Government
Valeria Luiselli on Sound, Memory, and New Beginnings
Sign up to receive it in your inbox. Your story in this week's issue, " Predictions and Presentiments," is drawn from your forthcoming book, " Beginning Middle End," which is coming out in July. The audio version will incorporate sounds that you and your team recorded in Sicily, where both the piece and the novel are set. How would you compare the creative processes of writing and recording, and the experiences of reading and listening? Recording sound and listening attentively have been an integral part of my writing process for a long time now.
Off-PolicyEvaluationforAction-Dependent Non-StationaryEnvironments
Methods for sequential decision making are often built upon a foundational assumption that the underlying decision process is stationary [Sutton and Barto, 2018]. While this assumption was a cornerstone when laying the theoretical foundations of the field, and while is often reasonable, it isseldom trueinpractice andcanbeunreasonable [Dulac-Arnold etal.,2019].
Transformation
Particularly important is the ability to incorporate domain knowledge of invariances, e.g., translational invariance ofimages. Kernels based onthemaximumsimilarity overagroup of transformations are not generally positive definite. Perhaps it is for this reason that they have not been studied theoretically. We address this lacuna and show thatpositivedefiniteness indeed holdswith high probabilityforkernels based on the maximum similarity in the small training sample set regime of interest, and that they do yield the best results in that regime.
Transformation
Particularly important is the ability to incorporate domain knowledge of invariances, e.g., translational invariance ofimages. Kernels based onthemaximumsimilarity overagroup of transformations are not generally positive definite. Perhaps it is for this reason that they have not been studied theoretically. We address this lacuna and show thatpositivedefiniteness indeed holdswith high probabilityforkernels based on the maximum similarity in the small training sample set regime of interest, and that they do yield the best results in that regime.