Joulani et al. (2013) have studied multi-armed bandits with delayed feedback under the assumption that the rewards are stochastic and the delays are sampled from a fixed distribution.

However, some recent studies haverecognized that most ofthese approaches failtoimprovethe performance over empirical risk minimization especially when applied to overparameterized neural networks.

It is structured as follows.

Althoughintuitive,sucha nativelabelassignment strategycannot revealtheunderlying semantic similarity between aquery anditspositivesandnegatives,andimpairs performance, since some negatives are semantically similar to the query or even share the same semantic class as the query.