- Advanced Memory and Neural Computing
- Sparse and Compressive Sensing Techniques
- Supply Chain and Inventory Management
- Auction Theory and Applications
- Graph theory and applications
- Advanced Combinatorial Mathematics
- Consumer Market Behavior and Pricing
- Advanced Graph Theory Research
- Advanced Neural Network Applications
- Blind Source Separation Techniques
- Physical Unclonable Functions (PUFs) and Hardware Security
- Commutative Algebra and Its Applications
- CCD and CMOS Imaging Sensors
- Advanced Optimization Algorithms Research
- Microwave Imaging and Scattering Analysis
- Data Visualization and Analytics
- Topology Optimization in Engineering
- Matrix Theory and Algorithms
- Innovative Human-Technology Interaction
- Graph Labeling and Dimension Problems
- Gear and Bearing Dynamics Analysis
- Design Education and Practice
- Neuroscience and Neural Engineering
- Complex Systems and Decision Making
- IoT and Edge/Fog Computing
Dr. Babasaheb Ambedkar Marathwada University
2023
Kokilaben Dhirubhai Ambani Hospital
2022
Indian Institute of Technology Gandhinagar
2022
Analytical Services
2020
Savitribai Phule Pune University
1993-2019
Tennessee Cancer Specialists
2019
University of Maryland, Baltimore County
2014-2018
Tata Consultancy Services (India)
2017
Romax Technology (United Kingdom)
2011-2013
National Aerospace Laboratories
2010
Fueled by ImageNet Large Scale Visual Recognition Challenge and Common Objects in Context competitions, the convolutional neural network (CNN) has become important computer vision natural language processing. However, state-of-the-art CNNs are computationally memory-intensive, thus energy-efficient implementation on embedded platform is challenging. Recently, VGGNet ResNet showed that deep networks with more convolution layers a few fully connected can achieve lower error rates, reducing...
This work presents a low-power, embedded ECG pattern recognition system for the purpose of biometric authentication. We believe that coupled with secondary marker such as fingerprint will play key role in wearable security wearables' popularity continues to grow. The objective this is implement reliable, robust, and fast while maintaining low area power footprint. A streamlined approach was devised utilized neural networks both identify QRS complex segments signal then perform user...
Orthogonal Matching Pursuit (OMP) is an important compressive sensing (CS) recovery and sparsity inducing algorithm, which has potential in various emerging applications ranging from wearable mobile computing to real-time analytics processing on servers. Thus application aware OMP algorithm implementation important. In this paper, we propose two different modifications named Thresholding technique for (tOMP) Gradient Descent (GDOMP) reduce hardware complexity of algorithm. tOMP modifies...
Hardware Trojans inserted during design or fabrication time by untrustworthy house foundry possesses important security concerns. These lead to un-desired change in functionality of the and provide easy access sensitive information. attacks malicious activities are triggered based on very rare conditions, which can evade test-time Trojan detection but arise long hours field operation. In this paper we propose a run-time architecture for custom many-core Machine Learning technique. We exploit...
In this paper, we address the applicability of Blockchain technology to ensure security data transmitted and received by nodes in an Internet Things (IoT) network. We propose a consensus model that is suitable for resource constrained devices. also implementing IoT on top model. simulate our proposed understand its feasibility.
Hardware Trojans inserted at the time of design or fabrication by untrustworthy house foundry, poses important security concerns. With increase in attacker's resources and capabilities, we can anticipate an unexpected new attack from attacker run-time. Therefore, challenge is not only to reduce hardware overhead added feature but also secure attacks introduced real-time. In this work, propose a Real-time Online Learning approach for Securing many-core design. order prevent attacks, provides...
The authors prove the stepwise stability for a finite difference scheme heat equation with an integral constraint. resulting matrix is nonsymmetric and does not have usual band structure. proof based on method of analysis. eigenvalues several matrices are found explicitly or their location described precisely. This relies upon relationship characteristic polynomials these orthogonal polynomials.
<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> Auction-based mechanisms are extremely relevant in modern day electronic procurement systems since they enable a promising way of automating negotiations with suppliers and achieve the ideal goals efficiency cost minimization. This paper surveys recent research current art area auction-based for e-procurement. The survey delineates different representative scenarios e-procurement where auctions...
In this article, we propose a real-time anomaly detection framework for an NoC-based many-core architecture. We assume that processing cores and memories are safe is included through communication medium (i.e., router). The article targets three different attacks, namely, traffic diversion, route looping, core address spoofing attacks. attacks detected by using machine-learning techniques. Comprehensive analysis on algorithms suggests Support Vector Machine (SVM) K-Nearest Neighbor (K-NN)...
Physically Unclonable Functions (PUFs) have proved to be an effective and low-cost measure against counterfeiting by providing device authentication secure key storage services. Memory-based PUF implementations are attractive option due the ubiquitous nature of memory in electronic devices requirement minimal (or no) additional circuitry. Dynamic Random Access Memory-- (DRAM) based PUFs particularly advantageous their large address space multiple controllable parameters during response...
FueXeA by ILSVRC and COCO competitions, Convolutional Neural Network (CNN) has become important in computer vision, natural language processing. However state-of-the-art CNNs are computationally memory intensive, thus energy efficient implementation on embedded platform is challenging. Recently VGGNet ResNet showed that deep neural networks with more convolution layers (CV) few fully connected layer (FC) can achieve lower error rates, reducing the complexity of utmost importance. To reduce...
Compressive Sensing (CS) signal reconstruction can be implemented using convex relaxation, non-convex, or local optimization algorithms. Though the optimization, such as Iterative Hard Thresholding algorithm, is more accurate than matching pursuit algorithms, most researchers focus on algorithms because they are less computationally complex. Orthogonal Matching Pursuit (OMP) a greedy which solves problem by choosing significant variable to reduce least square error. In this paper, we propose...
Decomposition of large engineering design problems into smaller subproblems enhances robustness and speed numerical solution algorithms. Design can be solved in parallel, using the optimization technique most suitable for underlying subproblem. This also reflects typical multidisciplinary nature system allows better interpretation results. Hierarchical overlapping coordination (HOC) simultaneously uses two or more problem decompositions, each them associated with different partitions...
In this article we provide a combinatorial description of an arbitrary minor the Laplacian matrix (L) mixed graph (a with some oriented and unoriented edges). This is generalized Matrix Tree Theorem. We also characterize non-singular substructures graph. The sign attached to nonsingular substructure described in terms labeling number edges included certain paths. Nonsingular may be viewed as matchings, because case disjoint vertex sets corresponding rows columns L, our Theorem provides...
Compressive Sensing (CS) is a novel scheme, in which signal that sparse known transform domain can be reconstructed using fewer samples. However, the reconstruction techniques are computationally intensive and power consuming, make them impractical for embedded applications. This work presents parallel reconfigurable architecture Orthogonal Matching Pursuit (OMP) algorithm, one of most popular CS algorithms. In this paper, we proposing first OMP take different image sizes with sparsity up to...
The variability of deep-submicron technologies creates systems with asymmetric cores from a frequency and leakage power viewpoint, which makes an opportunity for performance-power optimization. In particular, process variation can transform homogeneous many-core platform into heterogeneous system where the task mapping is NP-hard problem. this paper, we propose algorithm that selects appropriate along voltage assignment cluster cores. algorithm, based on simulated annealing, determines...