Abstract

A major problem in Acoustic Scene Classification (ASC) is a representation of an acoustic scene, which serves to be an important task for ASC. This study used Linear Prediction Cepstral Coefficients (LPCC) and Spectral Centroid Magnitude Cepstral Coefficients (SCMC) features along with log-Mel band energies for the representation of an acoustic scene. Deep Neural Networks (DNN) is being used to model the Acoustic Scene Classification (ASC). LPCCs are used to capture the changes in the auditory spectrum with time and SCMCs are used to capture the weighted average magnitude finely for a given acoustic scene subband. log-Mel band energies are used to capture the spectral envelopes of audio frame. The DNN architecture is used for audio track level classification. We have experimented on Detection and Classification of Acoustic Scenes and Events (DCASE) 2018 development dataset and DCASE 2017 dataset. We carried out experiments with individual feature sets, and also performed decision level DNN score fusions for improving the performance.

Abstract

An Electric Vehicle (EV) charger, also called Electric Vehicle Charging Station (EVCS) is an infrastructure element that supplies electrical energy for recharging EVs. The paper focuses on the design and fabrication of two-wheeler (escooter) electric vehicle charging equipment. Considering the future scenario of mass privatization of EV two-wheelers on Indian roads, the outline discusses the product design based on fabricability, affordability, and ability to mass manufacture. The proposed architecture follows the Level 2 charging standards (240 Volts), and is based on the open charge point protocol (OCPP). The suggested EVCS is constructed considering the safety prerequisites of system administrators, installers, consumers, government agencies and others. The design of EVCS links to three industries: equipment manufacturers, software industry and electric power networks. This paper presents design considerations by elaborating the hardware, software, and protocols followed to design the Level 2 charging standard EVCS.

Abstract

Accelerometers are among the most mature sensor technologies with a broad range of applications in multiple fields and industries. They represent the most widely used microelectromechanical system (MEMS) devices with excellent and reliable performance. MEMS acceleration sensors established dominance mainly in navigation and control applications. In recent years, however, recent technologies and materials have emerged that introduce novel sensing mechanisms, improve performance, enable customization, reduce cost, and reduce fabrication complexity. Herein, the recent advances in accelerometers based on MEMS and recent emerging technologies are reviewed. This work provides a comprehensive review of accelerometers’ sensing mechanisms and the main characteristics and features of each type of sensor, material, and fabrication strategies used to fabricate them. From the aspect of sensor application, this work focuses on reviewing applications that demonstrate the use of accelerometers manufactured using unconventional technologies and materials in prevailing fields such as healthcare monitoring, automotive industry, navigation, building, and structural monitoring. Moreover, challenges and future efforts needed to be addressed in this field are summarized

Abstract

Point clouds, being the simple and compact representation of surface geometry of 3D objects, have gained increasing popularity with the evolution of deep learning networks for classification and segmentation tasks. Unlike human, teaching the machine to analyze the segments of an object is a challenging task and quite essential in various machine vision applications. In this paper, we address the problem of segmentation and labelling of the 3D point clouds by proposing a inception based deep network architecture called PIG-Net, that effectively characterizes the local and global geometric details of the point clouds. In PIG-Net, the local features are extracted from the transformed input points using the proposed inception layers and then aligned by feature transform. These local features are aggregated using the global average pooling layer to obtain the global features. Finally, feed the concatenated local and global features to the convolution layers for segmenting the 3D point clouds. We perform an exhaustive experimental analysis of the PIG-Net architecture on two state-of-the-art datasets, namely, ShapeNet [1] and PartNet [2]. We evaluate the effectiveness of our network by performing ablation study.

Abstract

For decades, the revolution in design and fabrication methodology of flexible capacitive pressure sensors using various inorganic/organic materials has significantly enhanced the field of flexible and wearable electronics with a wide range of applications in aerospace, automobiles, marine environment, robotics, healthcare, and consumer/portable electronics. Mathematical modelling, finite element simulations, and unique fabrication strategies are utilized to fabricate diverse shapes of diaphragms, shells, and cantilevers which function in normal, touch, or double touch modes, operation principles inspired from microelectromechanical systems (MEMS) based capacitive pressure sensing techniques. The capacitive pressure sensing technique detects changes in capacitance due to the deformation/deflection of a pressure sensitive mechanical element that alters the separation gap of the capacitor. Due to advancement in state-of-the-art fabrication technologies, the performance and properties of capacitive pressure sensors are enhanced. In this review paper, recent progress in flexible capacitive pressure sensing techniques in terms of design, materials, and fabrication strategies is reported. The mechanics and fabrication steps of paper-based low-cost MEMS/flexible devices are also broadly reported. Lastly, the applications of flexible capacitive pressure sensors, challenges, and future perspectives are discussed.

