Dr. Nagender Aneja

Verified email at ieee.org

Researcher, Institute of Applied Data Analytics
Universiti Brunei Darussalam



                                              

http://researchid.co/naneja

I am a Deep Learning and Patent Professional. I recently developed projects in Histopathologic Cancer Detection, (ranked 65th among 1157 - top 6%) ISIC 2018: Skin Lesion Analysis Towards Melanoma Detection, https://github.com/naneja/isic2018 Automated Plant Species Identification, https://github.com/naneja/plants Device Fingerprinting to identify network devices using deep learning https://github.com/naneja/device-fingerprinting Applied CNN-LSTM for Image Captioning, Identified Key facial points. Most of my work is available at https://github.com/naneja

I am also the founder and developer of https://researchid.co

EDUCATION

Ph.D. Computer Engineering
M.E. Computer Technology and Applications

RESEARCH INTERESTS

Deep Learning
Computer Vision

FUTURE PROJECTS

DEEP LEARNING TRAINING WITH LIMITED DATA

Deep learning needs lots of data for training; however, in some industrial applications, the significant amount of data may not be available, limiting the deep learning approach. Modern techniques like transfer learning and generative adversarial networks show some hope to solve this challenge. The objective of the project is to propose new techniques for deep learning training.


Applications Invited
Remote Research Collaboration

DEEP LEARNING SECURITY

Deep-learning networks are susceptible to butterfly effect wherein small alterations in the input data can point to drastically distinctive outcomes, making the deep learning network inherently volatile. Thus, the output of deep learning network may be controlled by altering its input or by adding noise. Research has shown that it is possible to fool the deep learning network by adding an imperceptible amount of noise in the input.


Applications Invited
Remote Research Collaboration

GENERATIVE ADVERSARIAL NETWORKS - REVERSE IMAGE CAPTIONING - TEXT TO IMAGE AND SCALING GAN TRAINING WITH BATCH SIZE

Generative Adversarial Networks may have potential to solve the text-to-image problem, but there are challenges in using GANs for NLP. Image classification have got benefitted with large mini-batches and one of the open question the question https://distill.pub/2019/gan-open-problems/#batchsize is if they can also help to scale GANs


Applications Invited
Remote Research Collaboration
12

Scopus Publications

122

Google Scholar Citations

7

Google Scholar h-index

4

Google Scholar i10-index

Scopus Publications

  • Transfer Learning using CNN for Handwritten Devanagari Character Recognition
    Nagender Aneja and Sandhya Aneja

    1st IEEE International Conference on Advances in Information Technology, ICAIT 2019 - Proceedings, Pages: 293-296, Published: July 2019 IEEE
    This paper presents an analysis of pre-trained models to recognize handwritten Devanagari alphabets using transfer learning for Deep Convolution Neural Network(DCNN). This research implements AlexNet, DenseNet, Vgg, and Inception ConvNet as a fixed feature extractor. We implemented 15 epochs for each of AlexNet, DenseNet 121, DenseNet 201, Vgg 11, Vgg 16, Vgg 19, and Inception V3.Results show that Inception V3 performs better in terms of accuracy achieving 99% accuracy with average epoch time 16.3 minutes while AlexNet performs fastest with 2.2 minutes per epoch and achieving 98%accuracy.

  • IoT Device fingerprint using deep learning
    Sandhya Aneja, Nagender Aneja and Md Shohidul Islam

    Proceedings - 2018 IEEE International Conference on Internet of Things and Intelligence System, IOTAIS 2018, Pages: 174-179, Published: 3 January 2019 Institute of Electrical and Electronics Engineers Inc.
    Device Fingerprinting (DFP) is the identification of a device without using its network or other assigned identities including IP address, Medium Access Control (MAC) address, or International Mobile Equipment Identity (IMEI) number. DFP identifies a device using information from the packets which the device uses to communicate over the network. Packets are received at a router and processed to extract the information. In this paper, we worked on the DFP using Inter Arrival Time (IAT). IAT is the time interval between the two consecutive packets received. This has been observed that the IAT is unique for a device because of different hardware and the software used for the device. The existing work on the DFP uses the statistical techniques to analyze the IAT and to further generate the information using which a device can be identified uniquely. This work presents a novel idea of DFP by plotting graphs of IAT for packets with each graph plotting 100 IATs and subsequently processing the resulting graphs for the identification of the device. This approach improves the efficiency to identify a device DFP due to achieved benchmark of the deep learning libraries in the image processing. We configured Raspberry Pi to work as a router and installed our packet sniffer application on the Raspberry Pi. The packet sniffer application captured the packet information from the connected devices in a log file. We connected two Apple devices iPad4 and iPhone 7 Plus to the router and created IAT graphs for these two devices. We used Convolution Neural Network (CNN) to identify the devices and observed the accuracy of 86.7%.

