-
About Ikjyot Singh Kohli
Mathematical Physicist with a specialty in Dynamical Systems and General Relativity. Through my interests in dynamical systems theory, I became deeply interested in neural networks which led to me pursuing AI and related applications on a full-time basis. I am interested from a theoretical perspective in the dynamics of Continuous-Time Recurrent Neural Networks (CTRNN), particularly with respect to their geometry and potentials for quantization.
I have been applying both AI and more “classic” ML methods in industry solving problems pertaining to audience and demographic predictions, customer behaviour and personalization and financial modelling as well. I have applied these methods at Cineplex where I have been a lead/senior data scientist for the past number of years utilizing Microsoft’s CNTK, Keras, TensorFlow, R, and Python. I have also made substantial use of Mathematica and MATLAB for problems that are more mathematical in nature.
As part of this work at Cineplex, several projects were put into production with three major accomplishments:
1. A state-of-the-art AI-Based Attendance and Audience Projections Model for Cineplex Media that was output to a PowerBI dashboard which is used by the whole Media business and generates Millions of Dollars in Revenue for the company. (Average quarterly revenue of $30 M).
2. A stochastic attendance model that predicts hourly attendance for each theatre location that is used by theatre managers for attendance forecasting and workforce management that saves hundreds of man hours of workforce scheduling.
In addition to my work in industry, I have also been a professor and course developer for McMaster University’s Big Data Analytics Program, and more recently at York University’s Schulich School of Business teaching and implementing courses for the MBAN and MMAI programs. I have specifically taught AI1, Data Science 1 and 2, and Numerical Methods for AI.