Research


Focus

Our brain is the most sophisticated machine on earth, when it comes to recognize things. We don't even notice how amazingly fast our brain processes information to make sense of what we see. We need only couple of samples of a new object to recognize it later on. Recent development in machine learning (deep learning, convolutional neural networks (CNNs), to be more specific) shows that the hierarchical models (inspired by our brain) are also very good at object recognition. My research is in the interface between the two: machine learning and the brain. I study to uncover the longstanding research question: what makes CNNs to be able to capture aspects of the cortical processing surprisingly well? Finding answer to this question will help us to understand how our brain solves vision.



Publications/Talks/Workshops

Deep neural networks capture texture sensitivity in V2

Md Nasir Uddin Laskar, Luis G Sanchez Giraldo, and Odelia Schwartz, in Journal of Vision, vol 20, pp 1-21, 2020.

Normalization and pooling in hierarchical models of natural images

Luis G Sanchez Giraldo, Md Nasir Uddin Laskar, and Odelia Schwartz, in Current Openion in Neurobiology: Machine Learning, Big Data, and Neuroscience, impact factor 6.541, vol 55, pp 65-72, 2019.

Correspondence of Deep Neural Networks and the Brain for Visual Textures

Md Nasir Uddin Laskar, Luis G Sanchez Giraldo, and Odelia Schwartz, in ArXiv preprint, 2018.

Texture Selectivity: Comparing Deep Neural Networks and the Brain

Md Nasir Uddin Laskar, Luis G Sanchez Giraldo, and Odelia Schwartz, in Mutual Benefits of Cognitive and Computer Vision (MBCC2017w) at ICCV 2017, Venice (oral presentation).

Deep Learning Captures V2 Selectivity for Natural Textures

Md Nasir Uddin Laskar, Luis G Sanchez Giraldo, and Odelia Schwartz, in Computational and System Neuroscience (COSYNE), 2017, SLC.

Deep Learning Captures V2 Selectivity for Natural Textures

Md Nasir Uddin Laskar, Luis G Sanchez Giraldo, and Odelia Schwartz, in Brains and Bits: Neuroscience Meets Machine Learning at NIPS 2016, Barcelona.

Offsetting Obstacles of any Shape for Robot Motion Planning

Md Nasir Uddin Laskar, H. H. Viet, S. Y. Choi, S. Y. Lee, and TaeChoong Chung, Robotica, vol. 33, issue 04, pp. 865-883, Cambridge University Press, 2015.

BA*: An Online Complete Coverage Algorithm for Cleaning Robots

H. H. Viet, Viet H. D., Md Nasir Uddin Laskar, and TaeChoong Chung, Applied Intelligence, vol. 39, issue 2, pp. 217-235, Springer, 2013.

Geometric Modeling of any Obstacle Shapes for Robot Motion Planning

Md Nasir Uddin Laskar, Seung Y. Choi, and TaeChoong Chung, Int. Conf. on Intelligent Robotics and Applications (ICIRA), Springer LNCS, vol. 8102, pp. 693-705, 2013.



Theses

Understanding Texture Sensitivity in the Brain Visual System with Deep Neural Networks

PhD in Computer Science, University of Miami, 2020. Advisor: Professor Odelia Schwartz

Modeling Obstacles of any Shape for Robot Motion Planning

Masters in Computer Engineering, Kyung Hee University, Korea, 2013. Advisor: Professor TaeChoong Chung