Dr. Nagender Aneja

nagender.aneja@ubd.edu.bn

Researcher, Institute of Applied Data Analytics and Senior Manager, Innovation and Enterprise Office

               

                              

I am a Deep Learning and Patent Professional and have a passion for the applications in Medical Imaging. I participated in Histopathologic Cancer Detection, https://www.kaggle.com/naneja [achieved the rank of 65th among 1157 (top 6%)].

I also participated in ISIC 2018: Skin Lesion Analysis Towards Melanoma Detection, https://github.com/naneja/isic2018 and developed a prototype for Automated Plant Species Identification, https://github.com/naneja/plants and implemented Device Fingerprinting to identify network devices using deep learning https://github.com/naneja/device-fingerprinting I also did a project to apply CNN-LSTM for Image Captioning.

I developed Autobot that does web scrapping from online public Patents Databases to create the desired Prior Art Search Report and developed Expert Directory (http://expert.ubd.edu.bn) for Universiti Brunei Darussalam where faculty members can self-register and update their individual profile. The system has been designed to fetch the list of publications and citation information automatically so that profile is automatically updated without human intervention https://expert.ubd.edu.bn/nagender.aneja The expert system is also available at https://researchid.co for public use freely.

I have also expertise in writing patent claims and responding to the patent examiner on the patent objections for US Patent Applications.

EDUCATION

Ph.D. Computer Engineering
M.E. Computer Technology and Applications
Master of Management Studies
M.Sc. Mathematics
B.Sc. Physics, Chemistry, Mathematics

RESEARCH INTERESTS

Deep Learning, Computer Vision
CNN, RNN, LSTM
Applications of Computer Vision in Medical Imaging

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
for 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
for 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
for Remote Research Collaboration
11

Scopus Publications

85

Google Scholar Citations

6

Google Scholar h-index

2

Google Scholar i10-index

Scopus Publications

TOP PUBLICATIONS

1. Amit Kumar Jaiswal, Ivan Panshin, Dimitrij Shulkin, Nagender Aneja, and Samuel Abramov, "Semi-Supervised Learning for Cancer Detection of Lymph Node Metastases," https://s1155026040.github.io/mvd-2019-cvpr-workshop/

2. Nagender Aneja, Sapna Gambhir, "Profile-Based Ad Hoc Social Networking Using Wi-Fi Direct on the Top of Android", Mobile Information Systems, 2018, http://hindawi.com/journals/misy/2018/9469536/ (SCIE Indexed)

3. Nagender Aneja, Sapna Gambhir, "Social Profile Aware AODV Protocol for Ad-hoc Social Networks", Wireless Personal Communications, 2017, http://link.springer.com/article/10.1007/s11277-017-4718-x (SCIE Indexed)

4. Sapna Gambhir, Nagender Aneja, Liyanage Chandratilake De Silva "Piecewise Maximal Similarity for Ad-hoc Social Networks", Wireless Personal Communications, 2017, http://link.springer.com/article/10.1007/s11277-017-4683-4 (SCIE Indexed)

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, http://kaggle.com/naneja
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