Biomimetic Computer Vision Coprocessor

This work relates to unconventional problem mapping harnessing the innate power of physical systems. In particular, we  solve Perceptual Grouping which is a component of Pattern Recognition.  Perceptual grouping relates to human vision system's ability to identify salient features and relationship without any prior knowledge about the image. 

Biomimetic way of solving this problem relates to  energy minimization problem, which has origins in the early days of Gestalt psychology, a branch of human perception psychology. How do we perceive an object? It turns out objects in the world tend to exhibit high level of symmetry, convexity, parallelism. We solve the grouping problem using a system of nano-magnets. The Hamiltonian of a collection of bipolar nano magnets is governed by the pairwise dipolar interactions. The ground state of a nanomagnet collection minimizes this Hamiltonian. We have devised a computational method, based on multi-dimensional scaling, to decide upon the spatial arrangement of nanomagnets that matches a particular quadratic minimization problem. Each variable is represented by a nanomagnet and the distances between them are such that the dipolar interactions match the corresponding pairwise energy term in the original optimization problem.

1. Sarkar, Sudeep, and Sanjukta Bhanja. "Direct quadratic minimization using magnetic field-based computing." Design and Test of Nano Devices, Circuits and Systems, 2008 IEEE International Workshop on. IEEE, 2008.
2. Rajaram, S., Karunaratne, D. K., Sarkar, S., & Bhanja, S. (2013). Study of dipolar neighbor interaction on magnetization states of nano-magnetic disks.Magnetics, IEEE Transactions on, 49(7), 3129-3132.
3. Panchumarthy, R., Karunaratne, D. K., Sarkar, S., & Bhanja, S. (2013). Magnetic state estimator to characterize the magnetic states of nano-magnetic disks. Magnetics, IEEE Transactions on, 49(7), 3545-3548.
4. Kumari, A., Sarkar, S., Pulecio, J. F., Karunaratne, D. K., & Bhanja, S. (2011). Study of magnetization state transition in closely spaced nanomagnet two-dimensional array for computation. Journal of Applied Physics, 109(7), 07E513
5. Pulecio, J., Bhanja, S., & Sarkar, S. (2011, August). An experimental demonstration of the viability of energy minimizing computing using nano-magnets. In Nanotechnology (IEEE-NANO), 2011 11th IEEE Conference on(pp. 1038-1042). IEEE.
6. Panchumarthy, R., Karunaratne, D. K., Sarkar, S., & Bhanja, S. (2011, August). Tool for analysis and quantification of fabrication layouts in nanomagnet-based computing. In Nanotechnology (IEEE-NANO), 2011 11th IEEE Conference on (pp. 111-115). IEEE.