  • Profile-based Ad hoc social networking using Wi-Fi direct on the top of android
    Nagender Aneja and Sapna Gambhir

    Mobile Information Systems, ISSN: 1574017X, eISSN: 1875905X, Volume: 2018, Published: 2018 Hindawi Limited
    Ad hoc social networks have become popular to support novel applications related to location-based mobile services that are of great importance to users and businesses. Unlike traditional social services using a centralized server to fetch location, ad hoc social network services support infrastructure-less real-time social networking. It allows users to collaborate and share views anytime anywhere. However, current ad hoc social network applications either are not available without rooting the mobile phones or do not filter the nearby users based on common interests without a centralized server. This paper presents an architecture and implementation of social networks on commercially available mobile devices that allow broadcasting name and a limited number of keywords representing users’ interests without any connection in a nearby region to facilitate matching of interests. The broadcasting region creates a digital aura and is limited by the Wi-Fi region that is around 200 meters. The application connects users to form a group based on their profile or interests using the peer-to-peer communication mode without using any centralized networking or profile-matching infrastructure. The peer-to-peer group can be used for private communication when the network is not available.

  • Social Profile Aware AODV Protocol for Ad-Hoc Social Networks
    Nagender Aneja and Sapna Gambhir

    Wireless Personal Communications, ISSN: 09296212, eISSN: 1572834X, Pages: 4161-4182, Published: 1 December 2017 Springer New York
    Ad-hoc social networks are required to strengthen local communication between people. Mobile ad-hoc social networks have emerged as self-configuring and self-organizing social networks to facilitate interactions among different mobile users without Internet. Contextual routing based on social patterns has been proposed and advantageous for ad-hoc social networks. Social profile aware routing protocol proposed in this paper allows users to use social networking applications using social routing protocol. The protocol has been implemented on network simulator ns-2 and is also available as a patch file for other researchers. Results indicate protocol has low overhead with 64 nodes. Results have been presented for packet delivery ratio, and average end-to-end delay. The need of multi-hop social network was also studied and observed that probability of nodes being connected at mult-hop increases with increment of number of nodes and geographical area.

  • Piecewise Maximal Similarity for Ad-hoc Social Networks
    Sapna Gambhir, Nagender Aneja and Liyanage Chandratilake De Silva

    Wireless Personal Communications, ISSN: 09296212, eISSN: 1572834X, Pages: 3519-3529, Published: 1 December 2017 Springer New York
    Computing Profile Similarity is a fundamental requirement in the area of Social Networks to suggest similar social connections that have high chance of being accepted as actual connection. Representing and measuring similarity appropriately is a pursuit of many researchers. Cosine similarity is a widely used metric that is simple and effective. This paper provides analysis of cosine similarity for social profiles and proposes a novel method to compute Piecewise Maximal Similarity between profiles. The proposed metric is 6% more effective to measure similarity than cosine similarity based on computations on real data.

  • Security and privacy: Challenges and defending solutions for nosql data stores
    Ganesh Chandra Deka

    NoSQL: Database for Storage and Retrieval of Data in Cloud, Pages: 237-250, Published: 1 January 2017 CRC Press

  • Software design for social profile matching algorithm to create ad-hoc social network on top of Android
    Proceedings of the 10th INDIACom; 2016 3rd International Conference on Computing for Sustainable Global Development, INDIACom 2016, Pages: 3450-3454, Published: 27 October 2016

  • Need of ad-hoc social network based on users' dynamic interests
    Sapna Gambhir, Nagender Aneja and Samridhi Mangla

    International Conference on Soft Computing Techniques and Implementations, ICSCTI 2015, Pages: 52-56, Published: 10 June 2016 Institute of Electrical and Electronics Engineers Inc.
    Ad-hoc social network (ASN) is a location based network that makes use of ad-hoc network to connect interested users socially. Ad-hoc social network is a combination of social network that maintains profile and interests of users, and ad-hoc network that helps to connect nearby users without centralized access point. A survey was conducted to know users' perception and preferences for ASN. This paper presents survey and results for need of ad-hoc social network. Results also indicate that in users prefer 75% of average profile similarity to connect nearby users.

