of our algorithms in terms of classification performance.

Global Optimization of Graph Filters with Multiple Shift Matrices

Abstract: Using graphs to represent data sets that reside on irregular and complex structures can bring special advantages. Graph signal processing (DSP G ) converts traditional DSP operators, such as time shift, linear filters and Fourier transform, from time and frequency domain to the graph domain. In machine learning applications, DSP G provides an approach for semi-supervised classification. Different from conventional graph-filter-based classifiers, we propose a new graph filter with multiple graph shift matrices that can provide better classification performance when the feature quality is uneven. To solve the resulting non-convex problem, a tight and efficient convex relaxation approach is introduced. Through a branch and bound optimization method, we can find the mapping from the optimum relaxed parameter set to original parameter set, which technically provides the globally optimum solution. Simulation experiments corroborate our results.

Systems and methods for customizing kernel machines with deep neural networks

Abstract: A method including receiving an input data set. The input data set can include one of a feature domain set or a kernel matrix. The method also can include constructing dense embeddings using: (i) Nystrom approximations on the input data set when the input data set comprises the kernel matrix, and (ii) clustered Nystrom approximations on the input data set when the input data set comprises the feature domain set. The method additionally can include performing representation learning on each of the dense embeddings using a multi-layer fully-connected network for each of the dense embeddings to generate latent representations corresponding to each of the dense embeddings. The method further can include applying a fusion layer to the latent representations corresponding to the dense embeddings to generate a combined representation. The method additionally can include performing classification on the combined representation. Other embodiments of related systems and methods are also disclosed.

Distributed Spectral Radius Estimation in Wireless Sensor Networks

Abstract: A distributed algorithm to compute the spectral radius of the graph in the presence of additive channel noise is proposed. The spectral radius of the graph is the eigenvalue with the largest magnitude of the adjacency matrix, and is a useful characterization of the network graph. Conventionally, centralized methods are used to compute the spectral radius, which involves eigenvalue decomposition of the adjacency matrix of the underlying graph. We devise an algorithm to reach consensus on the spectral radius of the graph using only local neighbor communications, both in the presence and absence of additive channel noise. The algorithm uses a distributed max update to compute the growth rate in the node state values and then performs a specific update to converge on the logarithm of the spectral radius. The algorithm works for any connected graph structure. Simulation results supporting the theory are also presented.

Introducing Machine Learning in a Sophomore Signals and Systems Course

Abstract: This Innovative Practice Work in Progress Paper describes the experience and assessment of introducing machine learning concepts in a sophomore signals and systems course. Advanced machine learning concepts are typically covered in graduate level courses. However, as machine learning applications become more and more ubiquitous in our daily lives, it is important to expose students to machine learning concepts early at the undergraduate level. Signals and Systems I is a sophomore level course in the Electrical Engineering online bachelor degree curriculum. As the first course in signals and systems, it focuses on the basic concepts including signal transformation, linear time-invariant systems, Fourier series, Fourier transforms, Laplace and Z transforms. The course was taught using lecture videos and reading materials. MATLAB labs were also incorporated to introduce students to practical applications. Feedback from students showed their preference for more real world applications. This paper describes how a web-based simulation lab exercise was introduced to expose students to machine learning concepts. Specifically, students made the connection between machine learning and signals and systems concepts through a speech recognition application. In particular, students applied spectral analysis and identified voice features through pole/zero representation. To evaluate the effectiveness of the exercise, statistics from pre/post quizzes as well as student comments from a survey is analyzed.

Innovation Starts With Education: ICASSP 2019 Education Panel [SP Forum]

Abstract: The anniversary of a number of significant signal processing algorithms from the 1960s, including the least mean square algorithm and the Kalman filter, provided an opportunity at ICASSP 2019 to reflect on the links between education and innovation. This led ultimately to the proposal of some special sessions as well a panel session that would provide some insight, via a historical perspective, consideration of the current status, and an assessment of the emerging educational future.

Introducing machine learning concepts using hands-on Android-based exercises

Abstract: In this innovative practice work-in-progress paper, we discuss novel methods to teach machine learning concepts to undergraduate students. Teaching machine learning involves introducing students to complex concepts in statistics, linear algebra, and optimization. In order for students to better grasp concepts in machine learning, we provide them with hands-on exercises. These types of immersive experiences will expose students to the different stages of the practical uses of machine learning. The data collection apparatus is based on applications (apps) developed for the Android platform. Due to the accessible nature of the app and the exercises based on the app, this approach is useful for students across all majors.We provide the students with three different sets of activities, the first of which will introduce the basics of machine learning with specially designed artificial datasets. The second and third activities involve data collection, modeling, training, and testing, as applied to machine learning algorithms. The second activity will involve collecting touch/swipe data on mobile devices from students as they use a touch logger app. The third activity uses the Reflections app to collect cross-correlation data from rooms with different purposes. These hands-on activities guide the students through every step of the machine learning process. Student learning is assessed for each activity by holding workshops for undergraduate students. A workshop with the first activity outlining the basics of machine learning was given in the fall of 2018 and significant student learning was demonstrated. Workshops for the second and third activities are planned for the fall semester of 2019. Results from these workshops will be presented at the conference.

