Research

Dynamical systems and ergodic theory

Journal articles

1. 2023
   Slow entropy of some combinatorial constructions

   with P. Kunde and D. Wei

   Nonlinearity 36 (2023), no. 6, 2923–2974.

   Journal arXiv:2010.14472
2. 2023
   Slow entropy for some Anosov–Katok diffeomorphisms

   with P. Kunde and D. Wei

   Discrete and Continuous Dynamical Systems 43 (2023), 209–255.

   Journal arXiv:2109.08602
3. 2022
   Real-analytic realization of Uniform Circular Systems and some applications

   with P. Kunde

   Journal d’Analyse Mathématique 148 (2022), no. 2, 399–445.

   Journal arXiv:1705.05079
4. 2019
   Real-analytic AbC constructions on the torus

   with P. Kunde

   Ergodic Theory and Dynamical Systems 39 (2019), no. 10, 2643–2688.

   Journal arXiv:1611.06621
5. 2017
   Non-standard real-analytic realizations of some rotations of the circle

   Ergodic Theory and Dynamical Systems 37 (2017), no. 5, 1369–1386.

   Journal arXiv:1501.01531 Corrigendum

Preprints and notes

1. 2024
   Weak mixing behavior for the projectivized derivative extension

   with D. Khurana and P. Kunde

   Preprint.

   arXiv:2412.21041
2. 2017
   Real-analytic diffeomorphisms with homogeneous spectrum and disjointness of convolutions

   with P. Kunde

   Preprint.

   arXiv:1711.02537
3. Notes
   Information and entropy in dynamical systems: an introduction

   with F. Ledrappier

   Unpublished notes of the Będlewo lectures delivered by F. Ledrappier.

   PDF

Machine learning

Conference proceedings

1. 2022
   SAU: Smooth activation function using convolution with approximate identities

   K. Biswas, S. Kumar, S. Banerjee, A. Pandey

   Computer Vision – ECCV 2022, Lecture Notes in Computer Science 13681, Springer, Cham.

   Proceedings arXiv:2109.13210
2. 2022
   Smooth Maximum Unit: Smooth activation function for deep networks using smoothing maximum technique

   K. Biswas, S. Kumar, S. Banerjee, A. Pandey

   Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, 784–793.

   CVF open access IEEE arXiv:2111.04682
3. 2022
   ErfAct and Pserf: Non-monotonic smooth trainable activation functions

   K. Biswas, S. Kumar, S. Banerjee, A. Pandey

   Proceedings of the AAAI Conference on Artificial Intelligence 36 (2022), no. 6, 6097–6105.

   Proceedings arXiv:2109.04386
4. 2021
   EIS: Efficient and trainable activation functions for better accuracy and performance

   K. Biswas, S. Kumar, S. Banerjee, A. Pandey

   Artificial Neural Networks and Machine Learning – ICANN 2021, Lecture Notes in Computer Science 12892, Springer, Cham.

   Proceedings arXiv:2009.13501

Journal articles

1. 2021
   TanhSoft: Dynamic trainable activation functions for faster learning and better performance

   K. Biswas, S. Kumar, S. Banerjee, A. Pandey

   IEEE Access 9 (2021), 120613–120623.

   Journal arXiv:2009.03863

Preprints

1. 2021
   Orthogonal-Padé activation functions: Trainable activation functions for smooth and faster convergence in deep networks

   K. Biswas, S. Banerjee, A. K. Pandey

   Preprint.

   arXiv:2106.09693

Other areas

Journal articles

1. 2024
   Ergodic and mixing quantum channels: From two-qubit to many-body quantum systems

   S. Aravinda, S. Banerjee, R. Modak

   Physical Review A 110 (2024), no. 4, 042607.

   Journal arXiv:2310.02740
2. 2019
   Revisiting the use of squared magnitude function for the optimal approximation of (1+α)-order Butterworth filter

   S. Mahata, S. Banerjee, R. Kar, D. Mandal

   AEU – International Journal of Electronics and Communications 110 (2019).

   Journal

Talks and mini-courses