  • Geo-social semantic profile matching algorithm for dynamic interests in Ad-hoc social network
    Nagender Aneja and Sapna Gambhir

    Proceedings - 2015 IEEE International Conference on Computational Intelligence and Communication Technology, CICT 2015, Pages: 354-358, Published: 1 April 2015 Institute of Electrical and Electronics Engineers Inc.
    Ad-hoc Social Network (ASN) allows users to create social network connections using wireless ad hoc network. Various techniques have been proposed to create ASN by matching profiles of users. In order to create meaningful ASN, there is a need to dynamically set and match user profile based on changing user's interests. This paper provides an algorithm to semantically match users profiles based on geographic location and dynamic interests.

  • Verification of forecasts from high resolution numerical weather prediction model
    Nagender Aneja and Thomas George

    Research Journal of Applied Sciences, Engineering and Technology, ISSN: 20407459, eISSN: 20407467, Pages: 1255-1258, Published: 2014 Maxwell Science Publications
    Assessment of forecast quality is a critical component for weather model development as well as evaluating the impact on weather sensitive business applications such as renewable energy forecasting, agriculture, insurance etc. This study presents forecast quality results of a high resolution numerical weather model deployed for the country of Brunei at Universiti Brunei Darussalam. We present the monthly accuracy and probability of detection scores for precipitation as well as accuracy scores for Relative Humidity (RH) and Dew Point Temperature (DPT) for the year 2013.

  • Minimum Exposed Path to the Attack (MEPA) in Mobile Ad hoc Network (MANET)
    Sandhya Khurana, Neelima Gupta and Nagender Aneja

    Proceedings of the Sixth International Conference on Networking, ICN'07, Published: 2007 IEEE
    Lack of infrastructure, central controlling authority and the properties of wireless links make mobile ad hoc networks (MANETs) vulnerable to attacks. Several protocols have been proposed to make the routing protocols handle attacks in MANETs. These protocols detect the misbehaving nodes and re-route the data packets around them, mostly along the shortest such path. However, no single protocol handles all the attacks. A variant of the problem for routing around misbehaving nodes in ad hoc networks can be stated as: given a set of nodes under the danger of attack, one wishes to determine the path which is farthest from the endangered nodes. The problem does not address the problem of handling attack directly but tries to minimize the impact of attack. The problem also finds its applications in sensor networks. In this paper, we present a simple and efficient algorithm to solve the problem. The algorithm converges in O(d2) time where d is the diameter of the network.

  • Reliable ad-hoc on-demand distance vector routing protocol
    Sandhya Khurana, Neelima Gupta and Nagender Aneja

    Proceedings of the International Conference on Networking, International Conference on Systems and International Conference on Mobile Communications and Learning Technologies,ICN/ICONS/MCL'06, Volume: 2006, Published: 2006 IEEE
    Mobile Ad hoc Networks’ (MANETs) properties present major vulnerabilities in security. The threats considered in MANETS are due to maliciousness that intentionally disrupt the network by using variety of attacks and due to selfishness of node which do not perform certain operations due to a wish to save power. In this paper, a co-operative security scheme called Reliable Ad hoc On-demand Distance Vector (RAODV) routing protocol based on local monitoring has been proposed to solve the problem of attack by malicious node as well as selfish behavior. RAODV behaves as AODV in the absence of attack and, detects and isolates misbehaving nodes in the presence of attack. Also it recovers from the attack when a misbehaving node leaves the network or becomes good.

RECENT SCHOLAR PUBLICATIONS

  • Transfer Learning using CNN for Handwritten Devanagari Character Recognition
    N Aneja, S Aneja
    arXiv preprint arXiv:1909.08774 2019

  • Semi-Supervised Learning for Cancer Detection of Lymph Node Metastases
    AK Jaiswal, I Panshin, D Shulkin, N Aneja, S Abramov
    arXiv preprint arXiv:1906.09587 2019

  • Method and system for ad-hoc social networking and profile matching
    N Aneja, S Gambhir
    US Patent 10,264,609 2019

  • IoT Device Fingerprint using Deep Learning
    S Aneja, N Aneja, MS Islam
    2018 IEEE International Conference on Internet of Things and Intelligence 2018

  • Options and challenges in Providing Universal Access
    M Aledhari, N Aneja, AK Bashir, R Bennett, J Bielby, HA Hicks, R Johnson, ...
    Retrieved from Internet Initiative IEEE: https://internetinitiative. ieee 2018

  • Profile-Based Ad Hoc Social Networking Using Wi-Fi Direct on the Top of Android
    N Aneja, S Gambhir
    Mobile Information Systems 2018 2018

  • Piecewise Maximal Similarity for Ad-hoc Social Networks
    S Gambhir, N Aneja, LC De Silva
    Wireless Personal Communications, 1-11 2017

