The evaluation of spoken language understanding (SLU) systems is often restricted to assessing their global performance or examining predefined subgroups of interest. However, a more detailed analysis at the subgroup level has the potential to uncover valuable insights into how speech system performance differs across various subgroups. In this work, we identify biased data subgroups and describe them at the level of user demographics, recording conditions, and speech targets. We propose a new task-, model- and dataset-agnostic approach to detect significant intra- and cross-model performance gaps. We detect problematic data subgroups in SLU models by leveraging the notion of subgroup divergence. We also compare the outcome of different SLU models on the same dataset and task at the subgroup level. We identify significant gaps in subgroup performance between models different in size, architecture, or pre-training objectives, including multi-lingual and mono-lingual models, yet comparable to each other in overall performance. The results, obtained on two SLU models, four datasets, and three different tasks–intent classification, automatic speech recognition, and emotion recognition–confirm the effectiveness of the proposed approach in providing a nuanced SLU model assessment.

End-to-End Spoken Language Understanding models are generally evaluated according to their overall accuracy, or separately on (a priori defined) data subgroups of interest. We propose a technique for analyzing model performance at the subgroup level, which considers all subgroups that can be defined via a given set of metadata and are above a specified minimum size. The metadata can represent user characteristics, recording conditions, and speech targets. Our technique is based on advances in model bias analysis, enabling efficient exploration of resulting subgroups. A fine-grained analysis reveals how model performance varies across sub-groups, identifying modeling issues or bias towards specific subgroups.We compare the subgroup-level performance of models based on wav2vec 2.0 and HuBERT on the Fluent Speech Commands dataset. The experimental results illustrate how subgroup-level analysis reveals a finer and more complete picture of performance changes when models are replaced, automatically identifying the subgroups that most benefit or fail to benefit from the change.

We present results from a large-scale experiment on pretraining encoders with non-embedding parameter counts ranging from 700M to 9.3B, their subsequent distillation into smaller models ranging from 17M-170M parameters, and their application to the Natural Language Understanding (NLU) component of a virtual assistant system. Though we train using 70% spoken-form data, our teacher models perform comparably to XLM-R and mT5 when evaluated on the written-form Cross-lingual Natural Language Inference (XNLI) corpus. We perform a second stage of pretraining on our teacher models using in-domain data from our system, improving error rates by 3.86% relative for intent classification and 7.01% relative for slot filling. We find that even a 170M-parameter model distilled from our Stage 2 teacher model has 2.88% better intent classification and 7.69% better slot filling error rates when compared to the 2.3B-parameter teacher trained only on public data (Stage 1), emphasizing the importance of in-domain data for pretraining. When evaluated offline using labeled NLU data, our 17M-parameter Stage 2 distilled model outperforms both XLM-R Base (85M params) and DistillBERT (42M params) by 4.23% to 6.14%, respectively. Finally, we present results from a full virtual assistant experimentation platform, where we find that models trained using our pretraining and distillation pipeline outperform models distilled from 85M-parameter teachers by 3.74%-4.91% on an automatic measurement of full-system user dissatisfaction.

Interacting proteins and protein domains coevolve on multiple scales, from their correlated presence across species, to correlations in amino-acid usage. Genomic databases provide rapidly growing data for variability in genomic protein content and in protein sequences, calling for computational predictions of unknown interactions. We first introduce the concept of direct phyletic couplings, based on global statistical models of phylogenetic profiles. They strongly increase the accuracy of predicting pairs of related protein domains beyond simpler correlation-based approaches like phylogenetic profiling (80% vs. 30-50% positives out of the 1000 highest-scoring pairs). Combined with the direct coupling analysis of inter-protein residue-residue coevolution, we provide multi-scale evidence for direct but unknown interaction between protein families. An in-depth discussion shows these to be biologically sensible and directly experimentally testable. Negative phyletic couplings highlight alternative solutions for the same functionality, including documented cases of convergent evolution. Thereby our work proves the strong potential of global statistical modeling approaches to genome-wide coevolutionary analysis, far beyond the established use for individual protein complexes and domain-domain interactions.