contrastive learning

stMMR: accurate and robust spatial domain identification from spatially resolved transcriptomics with multi-modal feature representation

stMMR identifies spatial domains from multi-modal spatial transcriptomics data using geometric deep learning, showing improved performance in domain identification and gene discovery across biological contexts.

DiffGAN: a conditional generative adversarial network for phasing single molecule diffraction data to atomic resolution

DiffGAN estimates missing phases in single molecule cryo-electron diffraction data to determine protein structures at atomic resolution using a conditional GAN.

Contrastive Fitness Learning: Reprogramming Protein Language Models for Low-N Learning of Protein Fitness Landscape

This paper introduces ConFit, a method to train protein language models with limited experimental data for fitness prediction, showing improved accuracy and efficiency in low-N learning scenarios.

Unlocking the soundscape of coral reefs with artificial intelligence: pretrained networks and unsupervised learning win out

This paper demonstrates that pretrained neural networks and unsupervised learning effectively analyze coral reef soundscapes for ecological insights without manual annotation.

3 Directional Inception-ResUNet: deep spatial feature learning for multichannel singing voice separation with distortion

This paper proposes a 3 directional Inception-ResUNet for multichannel singing voice separation, achieving better performance than comparison models with an average NSDR of 11.55db.

CellDiffusion: Enhancing scRNA-seq data integration with accurate inference of cell-cell association

CellDiffusion improves scRNA-seq data integration by accurately inferring cell-cell associations and correcting batch effects across diverse datasets.

Synthetic Histology Images for Training AI Models: A Novel Approach to Improve Prostate Cancer Diagnosis

This paper generates synthetic histology images using GANs to train an AI model for prostate cancer grading, showing improved performance compared to models trained on real data.

Self-inspired learning to denoise for live-cell super-resolution microscopy

This paper introduces SN2N, a self-supervised denoising method for live-cell super-resolution microscopy that achieves competitive performance without requiring large clean datasets.

Self-supervised deep learning of gene-gene interactions for improved gene expression recovery

This paper improves gene expression recovery in scRNA-seq by using self-supervised deep learning to model gene-gene interactions.

Community-aware explanations in knowledge graphs with XP-GNN

This paper introduces XP-GNN, a method for explaining Graph Neural Networks in biological knowledge graphs by leveraging community structures, showing improved relevance in drug discovery use cases.

Multimodal CustOmics: A Unified and Interpretable Multi-Task Deep Learning Framework for Multimodal Integrative Data Analysis in Oncology

This paper introduces a deep learning framework for integrating multimodal cancer data, showing improved performance and interpretability across TCGA datasets.

Cross-institutional HER2 assessment via a computer-aided system using federated learning and stain composition augmentation

This paper develops a HER2 assessment system using federated learning and stain composition augmentation, achieving over 90% accuracy across two institutions with improved generalizability.

Self-supervised learning to predict intrahepatic cholangiocarcinoma transcriptomic classes on routine histology

This paper develops a self-supervised learning model to predict iCCA transcriptomic classes from histological images, achieving good to very good performance in multiple validation sets.

Generative AI for Cardiac Organoid Florescence Generation

This paper uses conditional GANs to colorize brightfield images of cardiac organoids, providing cell type-specific fluorescence information to enhance their characterization.

A Supervised Contrastive Framework for Learning Disentangled Representations of Cell Perturbation Data

This paper introduces SC-VAE, a supervised contrastive framework for analyzing cell perturbation data, showing improved sensitivity in detecting perturbation effects compared to existing methods.

Predicting cell types with supervised contrastive learning on cells and their types

This paper introduces SCLSC, a supervised contrastive learning method for cell type annotation in scRNA-seq data, achieving better accuracy and scalability than existing methods.

Multi-Modal CLIP-Informed Protein Editing

This paper introduces ProtET, a method for CLIP-informed protein editing using multi-modality learning, achieving significant improvements in protein stability and functionality.

spatiAlign: An Unsupervised Contrastive Learning Model for Data Integration of Spatially Resolved Transcriptomics

spatiAlign integrates spatially resolved transcriptomics datasets using contrastive learning, improving batch effect removal and enabling downstream analysis across multiple tissue sections.

Enhancing Whole Slide Image Classification with Discriminative and Contrastive Learning

This study improves whole slide image classification by using discriminative and contrastive learning to create WSI-level positive and negative samples, enhancing classification performance and feature robustness.

WildCLIP: Scene and animal attribute retrieval from camera trap data with domain-adapted vision-language models

WildCLIP retrieves camera trap images based on environmental and ecological attributes using a domain-adapted vision-language model.

Improving cell-type identification with Gaussian noise-augmented single-cell RNA-seq contrastive learning

This paper introduces a contrastive learning framework with Gaussian noise augmentation for improved cell-type prediction in scRNA-seq data.

Enhancing Opioid Bioactivity Predictions through Integration of Ligand-Based and Structure-Based Drug Discovery Strategies with Transfer and Deep Learning Techniques

This study uses transfer learning with ligand- and structure-based descriptors to improve opioid bioactivity prediction across OR subtypes.

Enhancing Antibody Language Models with Structural Information

This paper enhances antibody language models by integrating structural information through a contrastive learning approach, improving binding prediction accuracy and data efficiency.

A cross-attention transformer encoder for paired sequence data

The paper introduces a cross-attention transformer encoder for multi-sequence inputs and evaluates it on peptide-MHCII binding prediction.

ATAT: Automated Tissue Alignment and Traversal in Spatial Transcriptomics with Self-Supervised Learning

ATAT automates spatial transcriptomics analysis by aligning and traversing tissue samples using self-supervised learning, improving gene expression modeling without manual annotation.

Multi-Level Contrastive Learning for Protein-Ligand Binding Residue Prediction

This paper introduces MucLiPred, a deep learning model using dual contrastive learning to improve prediction of protein-ligand interactions and identify binding residues, showing superior performance over existing methods.

High-Activity Enhancer Generation based onFeedback GAN with Domain Constraint andCurriculum Learning

This paper introduces Enhancer-GAN, a GAN-based method for generating high-activity enhancer sequences, showing improved activity compared to benchmark datasets and identifying new motifs.

DNDesign: Enhancing Physical Understanding of Protein Inverse Folding Model via Denoising

DNDesign improves protein inverse folding by integrating denoising training to better capture the energy landscape, enhancing performance across multiple models.

Democratizing computational pathology: optimized Whole Slide Image representations for The Cancer Genome Atlas

This paper introduces Giga-SSL, a self-supervised method for learning Whole Slide Image representations from The Cancer Genome Atlas data, improving classification performance and enabling efficient analysis.

Improving generalizability for MHC-I binding peptide predictions through geometric deep learning

This paper improves MHC-I binding peptide prediction generalizability using geometric deep learning, outperforming sequence-based methods without binding affinity data.