  • Social Profile Aware AODV Protocol for Ad-Hoc Social Networks
    N Aneja, S Gambhir
    Wireless Personal Communications, 1-22 2017

  • Security and Privacy: Challenges and Defending Solutions for NoSQL Data Stores
    S Aneja, N Aneja
    NoSQLDatabase for Storage and Retrieval of Data in Cloud, 237-250 2017

  • Options and Challenges in Providing Universal Access
    P Mantri, M Aledhari, HA Hicks, R Nighot, N Aneja, S Mandal, AK Bashir, ...
    IEEE Internet Technology Policy Community White Paper 2017

  • Internet of Things (IoT) Security Best Practices
    GA Fink, M Aledhari, J Bielby, R Nighot, S Mandal, N Aneja, C Hrivnak, ...
    IEEE Internet Technology Policy Community White Paper 2017

  • Protecting Internet Traffic: Security Challenges and Solutions
    M Aledhari, S Mandal, N Aneja, M Dautrey, R Nighot, P Mantri, J Bielby
    IEEE Internet Technology Policy Community White Paper 2017

  • Options and Challenges in Providing Universal Access to the Internet: Part Two
    IJBI Helen Anne Hicks (North Park University), Rajesh Nighot (IEEE ...
    Newsletter 2017 2017

  • Middleware Architecture for Ad-hoc Social Network
    N Aneja, S Gambhir
    Research Journal of Applied Sciences, Engineering and Technology 13 (9), 690-695 2016

  • Software Design for Social Profile Matching Algorithm to create Ad-hoc social network on top of Android
    S Gambhir, S Mangla, N Aneja
    IEEE International Conference on Computing for Sustainable Global Development 2016

  • Anti-fraud computer implemented method for financial card transaction
    N Aneja, S Aneja
    US Patent App. 2015/0/339,657 2015

  • Geo-Social Semantic Profile Matching Algorithm for Dynamic Interests in Ad-hoc Social Network
    N Aneja, S Gambhir
    IEEE International Conference on Computational Intelligence & Communication 2015

  • Need of Ad-hoc social network based on Users’ Dynamic Interests
    S Gambhir, N Aneja, S Mangla
    IEEE International Conference on Soft Computing Techniques & Implementation 2015

  • Method and System for Profile Matching in Social Networks
    N Aneja, S Gambhir
    IN Patent App. 3105/DEL/2,015 2015

  • Geo-Social Profile Matching Algorithm for Dynamic Interests in Ad-Hoc Social Network
    N Aneja, S Gambhir
    Social Networking 3 (5), 240-247 2014

MOST CITED SCHOLAR PUBLICATIONS

  • Reliable ad-hoc on-demand distance vector routing protocol
    S Khurana, N Gupta, N Aneja
    IEEE International Conference on Networking, International Conference on 2006
    Citations: 38

  • Method and system for ad-hoc social networking and profile matching
    N Aneja, S Gambhir
    US Patent 10,264,609 2019
    Citations: 21

  • Minimum exposed path to the attack (MEPA) in mobile ad hoc network (MANET)
    S Khurana, N Gupta, N Aneja
    Sixth International Conference on Networking (ICN'07), 16-16 2007
    Citations: 11

  • Ad-hoc Social Network: A Comprehensive Survey
    S Gambhir, N Aneja
    International Journal of Scientific & Engineering Research 4 (8) 2013
    Citations: 10

  • Geo-Social Profile Matching Algorithm for Dynamic Interests in Ad-Hoc Social Network
    N Aneja, S Gambhir
    Social Networking 3 (5), 240-247 2014
    Citations: 9

  • Various Issues in Ad-hoc Social Networks
    N Aneja, S Gambhir
    National Conference on Recent Trends in Computer Science and Information 2012
    Citations: 8

  • Geo-Social Semantic Profile Matching Algorithm for Dynamic Interests in Ad-hoc Social Network
    N Aneja, S Gambhir
    IEEE International Conference on Computational Intelligence & Communication 2015
    Citations: 7

  • IoT Device Fingerprint using Deep Learning
    S Aneja, N Aneja, MS Islam
    2018 IEEE International Conference on Internet of Things and Intelligence 2018
    Citations: 4

  • Profile-Based Ad Hoc Social Networking Using Wi-Fi Direct on the Top of Android
    N Aneja, S Gambhir
    Mobile Information Systems 2018 2018
    Citations: 3

  • Piecewise Maximal Similarity for Ad-hoc Social Networks
    S Gambhir, N Aneja, LC De Silva
    Wireless Personal Communications, 1-11 2017
    Citations: 3

