In this work Dr Priya Ranjan demonstrated the role of numerical R/Matlab based optimization techniques in Deep Learning using some simple examples. Under this renewed context of deep learning,  basic notion of Gradient Descent algorithm was shown to be of interest in the framework of largely smooth optimization for both constrained and unconstrained cases.  Further, role of convexity in optimization was highlighted and  noisy version of Gradient Descent was illustrated towards minimizing the error function which appears naturally in the course of Deep learning. Primitive computational techniques and algorithmic implementations were also hinted at. Some hands on examples regarding these optimization schemes and their application to deep learning were demonstrated in action.