An REU Experience in Machine Learning and Computational Cameras

Abstract: In this work in progress paper, we describe an REU summer experience on imaging sensors that involved a female junior level Electrical Engineering student, a graduate student advisor, and three faculty. A research plan was designed to embed the student in a sensor and machine learning research with specific emphasis on energy-efficient cameras. The motivation for submitting this paper is the unique planning and the quality of the overall student experience which resulted in continuous engagement of the REU student with the faculty after the REU summer program completed. The program resulted in a major presentation at an international event, an NSF I/UCRC poster presentation, a research conference submission which is remarkable for an undergraduate student, and finally a new research direction for the graduate mentor and faculty. This paper describes successful strategies for research engagement for undergraduates in state-of-the-art research fields which yield positive outcomes for all participants, and is grounded in contemporary educational methodology and theoretical frameworks.

Interactive Signal Processing Education Applications for the Android Platform

Abstract: In this paper, we present a unique Android-DSP (AJDSP) application which was built from the ground up to provide mobile laboratory and computational experiences for educational use. AJDSP provides a mobile intuitive environment for developing and running signal processing simulations in a user-friendly. It is based on a block diagram system approach to support signal generation, analysis, and processing. AJDSP is designed for use by undergraduate and graduate students and DSP instructors. Its extensive functions enable instructors and students to simulate DSP concepts including convolution, Fourier transforms, z-transform and filter design. We describe the AJDSP functions and the process of using them in computer exercises, classroom demonstrations, and undergraduate remote laboratories. We include quantitative and qualitative assessments of AJDSP at the end of this paper. We also include descriptions of outreach efforts where we used AJDSP in middle schools and high schools to demonstrate how DSP algorithms enable several applications.

Board 162: Online Modules to Introduce Students to Solar Array Control using Neural Nets

