Dr. Chandranath Adak

Assistant Professor

Centre for Data Science, JISIASR

 

email: chandra@jisiasr.org, adak32@gmail.com

Google Scholar: https://scholar.google.co.in/citations?user=f49iYVQAAAAJ&hl=en

DBLP: https://dblp.uni-trier.de/pers/a/Adak:Chandranath.html

 

Postdoctoral Researcher: Computational Intelligence and Brain Computer Interface (CIBCI) Centre, University of Technology Sydney, Australia.

PhD: University of Technology Sydney, Australia.

 

Professional Engagements:

Jul. 2019 – Present: Assistant Professor, JIS Institute of Advanced Studies and Research, India.

Jul. 2017 – Sep. 2017: Visiting Researcher, DINFO, University of Florence, Italy.

Jun. 2014 – Mar. 2015: Research Project-linked Person, CVPR Unit, Indian Statistical Institute, India.

 

Research Topics:

Deep Learning; Reinforcement Learning; Image & Video Understanding; Medical Image/Video Analysis, Computer Forensics & Biometrics, Service Recommendation.

 

Research Focus:

 

  1. Forensics and Biometrics:

Forensics and biometrics studies are very pertinent research topics in today’s world, especially when the crime increases with a massive rate. Biometrics are also important to keep track of the identity of people, which ensures to prevent the fake identity.

We mostly focus on handwriting, eye, and face. The handwriting analysis comes under behavioural biometrics, whereas the eye and face image analysis are physical biometrics. Some of our researches in this domain include writer identification/verification, handwriting stroke analysis towards legibility and aesthetics, understanding writing difficulty and idiosyncrasy, sclera and iris segmentation/recognition, face identification/verification, liveness detection etc. Different cutting-edge machine learning technologies, innovative techniques are used for our research on forensics and biometrics.

 

  1. Video Surveillance and Understanding:

Video surveillance is a pivotal technology having benefits in making social life better. The increasing rate of crime is a crucial social problem. Video surveillance can trace the crime and provide proper evidence, as a result, the crime may be reduced drastically. Besides, for smart traffic systems, beach/water surveillance systems, such technology is also essential.

The state-of-the-art system works well on a less crowded scenario. However, it fails badly for a populated country such as India, where the roads are mostly densely crowded. In our research, we focus on a highly dense, unstructured crowd. Our research emphasises on such dense Indian traffic video for scene segmentation, object localization, detection and tracking, and hazard prediction automatically in real-time. Here, automation is required since manual analysis is usually error-prone owing to the short human-attention span.

We use U-Net++ for scene segmentation, Mask-RCNN for object localization, an attention-based convolutional neural network for object detection, Siamese architecture-based frame matching for object tracking, and deep regression for hazard prediction. We also plan to build a deep semantic architecture to understand the video by analysing various objects inside the video.

 

  1. Medical Image and Video Analysis:

Automated medical image analysis is one of the biggest concerns in today’s world due to the increasing number of digital medical imaging. Besides, the medical video analysis is a booming area owing to a growing amount of machine-assisted laparoscopic surgery.

We focus on digital X-rays, Computer Tomography (CT) scans to analyse, and diagnose various diseases. Very recently, we have proposed a Denser Network which inspects the COVID-19-affected patients’ chest X-rays and thorax CT scans.

Often, the laparoscopic videos are distorted due to various real-time noises. We work on dealing with such noise detection and removal using deep reinforcement learning. Such an area of research is an original attempt of its kind.

 

  1. Service Recommendation

With the proliferation of Internet-of-Things, Machine to Machine communication, and smart technologies, web services are becoming pervasive. The massive growth of the number of functionally equivalent services aiding various usages (e.g., online shopping, online payment, online car/hotel booking) in modern days introduces a challenge to the service recommendation research. One of the prime aspects influencing the service recommendation is the Quality-of-Service (QoS) parameter, which depicts the performance of a web service. In general, the service provider furnishes the value of the QoS parameters during service deployment. However, in reality, the QoS values of service vary across different users, time, locations, etc. Therefore, estimating the QoS value of service before its execution is an important task, and thus the QoS prediction has gained significant research attention. Multiple approaches are available in the literature for predicting service QoS. However, these approaches are yet to reach the desired accuracy level. Three major aspects to take into account while designing a prediction algorithm are: (a) accuracy, (b) robustness (i.e., prediction time), (c) scalability (i.e., the prediction algorithm is supposed to handle large dataset). This research aims to design novel prediction algorithms leveraging different machine learning approaches (e.g., collaborative filtering, matrix factorization, regression), while satisfying the above three criteria.

