Deepak Dudeja

Prof. Dr. Deepak Dudeja

Computer Science and Engineering, Maharishi Markandeshwar University, Mullana, Ambāla, India
orcid
Professor and Cluster Head

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
31 December 2022 in Journal of Imaging
Verified
Journal Article
Open access
PDF file
Arun Singh YadavArun Singh Yadav
Arun Singh Yadav
SKSurendra Kumar
Surendra Kumar
GKGirija Rani Karetla
Girija Rani Karetla
JCJuan Carlos Cotrina-Aliaga
Juan Carlos Cotrina-Aliaga
JGJosé Luis Arias Gonzales
José Luis Arias Gonzales
NTNisha Tatkar
Nisha Tatkar
Background and Objectives: Brain Tumor Fusion-based Segments and Classification-Non-enhancing tumor (BTFSC-Net) is a hybrid system for classifying brain tumors that combine medical image fusion, segmentation, feature extraction, and classification procedures. Materials and Methods: to reduce noise from medical images, the hybrid probabilistic wiener filter (HPWF) is first applied as a preprocessing step. Then, to combine robust edge analysis (REA) properties in magnetic resonance imaging (MRI) and computed tomography (CT) medical images, a fusion network based on deep learning convolutional neural networks (DLCNN) is developed. Here, the brain images’ slopes and borders are detected using REA. To separate the sick region from the color image, adaptive fuzzy c-means integrated k-means (HFCMIK) clustering is then implemented. To extract hybrid features from the fused image, low-level features based on the redundant discrete wavelet transform (RDWT), empirical color features, and texture characteristics based on the gray-level cooccurrence matrix (GLCM) are also used. Finally, to distinguish between benign and malignant tumors, a deep learning probabilistic neural network (DLPNN) is deployed. Results: according to the findings, the suggested BTFSC-Net model performed better than more traditional preprocessing, fusion, segmentation, and classification techniques. Additionally, 99.21% segmentation accuracy and 99.46% classification accuracy were reached using the proposed BTFSC-Net model. Conclusions: earlier approaches have not performed as well as our presented method for image fusion, segmentation, feature extraction, classification operations, and brain tumor classification. These results illustrate that the designed approach performed more effectively in terms of enhanced quantitative evaluation with better accuracy as well as visual performance.

27 June 2023 in Measurement: Sensors
Verified
Journal Article
Open access
VSVoore Subrahmanyam
Voore Subrahmanyam
SKSarvesh Kumar
Sarvesh Kumar
SATYAJEE SRIVASTAVASATYAJEE SRIVASTAVA
SATYAJEE SRIVASTAVA
ABAnkur Singh Bist
Ankur Singh Bist
BSBasant Sah
Basant Sah
PBPawan Bhambu
Pawan Bhambu

13 December 2023 in Engineering Proceedings
Verified
Conference Paper
Open access
PDF file
Deepak DudejaDeepak Dudeja
Deepak Dudeja
SSShweta Mayor Sabharwal
Shweta Mayor Sabharwal
YGYatish Ganganwar
Yatish Ganganwar
MSManoj Singhal
Manoj Singhal
NGNITIN GOYAL
NITIN GOYAL
ATAshish Tiwari
Ashish Tiwari
Recent trends have shown a greatly increasing number of users in the digital world, so there is a need for a large number of resources. To handle these resources, there is the need to manage and schedule in an optimized manner using artificial intelligence (AI) systems. These systems deal with the business-common method of managing offerings. Ordinary models consolidate inbound deals, outbound bargains, account-based offerings, or a mix of diverse models. An organization model may gather multiple choices that an organization makes over a long period of time, considering a system, cycle, or trade. In our approach, computational resources are treated as commodities that can be bought and sold in a decentralized marketplace. Agents representing AI tasks or workloads participate in resource auctions, competing for the resources they need. The allocation of resources is determined through competitive bidding, where the highest bidder secures the required resources. This approach encourages efficient resource utilization and fair distribution based on the tasks’ priorities and value. Our sales-based models for resource management and scheduling offer a promising solution for optimizing AI systems’ resource allocation. By applying principles from auction theory and market dynamics, AI systems can become more adaptive, responsive, and efficient in managing computational resources, ultimately leading to improved performance and resource utilization.

