Results 91 to 100 of about 14,201 (266)

Single‐cell DNA methylation profiling: Technologies, computation, and applications in precision oncology

Molecular Oncology, EarlyView.
Single‐cell DNA methylation (scDNAme) profiling maps epimutational clonal evolution, revealing mechanisms of malignancy and therapeutic resistance across diverse cancer types. By providing a high‐resolution landscape of intratumoral heterogeneity, these technologies empower precise patient stratification, guide the development of enhanced ...
Ik Soo Kim
wiley   +1 more source

Relaxed Contrastive Learning for Federated Learning

2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
We propose a novel contrastive learning framework to effectively address the challenges of data heterogeneity in federated learning. We first analyze the inconsistency of gradient updates across clients during local training and establish its dependence on the distribution of feature representations, leading to the derivation of the supervised ...
Seonguk Seo, Jinkyu Kim 0005, Geeho Kim, Bohyung Han   +3 more
openaire   +2 more sources

CEACAM1 participation in breast cancer progression

Molecular Oncology, EarlyView.
In invasive breast cancer (BC), CEACAM1 shifts from an apical to a uniform membranous/cytoplasmic pattern, or is lost, as tumors dedifferentiate, inversely tracking the Ki‐67 proliferative index. In MCF‐7 cells, only CEACAM1‐4L suppresses proliferation, repressing cell cycle and growth factor genes.
Mykola Lyndin, Irina Kube‐Golovin, Anatolii Romaniuk, Gunther Wennemuth   +3 more
wiley   +1 more source

Partial contrastive point cloud self-supervised representation learning

Scientific Reports
Annotating 3D point cloud data is labor-intensive. Self-supervised representation learning can reduce the intense demand of manual annotation. However, the sparsity of point cloud, while containing rich geometric structural information, makes the self ...
Zijun Cheng, Yiguo Wang
doaj   +1 more source

Directed evolution of enzymes at the crossroads of tradition and innovation

FEBS Open Bio, EarlyView.
An iterative cycle of data‐driven enzyme optimization comprising four stages: genetic diversification of a template enzyme, expression of protein variants, high‐throughput evaluation, and machine‐learning‐guided redesign of the next variant library.
Maria Tomkova, Andrej Mirossay, Erik Sedlak   +2 more
wiley   +1 more source

Line graph contrastive learning for node classification

Journal of King Saud University: Computer and Information Sciences
Existing graph contrastive learning methods often rely on differences in node features within subgraphs, lacking effective capture of the global structural information of the graph.
Mingyuan Li, Lei Meng, Zhonglin Ye, Yuzhi Xiao, Shujuan Cao, Haixing Zhao   +5 more
doaj   +1 more source

From energy provision to protein synthesis: Tunnelling nanotubes as mediators of intercellular metabolic cooperation in cancer

FEBS Open Bio, EarlyView.
The cytoskeleton‐mediated transport of mitochondria via tunnelling nanotubes restores respiration, increases ATP production, rescues cells from apoptosis, activates the AKT/mTOR signalling pathway, promotes cell migration and invasiveness, contributes to cancer progression and treatment resistance.
Stanislava Martínková, Jan Trnka
wiley   +1 more source

Edge Contrastive Learning: An Augmentation-Free Graph Contrastive Learning Model

Proceedings of the AAAI Conference on Artificial Intelligence
Graph contrastive learning (GCL) aims to learn representations from unlabeled graph data in a self-supervised manner and has developed rapidly in recent years. However, edge-level contrasts are not well explored by most existing GCL methods. Most studies in GCL only regard edges as auxiliary information while updating node features.
Yujun Li, Hongyuan Zhang, Yuan Yuan
openaire   +2 more sources

Pathways and pitfalls: a qualitative study of student experiences in biomedical science education

FEBS Open Bio, EarlyView.
Biomedical science students from underrepresented backgrounds face barriers including financial strain, disrupted laboratory access and cultural exclusion. Peer networks provide vital support when institutional systems are difficult to navigate. To create inclusive learning environments and achieve academic success, educators should blend active, hands‐
Olivia J. Russell, Cassy Ross, Panagiota S. Filippou, Bruno Silvester Lopes, Dominic Gilchrist, Paul Dean, Kamar Ameen‐Ali, Tora K. Smulders‐Srinivasan, Ahmad A. Khundakar   +8 more
wiley   +1 more source

Contrastive Learning with Synthetic Positives

Contrastive learning with the nearest neighbor has proved to be one of the most efficient self-supervised learning (SSL) techniques by utilizing the similarity of multiple instances within the same class. However, its efficacy is constrained as the nearest neighbor algorithm primarily identifies "easy" positive pairs, where the representations are ...
Dewen Zeng, Yawen Wu, Xinrong Hu, Xiaowei Xu 0004, Yiyu Shi 0001   +4 more
openaire   +2 more sources