Our initial research on QoS prediction focuses on the problem across different users. Our proposal includes two key technologies: (a) hybrid filtering and (b) hierarchical prediction mechanism. On the one hand, the hybrid filtering method aims to obtain a set of similar users and services, given a target user and a service. On the other hand, the goal of the hierarchical prediction mechanism is to estimate the QoS value accurately by leveraging hierarchical neural-regression.

We further intend to extend our work on a multi-dimensional context-aware QoS prediction problem, which helps in real-time service provisioning.

 

Publications:

 

Journals:

  1. Adak, B. B. Chaudhuri, C. T. Lin, M. Blumenstein, “Intra-Variable Handwriting Inspection Reinforced with Idiosyncrasy Analysis”, IEEE Trans. on Information Forensics & Security, vol. 15, pp. 3567-3579, 2020.
  2. Adak, B. B. Chaudhuri, M. Blumenstein, “An Empirical Study on Writer Identification and Verification from Intra-variable Individual Handwriting”, IEEE Access, vol. 7, no. 1, pp. 24738-24758, 2019.
  3. B. Chaudhuri*, C. Adak*, “An Approach for Detecting and Cleaning of Struck-out Handwritten Text”, Pattern Recognition, vol. 61, pp. 282-294, January 2017. *Joint first author.

 

Book Chapter:

  1. Adak, D. Ghosh, R. R. Chowdhury, S. Chattopadhyay, “COVID-19-affected Medical Image Analysis using Denser Net”, in Book Chapter of Data Science for COVID-19, Eds. U. Kose, D. Gupta, V.H.C. de Albuquerque, A. Khanna, Elsevier, ISBN: 9780128245361, 2020. (in print)

 

Conferences:

  1. Adak, B. B. Chaudhuri, C. T. Lin, M. Blumenstein, “Why Not? Tell us the Reason for Writer Dissimilarity”, 33rd Int. Joint Conference on Neural Networks (IJCNN), pp. 1-7, Glasgow, Scotland, UK, 19-24 July, 2020.
  2. Adak, B. B. Chaudhuri, C. T. Lin, M. Blumenstein, “Detecting Named Entities in Unstructured Bengali Manuscript Images”, 15th Int. Conf. on Document Analysis and Recognition (ICDAR), pp. 196-201, Sydney, Australia, 20-25 Sep., 2019.
  3. Adak, B. B. Chaudhuri, M. Blumenstein, “A Study on Idiosyncratic Handwriting with Impact on Writer Identification”, 16th Int. Conf. on Frontiers in Handwriting Recognition (ICFHR), pp. 193-198, Niagara Falls, USA, 5-8 Aug., 2018.
  4. Adak, B. B. Chaudhuri, M. Blumenstein, “Cognitive Analysis for Reading and Writing of Bengali Conjuncts”, 31st Int. Joint Conference on Neural Networks (IJCNN), pp. 1-7, Rio de Janeiro, Brazil, 8-13 July, 2018.
  5. Adak, S. Marinai, B. B. Chaudhuri, M. Blumenstein, “Offline Bengali Writer Verification by PDF-CNN and Siamese Net”, 13th IAPR Int. Workshop on Document Analysis Systems (DAS), pp. 381-386, Vienna, Austria, 24-27 Apr., 2018.
  6. Adak, B. B. Chaudhuri, M. Blumenstein, “Legibility and Aesthetic Analysis of Handwriting”, 14th Int. Conf. on Document Analysis and Recognition (ICDAR), pp. 175-182, Kyoto, Japan, 9-15 Nov., 2017.
  7. Adak, B. B. Chaudhuri, M. Blumenstein, “Impact of Struck-out Text on Writer Identification”, 30th Int. Joint Conference on Neural Networks (IJCNN), pp. 1465-1471, Anchorage, Alaska, USA, 14-19 May, 2017.
  8. Adak, B. B. Chaudhuri, M. Blumenstein, “Writer Identification by Training on One Script but Testing on Another”, 23rd Int. Conference on Pattern Recognition (ICPR), pp. 1148-1153, Cancun, Mexico, 4-8 Dec., 2016.
  9. Adak, B. B. Chaudhuri, M. Blumenstein, “Offline Cursive Bengali Word Recognition using CNNs with a Recurrent Model”, 15th Int. Conf. on Frontiers in Handwriting Recognition (ICFHR), pp. 429-434, Shenzhen, China, 23-26 Oct., 2016.
  10. Adak, B. B. Chaudhuri, M. Blumenstein, “Named Entity Recognition from Unstructured Handwritten Document Images”, 12th IAPR Int. Workshop on Document Analysis Systems (DAS), pp. 375-380, Santorini, Greece, 11-14 Apr., 2016.
  11. Adak, P. Maitra, B. B. Chaudhuri, M. Blumenstein, “Binarization of Old Halftone Text Documents”, IEEE TENCON, IEEE Conf. # 35439, pp. 1-5, Macau, China, 1-4 Nov., 2015.
  12. Adak, B. B. Chaudhuri, “Writer Identification from Offline Isolated Bangla Characters and Numerals”, 13th Int. Conf. on Document Analysis and Recognition (ICDAR), pp. 486-490, Nancy, France, 23-26 Aug., 2015.
  13. Adak, B. B. Chaudhuri, “An Approach of Strike-through Text Identification from Handwritten Documents”, 14th Int. Conf. on Frontiers in Handwriting Recognition (ICFHR), pp. 643-648, Crete Island, Greece, 1-4 Sep., 2014.

