Chuan Sun is a VP Data Scientist in the Digital Intelligence team at JPMorgan Chase. He architected, designed, and implemented an in-house machine learning platform to build predictive models to drive user growth/engagement for ChasePay and QuickPay. Previously, he was a SDE in Amazon Go, working on challenging Computer Vision problems. He received Ph.D. in Computer Science from University of Central Florida in 2014. His research topics involved classifying complex human actions from video via representation learning and tensor approximation. Chuan is interested in uncovering the relationship of things and seeking order from chaos.
"Unity of Opposites" for Machine Learning in Financial Services
by VP Data Scientist, JPMorgan Chase & Co.
In Philosophy, the Unity of Opposites (UOO) defines a situation in which the existence of a situation depends on the co-existence of at least two conditions which are opposite to each other, yet dependent on each other and presupposing each other, within a field of tension (Wikipedia). As a machine learning practitioner and data scientist, I gradually realize that, surprisingly, this UOO is well applicable in a wide range of important ML concepts, in which two opposite conditions co-exist. For example, the model fitting may cause underfitting or overfitting; generalization error contains bias and variance, model selection involves exploration and exploitation, etc. Furthermore, when digging deeper, there are even more key concepts possessing this philosophical trait of "unity of opposites". Understanding those opposites is indispensable to obtain a holistic perspective and deep understanding of ML. In providing practical examples of UOO, I will further decompose ML projects in financial services into "5 Pillars", namely, boundary, data, people, algorithm, and platform. I will show, under UOO, how these 5 pillars achieve the philosophical oneness (and twoness) and help to drive ML decisions in financial services.