Conventional Federated Learning (FL) systems inherently assume a uniform processing capacity among clients for deployed models. However, diverse client hardware often leads to varying computation resources in practice.

Do feature attribution methods correctly attribute features?

To resolvethis issue, we propose asimple alteration to the model training process, which results in more socially aligned explanations and metrics.

Determining exactly which relationships to include when representing an object as agraph is not alwaysstraightforward.

Compositional generalization is a key challenge in grounding natural language tovisual perception. While deeplearning models haveachievedgreatsuccess in multimodal tasks likevisual question answering, recent studies haveshownthat they fail to generalize to new inputs that are simply an unseen combination of those seen inthetraining distribution [6].

We introduce the Globally Normalized Autoregressive Transducer (GNAT) for addressing thelabel biasproblem instreaming speech recognition. Oursolution admits a tractable exact computation of the denominator for the sequence-level normalization.