 

Achievements:

 

Grants:

  1. Centre for AI and School of Software, FEIT, Publication Grant, University of Technology Sydney, Australia, 2019.
  2. School of Software, FEIT, Higher level of Funding Support; and Centre for AI, Conference Travel Grant, University of Technology Sydney, Australia, 2018.
  3. Vice-Chancellor’s Postgraduate Research Student Conference Fund, University of Technology Sydney, Australia, 2018.
  4. IEEE CIS (Computational Intelligence Society) Graduate Student Research Grants, 2017.
  5. GGRS-IEIS (Griffith Graduate Research School and International Experience Incentive Scheme) Conference Travel Grants, Australia, 2017.
  6. School of ICT (Information and Communication Technology), Griffith University, Conference Funding, Australia, 2017.
  7. IIIS (Institute for Integrated and Intelligent Systems), Griffith University, Conference Travel Grant, Australia, 2016.
  8. Gold Coast Association of Postgraduates (GCAP) Professional Development and Conference (PDAC) Funding, Australia, 2015.

 

Competitions:

  1. Winner of ICDAR 2019 Competition on Recognition of Early Indian Printed Documents – REID2019 (conjunction with IAPR International Conference on Document Analysis and Recognition (ICDAR-2019)), Australia, Sep. 2019.
  2. Winner of the signature and name component recognition task of ICFHR-2018 Competition on Thai Student Signature and Name Component Recognition and Verification – TSNCRV2018 (conjunction with IAPR International Conference on Frontiers in Handwriting Recognition (ICFHR-2018)), USA, Aug. 2018.
  3. 1st runner-up in ICFHR 2018 Competition on Handwritten Document Image Binarization (H-DIBCO 2018), USA, Aug. 2018.
  4. 1st runner-up in ICDAR 2017 Competition on Recognition of Early Indian Printed Documents – REID2017 (conjunction with IAPR International Conference on Document Analysis and Recognition (ICDAR-2017)), Japan, Nov. 2017.
  5. Entrepreneur’s Award and Best of Gold Coast Award, GovHack-2016, Australia.
  6. Top-10 finalist in ISTAS (Invitational Symposium on Trusted Autonomous Systems) Student Poster Competition, organized by Defence Science and Technology Group, Dept. of Defence, Australian Government, 2016.
  7. Winner of the recognition task, and 1st runner-up in segmentation task of ICB-2016 Sclera Segmentation and Recognition Benchmarking Competition: SSRBC-2016 (conjunction with IAPR International Conference on Biometrics (ICB-2016)), Sweden, Jun. 2016.

 

Scholarships:

  1. HDR Post-Thesis Industry Scholarship, FEIT, University of Technology Sydney, Australia, 2019.
  2. Kalam Doctoral Scholarship and International Research Scholarship (IRS), University of Technology Sydney, Australia, 2018.
  3. Griffith University International Postgraduate Research Scholarship (GUIPRS) and Postgraduate Research Scholarship (GUPRS), Australia, 2015-2018.
  4. National Scholarship for qualifying in Graduate Aptitude Test in Engineering (GATE-2012 in CS), India, 2012-2014.
  5. National Merit Scholarship from MHRD (Ministry of Human Resource Development), India, based on Higher-Secondary level exam, 2008-2012.
  6. National Merit Scholarship from MHRD (Ministry of Human Resource Development), India, based on Secondary level exam, 2006-2008.

 

 

For more details, please follow this link: https://sites.google.com/site/chandranathadak1989