Khardon’s proposal, “Effective Decision Making in Large Action Spaces,” earned a $50,000 grant for promoting the understanding and use of data science in the area of marketing. The Data Science Research Award program encourages both theoretical and empirical development of solutions to problems in customer experience and marketing. Khardon models decision problems and machine learning problems using probabilistic models and develops algorithms for them using methods from approximate Bayesian inference.
“My work is relevant for optimizing customer experience in online systems, which is a sequential process where the system can potentially vary multiple aspects at each step and has a huge number of possible ‘actions’.” Khardon said. “This applies in many applications such as web-site configuration, journey optimization, or advertising. My work in this project will explore reinforcement learning scenarios where a model of the environment is not given.”
Khardon hopes to develop algorithms that learn a model and optimize the actions simultaneously. His proposed algorithms make use of the connections between planning and inference in probabilistic models.
“The power of data makes an impact on such a variety of areas, and our data science program at SICE is leading the way in finding innovative ways to analyze the flood of data,” said Raj Acharya, dean of SICE. “Roni’s recognition by Adobe is another signal that some of the top firms in technology understand the value of our research, and his work is richly deserving of this award.”