2026-10-13T07:40:56Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33989007 2026-09-24T17:44:41Z category_1 category_14 category_734 category_931 category_811 portal_5 item_type_3 month_year_09_2026

<p>Hyperparameter settings of the Transformer model.</p> Hongyi Yuan (1943686) Chunyan Wang (306606) Qiucheng Sun (627705) Mingyu Sun (397883) Mujie Fan (25105305) Biophysics Molecular Biology Biological Sciences not elsewhere classified Information Systems not elsewhere classified Mathematical Sciences not elsewhere classified world datasets demonstrate iterative representation learning existing deep learning driving significant breakthroughs deep learning models custom &# 8220 7 %&# 8211 1 %; meanwhile training clustering method final clustering results clustering accuracy compared fast louvain algorithm training clustering framework high clustering performance scale clustering ). xlink "> single 9 %&# 8211 cell rna sequencing based clustering cell louvain cell rna transformer framework comparable performance scale single transformer model traditional graph thus providing several state reliable tool related fields recent years rapidly advanced multiple real evaluation metrics developmental biology computational efficiency cancer research biomedical research based methods approximately 3 <div><p>Single-cell RNA sequencing (scRNA-seq) technology has rapidly advanced in recent years, driving significant breakthroughs in developmental biology, cancer research, immunology, and other related fields. However, existing clustering methods still face performance bottlenecks when handling large-scale scRNA-seq data. To address this challenge, we propose a self-training clustering method for large-scale single-cell RNA-seq data, termed scLTF (Single-cell Louvain-Transformer Framework for Large-scale Clustering). This method first generates preliminary clusters using a fast Louvain algorithm and then selects key cells based on a custom “Cluster Representativeness Index”. Subsequently, a Transformer model is employed for iterative representation learning and optimization to optimize the final clustering results. Experiments on multiple real-world datasets demonstrate that scLTF achieves overall superior or comparable performance in clustering accuracy compared with several state-of-the-art deep clustering methods, with evaluation metrics (ARI, NMI, and ACC) improving on average by approximately 3.9%–24.1%; meanwhile, its runtime is only about 6.7%–43.9% of existing deep learning-based methods. scLTF integrates the advantages of traditional graph-based clustering and deep learning models, achieving both high clustering performance and computational efficiency, thus providing a practical and reliable tool for biomedical research and cellular analysis.</p></div> 2026-09-24T17:44:21Z Dataset Dataset 10.1371/journal.pone.0359153.t001 https://figshare.com/articles/dataset/_p_Hyperparameter_settings_of_the_Transformer_model_p_/33989007 CC BY 4.0