Abstract: The growth in the field of machine learning (ML) can be attributed in part to the success of several algorithms such as neural networks as well as the availability of cloud computing resources. Recently, neural networks combined with signal processing analytics have found applications in renewable energy systems. With machine learning tools for solar array systems becoming popular, there is a need to train undergraduate students on these concepts and tools. In our undergraduate signal processing classes, we have developed self-contained modules to train students in this field. We specifically focused on developing modules with built-in software for applying neural nets (NN) to solar array systems where the NNs are used for solar panel fault detection and solar array connection topology optimization which are essentially ML classification tasks. We initially developed software modules in MATLAB and also developed these models on the user-friendly HTML-5 JavaDSP (JDSP) online simulation environment. J-DSP allows us to create and disseminate web-based laboratory exercises to train undergraduate students from different disciplines, in neural network applications. In this paper, we describe our efforts to enable students understand the properties of the main features of the data used, the types of ML algorithms that can be applied on solar energy systems, and the statistics of the overall results. The modules are injected in our undergraduate DSP class. The project outcomes are assessed using pre and post quizzes and student interviews. Introduction The introduction of machine learning algorithms to optimize renewable energy generation is emerging as an important topic in university education and industrial research. Although there are distinct courses in solar energy generation and machine learning, the creation of content at the overlap of these two areas offers several opportunities for education and research. Machine learning [1] promises to solve several problems in solar energy generation including a) fault detection [2,3,4,5], b) shading prediction [6], and c) topology optimization [7,8,9]. Preparing students early in their plans of study to tackle these problems requires: a) training in machine learning, b) exposition to solar energy systems simulation [10], c) skill building in terms of developing or using software to integrate machine learning to obtain solar system analytics and control the overall system [11]. In this paper, we describe an educational program developed to bring to undergraduate classes [12,13] select topics on the utility of machine learning in solar energy generation. The program consists of online modules developed specifically to: a) provide machine learning training to undergraduate students and b) show how ML can be applied to improve the performance of solar energy systems. The program was motivated in part by our Cyber-physical system research [6] and projects developed in our REU site [14,15] in which we embedded undergraduate students in machine learning tasks for solar energy applications. In our program, we used online modules in undergraduate classes and also adapted these modules for use in high school outreach. We describe in the paper online machine learning software that we developed specifically for online education purposes. This software is based on the HTML 5 J-DSP environment [16] which can run on any browser and is accessible freely by students and faculty. We also describe the content that accompanies the J-DSP software modules and computer exercises. To assess the modules and software, we develop preand postquizzes which we provide to students before and after the module experience. The rest of the paper is organized as follows. Section 2 describes an overview of the functionalities of the solar monitoring system developed by the SenSIP Center [3] and the need for signal processing and machine learning techniques for solar array analytics. Section 3 introduces machine learning algorithms that have been utilized in the exercises on solar array systems. Section 4 and 5 elucidates the exercise on fault detection and connection topology reconfiguration on solar array panels using machine learning. Section 6 provides the proposed evaluation method of the exercises given to the students and section 7 provides the conclusion and summarizes the contributions of the paper. Description of the Solar Monitoring and Control System The SenSIP Center has developed an 18kW solar array cyber-physical system research facility which consists of 104 PV panels [3]. Figure 1a and 1b show the solar array research facility with “smart monitoring devices” (SMDs) on each individual panel for monitoring the PV array. Every panel has an SMD associated with it. Each SMD has dedicated sensors that take measurements from every panel, actuators that can change the connections, and a radio that enables communications with a central hub and eventually the Internet. Sensor data includes voltage, current, irradiance and temperature. Data received from each SMD attached on every panel will be used to obtain analytics [17] and train ML algorithms for fault detection and solar array connection topology reconfiguration. The availability of sensor data from a solar site has great educational value as the students can learn about the behaviour of solar panels relative to environmental and other conditions. Each SMD has relay switches that enable real time dynamical reconfiguration of the PV array. i.e., an underperforming PV module can be bypassed or reconfigured in certain topologies to maximize power output. Students can learn that reconnecting solar panels in different configurations under shading can improve efficiency. The SMDs also have Zigbee wireless communicators which allow for wireless transmission of data to an access point or fusion centre. The data from the panels is then transmitted wirelessly over the Internet thus allowing obtaining analytics and enabling remote control of the entire array. Thus, we explore the possibility of fault detection and identification and array topology reconfiguration remotely. The functional overview of our CPS solar monitoring system is illustrated in Fig.2. It can be seen that by using machine learning algorithms such as neural networks that leverage data from the panels along with a server, issues such as solar panel fault detection and array topology reconfiguration can be solved with better ease involving fewer manual interactions. (a) (b) Fig.1: The Cyber physical (CPS) system allowing transmission of analytics and enabling the control of the system from remote sites. The figure shows (a) Solar array research facility with SMDs attached on each panel. (b) The SenSIP solar array facility. The educational value of this CPS facility is that students can learn about sensors and sensor fusion, the use of machine learning to optimize solar energy production. Fig. 2. Overview diagram of the CPS solar monitoring system. The CPS systems consists of sensors on each panel which measure the voltage, current, irradiance and temperature of the module. This data is transmitted wirelessly to the fusion centre allowing for fault detection and topological reconfiguration. Students learn the workings of a true CPS system and the utility of computers and ML in controlling solar energy production. Recent work [2,6,7,17] shows how various machine learning and signal processing techniques are used to detect faults, predict shading, and select connection topologies. Certain machine learning methods provide a generalized system that learns a useful mapping function between the input features and the output. The algorithms can also “learn” the non-linearities of the data which is specifically useful for solar array analytics due to the non-linear PV characteristics. In the case of solar panel fault detection, algorithms such as neural networks can classify a variety of dependent faults with high accuracy [5]. On the other hand, machine learning algorithms [8] can provide an endto-end system with reduced number of switching between the panels, as in the case of the connection topology reconfiguration of solar panels. The advantages of machine learning systems mentioned above emphasize the need for such algorithms for solar analytics. The exposition of undergraduate students to the basics and properties of ML algorithms through the solar energy application offers several opportunities for building skills in using intelligent algorithms for green energy applications. In the following sections, we introduce our modules in machine learning. Module on Machine Learning Algorithms There are several applications of ML including Internet of Things (IoT) [1], Energy [24], Smart home [25], natural language [26] and biomedical systems [27]. These applications are becoming part of everyday life and it is important for students to learn about the enabling ML technology. Learning from data enables a system to make decisions and activate devices to accomplish specific objectives. For example, in energy related applications one can train an ML model to detect faults or inefficiencies and then command the actuators embedded in the SMDs to reconnect modules and optimize the system or make it more resilient. Also, students can learn how ML algorithms often learn and reveal hidden insights from the data that is unknown otherwise. Many complex engineering tasks run ML algorithms in the background which attempt to understand and model system behaviour. By understanding both system characteristics and ML algorithm properties, one can provide an effective solution for a particular task. With this intent, we introduce various ML algorithms to undergraduate DSP students at an early stage in their curriculum in order to enable them to match algorithms to applications. Algorithms such as k-means clustering and Artificial Neural Networks (ANN) can be introduced and compared, and subsequently applied to fault

Spatially-Varying Sharpness Map Estimation Based on the Quotient of Spectral Bands

Abstract: Natural images suffer from defocus blur due to the presence of objects at different depths from the camera. Automatic estimation of spatially-varying sharpness has several applications including depth estimation, image quality assessment, information retrieval, image restoration among others. In this paper, we propose a sharpness metric based on the quotient of high- to low-frequency bands of the log-spectrum of the image gradients. Using the proposed sharpness metric, we obtain a descriptive dense sharpness map. We also propose a simple yet effective method to segment out-of-focus regions using a global threshold which is defined using weak textured regions present in the input image. Results over two publicly available databases show that the proposed method provides competitive performance when compared with state-of-the-art methods.