1 April 2016
Verified
Conference Paper
Closed access
NCNeetu Chaudhari
Neetu Chaudhari
SATYAJEE SRIVASTAVASATYAJEE SRIVASTAVA
SATYAJEE SRIVASTAVA
We have entered in data deluge already. Data Deluge means data generated by IoT devices and humans simultaneously. The data deluge is a Big threat for technologist but beneficial for end users. Now the coming problem is the security of this data. Big Data is too big, too fast and too diverse that does not compile with traditional data base system. Traditional data base systems are very good to analyze structured data but these systems are not enough to analyze unstructured data. In this paper we discourse the possible challenges and security issues related to Big Data characteristics and possible solutions.

1 January 2022 in Wireless Communications and Mobile Computing
Verified
Journal Article
Open access
PDF file
Deepak DudejaDeepak Dudeja
Deepak Dudeja
SHSabeena Yasmin Hera
Sabeena Yasmin Hera
NDNitika Vats Doohan
Nitika Vats Doohan
NDNilesh Dubey
Nilesh Dubey
MRMAHAVEERAKANNAN R
MAHAVEERAKANNAN R
SHSimon Karanja Hinga
Simon Karanja Hinga
The extensive use of sensor technology in every sphere of life, along with the continuous digitization of society, makes it realistic to anticipate that the planet will soon be patched with small‐sized devices all over the place in the not‐too‐distant future. These devices give monitoring and surveillance capabilities, as well as access to a vast digital universe of information, among other things. Finding data and information, as well as processing enquiries, is made much easier thanks to the seamless transmission of information over wireless media, which is based on the “anywhere, anytime, everywhere” paradigm that allows information to be sent anywhere, at any time. Sensing networks came into existence as a consequence of the downsizing of wireless devices that are capable of receiving information from a source, transferring it, and processing it. Sensor networks, although they share many of the features, uses, and limits of ad hoc networks, have their own set of capabilities that are unique to them. While performing their responsibilities, sensor networks must contend with a variety of security issues, including unsecured wireless channels, physical compromise, and reprogramming. Because of the small size and ubiquitous availability of wireless sensor networks, compromise attacks are the most severe kind of attack in this environment (WSNs). With the proliferation of wireless sensor networks (WSNs), it is becoming more difficult to rely only on machine learning techniques. We sought to tackle the security challenge by developing a key management system as well as a secure routing mechanism. We are building scalable key management approaches that are resistant to node compromise and node replication attacks, which we will demonstrate in our proposed study, by using deployment‐driven localization of security information and leveraging distributed key management. Using a security‐aware selection of the next hop on the route to the destination, we were able to design safe routing algorithms that were effective.

11 December 2023 in Engineering Proceedings
Verified
Conference Paper
Open access
PDF file
SPSunita Pachar
Sunita Pachar
Deepak DudejaDeepak Dudeja
Deepak Dudeja
NBNeha Batra
Neha Batra
VTVinam Tomar
Vinam Tomar
JBJohn Philip Bhimavarapu
John Philip Bhimavarapu
ASAvadh Kishore Singh
Avadh Kishore Singh
To be effective, evidence-driven disaster risk management (DRM) relies on a wide variety of data types, information sources, and models. Weather modeling, the rupture of earthquake fault lines, and the creation of dynamic urban exposure measures all require extensive data collection from a variety of sources in addition to complex science. There are various methodologies to utilize AI to recognize necessities and asset accessibility by the likes of Twitter; however, the foremost broadly recognized and exact strategies remain cloudy. Within the occurrence of a catastrophe, machine learning apparatuses for designating assets are required to instantly help those in need. This overview appears to be necessary for additional examination with respect to an assertion on endorsed methods for calculation to demonstrate assurance, benchmarking datasets, crisis word references, word embedding techniques, and evaluation methods. As fiascos of all sorts become more common, these devices have the potential to improve real-time crisis administration over all stages of a catastrophe. This study aims to provide readers, including data scientists, with a clear and uncomplicated reference on how disaster risk management systems can benefit from machine learning. There are numerous sources of information on this set of technologies, which are both complicated and constantly changing. The volume of sensor data that can be analyzed has increased exponentially because of enormous increases in computational speed and capacity over the past few decades.