Key publications and related code


Interfacing adaptive control theory, machine learning and neuroscience:

A. Gilra and W. Gerstner. ‘Predicting non-linear dynamics by stable local learning in a recurrent spiking neural network’. eLife, 6:e28295, 2017. Recommended by Faculty of 1000 Access the recommendation on F1000Prime.

Code at: https://github.com/adityagilra/FOLLOW.

 

A. Gilra and W. Gerstner. ‘Non-linear motor control by local learning in spiking neural networks’. Proceedings of the 35th International Conference on Machine Learning, PMLR 80:1768-1777, 2018. See also my ‘long talk’ at ICML 2018.

Code at: https://github.com/adityagilra/FOLLOWControl.

  

C. Klos, Y. F. K. Kossio, S. Goedeke, A. Gilra, and R.-M. Memmesheimer. ‘Dynamical Learning of Dynamics’. arXiv:1902.02875 [q-Bio], 2019.

  

Learning cognitive tasks without interference and forgetting:

R. V. Rikhye, A. Gilra & M. M. Halassa. ‘Thalamic regulation of switching between cortical representations enables cognitive flexibility’. Nature Neuroscience, volume 21, pages 1753–1763, 2018.

Code at: https://github.com/adityagilra/PFC_MD_weights_stability.

  

Memory-augmented neural networks for reinforcement learning:

M. Martinolli, W. Gerstner and A. Gilra, ‘Multi-timescale memory dynamics in a reinforcement learning network with attention-gated memory’. Front. Comput. Neurosci. | doi: 10.3389/fncom.2018.00050.

Code at: https://github.com/martin592/hybrid_AuGMEnT. This is part of a larger project archibrain on github, to which Marco Martinolli and Vineet Jain contributed, to synthesize biologically plausible neural architectures that perform complex cognitive tasks.

 

Computation by interneurons in the first sensory area (olfactory bulb) in the brain, of the sense of smell:

A. Gilra and U. S. Bhalla. ‘Bulbar microcircuit model predicts connectivity and roles of interneurons in odor coding’. PLOS ONE 10 (5): e0098045, 2015.

Code at: https://github.com/adityagilra/olfactory-bulb.

  

See my Google Scholar page for a complete list.

See also my code for general-purpose neural simulation.