Vishwa Goudar

This study investigates learning-to-learn in a neural network model, showing schema formation in a low-dimensional subspace that accelerates learning across similar problems.

This paper trains biologically plausible RNNs on cognitive tasks with long-term dependencies using specialized skip-connections, achieving faster training and better performance than conventional methods.

This paper shows how recurrent neural networks can encode and transcribe spoken digits by generating time-invariant neural trajectories that generalize across spatial and temporal variations.