This paper discusses the problem of weakly supervised classification, in which instances are given weak labels that are produced by some label-corruption process. The goal is to derive conditions under which loss functions for weak-label learning are proper and lower-bounded – two essential requirements for the losses used in class-probability estimation. To this end, we derive a representation theorem for proper losses in supervised learning, which dualizes the Savage representation. We use this theorem to characterize proper weak-label losses and find a condition for them to be lower-bounded. From these theoretical findings, we derive a novel regularization scheme called generalized logit squeezing, which makes any proper weak-label loss bounded from below, without losing properness. Furthermore, we experimentally demonstrate the effectiveness of our proposed approach, as compared to improper or unbounded losses. The results highlight the importance of properness and lower-boundedness.

«Physical Review Letters Editor’s Suggestion»
We investigate the problem of an infinitely heavy impurity interacting with a dilute Bose gas at zero temperature. When the impurity-boson interactions are short-ranged, we show that boson-boson interactions induce a quantum blockade effect, where a single boson can effectively block or screen the impurity potential. Since this behavior depends on the quantum granular nature of the Bose gas, it cannot be captured within a standard classical-field description. Using a combination of exact quantum Monte Carlo methods and a truncated basis approach, we show how the quantum correlations between bosons lead to universal few-body bound states and a logarithmically slow dependence of the polaron ground-state energy on the boson-boson scattering length. Moreover, we expose the link between the polaron energy and the spatial structure of the quantum correlations, spanning the infrared to ultraviolet physics.