Contrastive learning enables rapid mapping to multimodal single-cell atlas of multimillion scale

Single-cell datasets continue to grow in size, posing computational challenges for dealing with expanded scale, extended modality and inevitable batch effects. Deep learning-based approaches have recently emerged to address these points by deriving nonlinear cell embeddings. Here we present contrastive learning of cell representations, Concerto, which leverages a self-supervised distillation framework to model multimodal single-cell atlases. Simply by discriminating each cell from the others, Concerto can be adapted to various downstream tasks such as automatic cell type classification, data integration and especially reference mapping. Unlike current mainstream packages, Concerto’s contrastive setting well supports operating on all genes to preserve biological variations. Concerto can flexibly generalize to multiomics to obtain unified cell representations. Benchmarking on both simulated and real datasets, Concerto substantially outperforms competing methods. By mapping to a comprehensive reference, Concerto recapitulates differential immune responses and discovers disease-specific cell states in patients with COVID-19. Concerto is easily parallelizable and efficiently scalable to build a 10-million-cell reference within 1.5 h and query 10,000 cells within 8 s. Overall, Concerto will facilitate biomedical research by enabling iteratively constructing single-cell reference atlases and rapidly mapping novel dataset against them to transfer relevant cell annotations. © 2022, The Author(s), under exclusive licence to Springer Nature Limited.

Question Classification Using Universal Sentence Encoder and Deep Contextualized Transformer

One of the most vital steps in automatic Question Answer systems is Question classification. The Question classification is also known as Answer type classification, identification, or prediction. The precise and accurate identification of answer types can lead to the elimination of irrelevant candidate answers from the pool of answers available for the question. High accuracy of Question Classification phase means highly accurate answer for the given question. This paper proposes an approach, named Question Sentence Embedding(QSE), for question classification by utilizing semantic features. Extracting a large number of features does not solve the problem every time. Our proposed approach simplifies the feature extraction stage by not extracting features such as named entities which are present in fewer questions because of their short length and features such as hypernyms and hyponyms of a word which requires WordNet extension and hence makes the system more external sources dependent. We encourage the use of Universal Sentence Embedding with Transformer Encoder for obtaining sentence level embedding vector of fixed size and then calculate semantic similarity among these vectors to classify questions in their predefined categories. As it is the time of the Global pandemic COVID-19 and people are more curious to ask questions about COVID. So, our experimental dataset is a publicly available COVID-Q dataset. The acquired result highlights an accuracy of 69% on COVID questions. The approach outperforms the baseline method manifesting the efficacy of the QSE method. © 2021 IEEE.