Zhihan Zhou, Dejiao Zhang, Wei Xiao, Nicholas Dingwall, Xiaofei Ma, Andrew Arnold, Bing Xiang. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Xisen Jin, Dejiao Zhang, Henghui Zhu, Wei Xiao, Shang-Wen Li, Xiaokai Wei, Andrew Arnold, Xiang Ren. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Virtual assistants such as Google Assistant and Amazon Alexa host thousands of voice applications (skills) that han-dle a very large and diverse array of customer utterances. However, the number of supported skills may be much lower in some locales, particularly in countries other than the United States. Accordingly, customer utterances handled in a popular locale may be going unclaimed in another locale. Moreover, locales with smaller skill ecosystems also suffer from lim-ited labeled data for training systems to route utterances to skills. To tackle these aforementioned challenges, we propose a Cross-locale SHard-based cO-Training model (X-SHOT) that uses an iterative label augmentation approach to retrieve relevant skills in a source locale for unclaimed utterances in a target locale. The obtained results could be further used by skill developers in the source locale to gauge the latent de-mand for their skills in other locales and therefore to prioritize the internationalization of their skills accordingly.

Many recent successes in sentence representation learning have been achieved by simply fine-tuning on the Natural Language Inference (NLI) datasets with triplet loss or siamese loss.Nevertheless, they share a common weakness: sentences in a contradiction pair are not necessarily from different semantic categories.Therefore, optimizing the semantic entailment and contradiction reasoning objective alone is inadequate to capture the high-level semantic structure.The drawback is compounded by the fact that the vanilla siamese or triplet losses only learn from individual sentence pairs or triplets, which often suffer from bad local optima.In this paper, we propose PairSupCon, an instance discrimination based approach aiming to bridge semantic entailment and contradiction understanding with high-level categorical concept encoding.We evaluate PairSupCon on various downstream tasks that involve understanding sentence semantics at different granularities.We outperform the previous state-of-theart method with 10%-13% averaged improvement on eight clustering tasks, and 5%-6% averaged improvement on seven semantic textual similarity (STS) tasks.

Robust principal component analysis (PCA) is a key technique for dynamical high-dimensional data analysis, including background subtraction for surveillance video. Typically, robust PCA requires all observations to be stored in memory before processing. The batch manner makes robust PCA inefficient for big data. In this paper, we develop an efficient online robust PCA method, namely, online moving window robust principal component analysis (OMWRPCA). Unlike the existing algorithms, OMWRPCA can successfully track not only slowly changing subspaces but also abruptly changing subspaces. Embedding hypothesis testing into the algorithm enables OMWRPCA to detect change points of the underlying subspaces. Extensive numerical experiments, including real-time background subtraction, demonstrate the superior performance of OMWRPCA compared with other state-of-the-art approaches.

Nonparametric correlations such as Spearman's rank correlation and Kendall's tau correlation are widely applied in scientific and engineering fields. This paper investigates the problem of computing nonparametric correlations on the fly for streaming data. Standard batch algorithms are generally too slow to handle real-world big data applications. They also require too much memory because all the data need to be stored in the memory before processing. This paper proposes a novel online algorithm for computing nonparametric correlations. The algorithm has O (1) time complexity and O (1) memory cost and is quite suitable for edge devices, where only limited memory and processing power are available. You can seek a balance between speed and accuracy by changing the number of cut-points specified in the algorithm. The online algorithm can compute the nonparametric correlations 10 to 1,000 times faster than the corresponding batch algorithm, and it can compute them based either on all past observations or on fixed-size sliding windows. The algorithm is applied to analyze real-world sensor data from an industrial plant1.

Parkinson's disease (PD) is a debilitating neurodegenerative disease excessively affecting millions of patients. Early diagnosis of PD is critical as manifestation of symptoms occur many years after the onset of neurodegenration, when more than 60\% of dopaminergic neurons are lost. Since there is no definite diagnosis of PD, the early management of disease is a significant challenge in the field of PD therapeutics. Therefore, identifying valid biomarkers that can characterize the progression of PD has lately received growing attentions in PD research community. In this paper, we employ a multi-task learning regression framework for prediction of Parkinson's disease progression, where each task is the prediction of PD rating scales at one future time point. We then use the model to identify the important biomarkers predictive of disease progression. We adopt a graph regularization approach to capture the relationship between different tasks and penalize large variations of the model at consecutive future time points. We have carried out comprehensive experiments using different categories of measurements at baseline from Parkinson's Progression Markers Initiative (PPMI) database to predict the severity of PD, measured by unified PD rating scale. We use the learned model to identify the biomarkers with significant contribution in prediction of PD progression. Our results confirm some of the important biomarkers identified in existing medical studies, validate some of the biomarkers that have been observed as a potential marker of PD and discover new biomarkers that have not yet been investigated.