  • Need of Ad-hoc social network based on Users’ Dynamic Interests
    S Gambhir, N Aneja, S Mangla
    IEEE International Conference on Soft Computing Techniques & Implementation 2015
    Citations: 3

  • Transfer Learning using CNN for Handwritten Devanagari Character Recognition
    N Aneja, S Aneja
    arXiv preprint arXiv:1909.08774 2019
    Citations: 1

  • Options and challenges in Providing Universal Access
    M Aledhari, N Aneja, AK Bashir, R Bennett, J Bielby, HA Hicks, R Johnson, ...
    Retrieved from Internet Initiative IEEE: https://internetinitiative. ieee 2018
    Citations: 1

  • Middleware Architecture for Ad-hoc Social Network
    N Aneja, S Gambhir
    Research Journal of Applied Sciences, Engineering and Technology 13 (9), 690-695 2016
    Citations: 1

  • Software Design for Social Profile Matching Algorithm to create Ad-hoc social network on top of Android
    S Gambhir, S Mangla, N Aneja
    IEEE International Conference on Computing for Sustainable Global Development 2016
    Citations: 1

  • Verification of Forecasts from High Resolution Numerical Weather Prediction Model
    N Aneja, T George
    Research Journal of Applied Sciences, Engineering and Technology 8 (10 2014
    Citations: 1

Publications

Nagender Aneja and Sapna Gambhir, "Profile-based Ad hoc social networking using Wi-Fi direct on the top of android", Mobile Information Systems, 2018, Hindawi

Nagender Aneja and Sapna Gambhir, "Social Profile Aware AODV Protocol for Ad-Hoc Social Networks", Wireless Personal Communications, 2017, Springer

Sapna Gambhir, Nagender Aneja and Liyanage Chandratilake de Silva, "Piecewise Maximal Similarity for Ad-hoc Social Networks", Wireless Personal Communications, 2017, Springer New York LLC

RESEARCH OUTPUTS (PATENTS, SOFTWARE, PUBLICATIONS, PRODUCTS)

Patents
Nagender Aneja and Sapna Gambhir, "Method and System for Ad-Hoc Social Networking and Profile Matching" US 10,264,609 Granted Apr 16, 2019

Histopathologic Cancer Detection
I am participating in this Kaggle competition to create an algorithm to identify metastatic cancer in small image patches taken from larger digital pathology scans. The data for this competition is a slightly modified version of the PatchCamelyon (PCam) benchmark dataset. Currently, my rank is in the top 6% (65th from 1157).

ISIC 2018: Skin Lesion Analysis Towards Melanoma Detection, http://github.com/naneja/isic2018
The International Skin Imaging Collaboration (ISIC) is an international effort to improve melanoma diagnosis, sponsored by the International Society for Digital Imaging of the Skin (ISDIS). The ISIC Archive contains the largest publicly available collection of quality controlled dermoscopic images of skin lesions. The goal of this challenge is to help participants develop image analysis tools to enable the automated diagnosis of melanoma from dermoscopic images. This challenge is broken into three separate tasks: Task 1: Lesion Segmentation Task 2: Lesion Attribute Detection Task 3: Disease Classification. I participated in Task 3 after the challenge is over and was able to get around 70% accuracy.

Automated Plant Species Recognition, http://github.com/naneja/plants
We created a dataset of 740 images from 11 different plants species and the dataset was divided into 596 images for training the model and 144 images used for testing the model. I implemented transfer learning using Alexnet with PyTorch. The following hyperparameters Arch = ’alexnet’; Batch = 32; Hidden_units = 4096; Epochs = 200; Dropout = 0.5; Learning Rate = 0.01, Optimizer = SGD, Momentum = 0.9 provided best accuracy of 91.7%. We are looking for external funding to develop the project at International Level where we can have a database of medicinal plants from multiple countries.

Device Fingerprinting for Access Control, https://github.com/naneja/device-fingerprinting
Device Fingerprinting (DFP) is a technique to identify devices using Inter-Arrival Time (IAT) of packets and without using any other unique identifier. Our experiments include generating graphs of IATs of 100 and 1000 packets and using Convolutional Neural Network on the generated graphs
to identify a device. We implemented CNN on the IATs graphs for two datasets. The first data set was collected by us for two devices and another dataset is standard public dataset available at http://crawdad.org/gatech/fingerprinting/20140609. We achieved 86% accuracy in the first set and 95% accuracy in the second dataset. Initial results have been published at http://ieeexplore.ieee.org/abstract/document/8600824

STARTUP

Founder and Developer, http://ResearchID.co