Dr. Shabbeer Basha

Associate Professor

shabbeerb@rvu.edu.in

About
Publication & Works
Research Summary
Awards & Achievements

Shabbeer has an industry experience of 3 years in computer vision, machine learning and related areas. He started his career after M.Tech. as an Assistant Professor in MITS Madanapalle and served four years. During this period, he taught core computer science courses like Data Structures, Design and Analysis of Algorithms, Theory of Computation, Optimization Techniques, Compiler Design, etc.,. After which, he joined Indian Institute of Information Technology Sri City to pursue his Ph.D. He worked on “Neural Architecture Learning for Image Classification” problem under the supervision of Prof. Viswanath. As part of his dissertation, he published 4 peer reviewed journals (Neural Networks, Neurocomputing, Neural Computing and Applications, Multimedia Tools, and Applications) and one conference (ICARCV). After his Ph.D., he joined PathPartner Technology Pvt Ltd, Bangalore as a lead engineer. Later, he joined Lytx Inc as Senior Machine Learning Scientist. During his industry journey, he worked on building low compute deep learning models, neural network compression, multitask learning, active learning.

“If we knew what we were doing, it would not be called research, would it?” - Albert Einstein

1. Journals:

AdaInject: Injection based adaptive gradient descent optimizers for convolutional neural networks, Shiv Ram Dubey, SH Shabbeer Basha, Satish Kumar Singh, and Bidyut Baran Chaudhuri, IEEE Transactions on Artificial Intelligence.

An Information-rich Sampling Technique over Spatio-Temporal CNN for Classification of Human Actions in Videos, Shabbeer Basha, Viswanath Pulabaigari, Snehasis Mukherjee, Multimedia Tools & Applications (2.75), March 2022.

HRel: Filter Pruning based on high relevance between activation maps and class labels, Sarvani CH, Mrinmoy Ghorai, Shiv Ram Dubey, Shabbeer Basha, Neural Networks (8.05), Jan 2022, (Elsevier).

AutoFCL: Automatically Tuning Fully Connected Layers for Hadnling Small Dataset. Shabbeer Basha, Sravan Kumar Vinakota, Shiv Ram Dubey, Viswanath Pulabaigari and Snehasis Mukherjee. Neural Computing and Applications (IF:4.77), Nov 2020. (Springer).

AutoTune: Automatically Tuning Convolutional Neural Networks for Improved Transfer Learning. SH Shabbeer Basha, Sravan Kumar Vinakota, Viswanath Pulabaigari, Snehasis Mukherjee and Shiv Ram Dubey. Neural Networks (IF:5.53), 133:112-122, Jan 2021. (Elsevier).

Impact of Fully Connected Layers on Performance of Convolutional Neural Networks for Image Classification. SH Shabbeer Basha, Shiv Ram Dubey, Viswanath Pulabaigari and Snehasis Mukherjee. Neurocomputing (IF: 4.43), 378:112-119, Feb 2020. (Elsevier).

2. Conferences:

A Simple Hybrid Filter Pruning for Efficient Edge Inference, SH Shabbeer Basha, Sheethal N Gowda, Jayachandra Dakala, In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 3398-3402. IEEE, 2022.

RCCNet: An Efficient Convolutional Neural Network for Histological Routine Colon Cancer Nuclei Classification. SH Shabbeer Basha, Soumen Ghosh, Kancharagunta Kishan Babu, Shiv Ram Dubey, Viswanath Pulabaigari and Snehasis Mukherjee. Fifteenth International Conference on Control, Automation, Robotics and Vision (ICARCV), Singapore, Nov 2018. (IEEE).

3. Patents:

A System and Method for Building Efficient Neural Networks

A System and Method for Deep Active Learning (both are under review)

Deep Neural Networks are used in most of the applications due to their implicit power of learning features and decision making. My research work focuses on building cost effective methods that enable us to use it for any application and improving generalization ability of these networks. My research interests include: Active Learning, Neural Network Compression, Neural Architecture Search, Multitask learning, Domain Adaptation.

AutoTune paper selected for VISION INDIA, ICVGIP-2021, IIT Kharagpur.
Qualified in APSET-2016.