Abstract

enote the sum of inverses of odd divisors of a positive integer n, and let cr (n) be the number of representations of n as a sum of r squares where representations with different orders and different signs are counted as distinct. The aim is of this note is to prove the following interesting combinatorial identity: ∑ d|n d≡1(2)

Abstract

Earthquake response control of base-isolated bridges using supplementary passive dampers is presented. A benchmark highway overcrossing located in California is selected to indicate the effectiveness of such hybrid response control schemes. The bridge is modeled as a lumped mass three-dimensional stick model. The coupled differential equations of motion for the system are derived and solved using Newmark’s step-by-step iteration method. Considering a bidirectional interaction of the seismic forces, the bridge responses are investigated using various combinations of base isolators and damper systems, called hybrid systems. Further, a parametric study is carried out to investigate the dynamic response of the highway bridge by changing the characteristic parameters of the control devices and the isolation period of the structure for different hybrid systems under various real earthquake ground motions. The variations in seismic responses, such as bearing displacement and base shear in the bridge pier, are computed and compared to study the effectiveness of the various damper-isolator combinations. The present study concludes better performance of the fluid viscous damper when used as supplementary dampers with the lead-core rubber bearing or friction pendulum system for the base isolation, leading to reduced bearing displacements and pier base shear.

Abstract

Surface normal estimation is an essential component of several computer and robot vision pipelines. While this problem has been extensively studied, most approaches are geared towards indoor scenes and often rely on multiple modalities (depth, multiple views) for accurate estimation of normal maps. Outdoor scenes pose a greater challenge as they exhibit significant lighting variation, often contain occluders, and structures like building facades are often ridden with numerous windows and protrusions. Conventional supervised learning schemes excel in indoor scenes, but do not exhibit competitive performance when trained and deployed in outdoor environments. Furthermore, they involve complex network architectures and require many more trainable parameters. To tackle these challenges, we present an adversarial learning scheme that regularizes the output normal maps from a neural network to appear more realistic, by using a small number of precisely annotated examples. Our method presents a lightweight and simpler architecture, while improving performance by at least 1.5x across most metrics. We evaluate our approaches against the state-of-the-art on normal map estimation, on a synthetic and a real outdoor dataset, and observe significant performance enhancements.

Abstract

The explosive growth of internet users and connected devices increased the threat vector surface. However, there is no single website or a search engine that provides information on vulnerabilities, threats, attacks, controls, etc. Ambiguity, bias and lack of credibility are some of the alarming issues while dealing with generic search engines on sensitive topics such as ‘Health’ and ‘Information Security’. A dedicated information security specific search engine benefits various stakeholders including security professionals, researchers, government, regulators and others. We implemented a fine grained approach that identifies sub-domains of information security, extracts related URLs and content and assesses search results credibility to enhance adoption of information security specific search engine. To identify sub-domains and extract seed and child URLs, a fine grained approach that extends an efficient Artificial Bee Colony algorithm was implemented. About 34,007 seed URLs and 400,726 child URLs of various sub-domains of the information security were extracted. The results of the proposed approach identified more URLs (seed and child) of sub-domains as compared to existing approaches while consuming less computing resources.

Abstract

Explainable AI aims to build systems that not only give high performance but also are able to provide insights that drive the decision making. However, deriving this explanation is often dependent on fully annotated (class label and local annotation) data, which are not readily available in the medical domain. Approach: This paper addresses the above-mentioned aspects and presents an innovative approach to classifying a lung nodule in a CT volume as malignant or benign, and generating a morphologically meaningful explanation for the decision in the form of attributes such as nodule margin, sphericity, and spiculation. A deep learning architecture that is trained using a multi-phase training regime is proposed. The nodule class label (benign/malignant) is learned with full supervision and is guided by semantic attributes that are learned in a weakly supervised manner. Results: Results of an extensive evaluation of the proposed system on the LIDC-IDRI dataset show good performance compared with state-of-the-art, fully supervised methods. The proposed model is able to label nodules (after full supervision) with an accuracy of 89.1% and an area under curve of 0.91 and to provide eight attributes scores as an explanation, which is learned from a much smaller training set. The proposed system's potential to be integrated with a sub-optimal nodule detection system was also tested, and our system handled 95% of false positive or random regions in the input well by labeling them as benign, which underscores its robustness. Conclusions: The proposed approach offers a way to address computer-aided diagnosis system design under the constraint of sparse availability of fully annotated images.