RT Conference Proceedings
T1 A transfer-learning approach to feature extraction from cancer transcriptomes with deep autoencoders
A1 López-García, Guillermo
A1 Jerez-Aragonés, José Manuel
A1 Franco, Leonardo
A1 Veredas-Navarro, Francisco Javier
K1 Cáncer - Diagnóstico
K1 Computación, Teoría de la
K1 Congresos y conferencias
AB The diagnosis and prognosis of cancer are among the morechallenging tasks that oncology medicine deals with. With the main aimof fitting the more appropriate treatments, current personalized medicinefocuses on using data from heterogeneous sources to estimate the evolu-tion of a given disease for the particular case of a certain patient. In recentyears, next-generation sequencing data have boosted cancer prediction bysupplying gene-expression information that has allowed diverse machinelearning algorithms to supply valuable solutions to the problem of cancersubtype classification, which has surely contributed to better estimationof patient’s response to diverse treatments. However, the efficacy of thesemodels is seriously affected by the existing imbalance between the highdimensionality of the gene expression feature sets and the number of sam-ples available for a particular cancer type. To counteract what is knownas the curse of dimensionality, feature selection and extraction methodshave been traditionally applied to reduce the number of input variablespresent in gene expression datasets. Although these techniques work byscaling down the input feature space, the prediction performance of tradi-tional machine learning pipelines using these feature reduction strategiesremains moderate. In this work, we propose the use of the Pan-Cancerdataset to pre-train deep autoencoder architectures on a subset com-posed of thousands of gene expression samples of very diverse tumortypes. The resulting architectures are subsequently fine-tuned on a col-lection of specific breast cancer samples. This transfer-learning approachaims at combining supervised and unsupervised deep learning modelswith traditional machine learning classification algorithms to tackle theproblem of breast tumor intrinsic-subtype classification.
YR 2019
FD 2019-06-18
LK https://hdl.handle.net/10630/17831
UL https://hdl.handle.net/10630/17831
LA eng
NO Universidad de Málaga. Campus de Excelencia Internacional Andalucía Tech.
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 19 ene 2026